Transcriptional profiling of human smooth muscle cells infected with gingipain and fimbriae mutants of Porphyromonas gingivalis

Porphyromonas gingivalis (P. gingivalis) is considered to be involved in the development of atherosclerosis. However, the role of different virulence factors produced by P. gingivalis in this process is still uncertain. The aim of this study was to investigate the transcriptional profiling of human aortic smooth muscle cells (AoSMCs) infected with wild type, gingipain mutants or fimbriae mutants of P. gingivalis. AoSMCs were exposed to wild type (W50 and 381), gingipain mutants (E8 and K1A), or fimbriae mutants (DPG-3 and KRX-178) of P. gingivalis. We observed that wild type P. gingivalis changes the expression of a considerable larger number of genes in AoSMCs compare to gingipain and fimbriae mutants, respectively. The results from pathway analysis revealed that the common differentially expressed genes for AoSMCs infected by 3 different wild type P. gingivalis strains were enriched in pathways of cancer, cytokine-cytokine receptor interaction, regulation of the actin cytoskeleton, focal adhesion, and MAPK signaling pathway. Disease ontology analysis showed that various strains of P. gingivalis were associated with different disease profilings. Our results suggest that gingipains and fimbriae, especially arginine-specific gingipain, produced by P. gingivalis play important roles in the association between periodontitis and other inflammatory diseases, including atherosclerosis.

atherosclerosis 13 . Gingipains are cysteine proteases, which include arginine gingipains (Rgp) and lysine gingipain (Kgp). Gingipains, the main virulence factors produced by P. gingivalis, modulate the expression of cytokines and immunoglobulins and thus affect the immune responses of the host cells [14][15][16][17] . Fimbriae are hair-like protein structures of the outer surfaces of bacteria. Studies have shown that fimbriae facilitate bacteria to attach and invade to the host cells 18,19 . There are two groups of fimbriae produced by P. gingivalis, major fimbriae and minor fimbriae, and both are involved in the development of periodontitis 20,21 .
Up to now, no study has been conducted to elucidate the signaling mechanisms of the virulence factors during P. gingivalis infection of SMCs, and thereby their role in atherosclerosis. The aim of this study was to investigate the effects of gingipains and fimbriae in the regulation of gene expression profiling in human aortic smooth muscle cells.

Distinct gene lists regulated by different strains of P. gingivalis. The microarray experiments
were performed to analyze the gene expression in AoSMCs treated with different P. gingivalis strains, including ATCC33277 (wild type), W50 (wild type), 381 (wild type), E8 (W50 derived Rgp mutant), K1A (W50 derived Kgp mutant), DPG3 (381 derived major fimbriae mutant), and KRX178 (381 derived minor fimbriae mutant). Through analyzing microarray raw data using limma package, we got 7 lists of differentially expressed genes of interest based on setting the threshold of fold change >1.5 with adjust p-value (Benjamini-Hochberg) <0.05. The wild type and gingipain mutants infected groups were compared to uninfected control group and the fimbriae mutants infected groups were compared to erythromycin treated group. The wild type P. gingivalis strains, ATCC33277 (Table S1), W50 (Table S2), and 381 (Table S3) showed more power to regulate the gene expression than E8 (Table S4), K1A (Table S5), DPG3 (Table S6), and KRX178 (Table S7) in AoSMCs.
Disease ontology (DO) analysis for genes differentially expressed from AoSMCs infected with P. gingivalis. To identify the differentially expressed genes correlated diseases, DO analysis was carried out by input the Entrez Gene identifiers (Entrez Gene IDs) from the gene lists into clusterProfiler package. We found that the differentially expressed genes regulated by P. gingivalis strain ATCC33277 were enriched in less DO categories compared to other wild type strains W50 and 381. In addition, for AoSMCs infected with P. gingivalis strain KRX178, the largest number of DO categories were found compared to other strains. However, the Rgp mutants E8 only significantly correlated to one DO term. (Fig. 1) The summary of DO analysis was list on Table S8.
we found 54 common differentially express genes ( Fig. 2A). In order to understand the roll of Rgp in the process of P. gingivalis infection of AoSMCs, the uncommon expression genes were picked out and insert into R platform and analyzed by SPIA package. These uncommon genes were significantly enriched in 27 KEGG pathways, including focal adhesion pathway, NOD-like receptor signaling pathway, MAPK signaling pathway, TGF-beta signaling pathway and several pathways related to different cancers (Fig. 2B, Table S9). The SPIA analysis of the uncommon differentially expressed genes (Fig. 2C) between the wild type W50 and W50-derived Kgp gingipain mutant K1A infected AoSMCs indicated that 9 KEGG pathways were significantly enriched, which are all included in KEGG pathways that derived from the uncommon genes when comparing W50 regulated genes to E8 regulated genes in AoSMCs (Fig. 2D, Table S9).
Functional analysis for genes differentially expressed from AoSMCs infected with P. gingivalis fimbriae mutants. Comparing the wild type P. gingivalis strain 381 with its corresponding major fimbriae mutant DPG3, the differentially expressed uncommon genes (Fig. 3A) were significantly enriched in 5 pathways, which include pathways in cancer, small cell lung cancer, pancreatic cancer, pathogenic Escherichia coli infection, and bladder cancer (Fig. 3B, Table S10). There are less differentially expressed genes between 381 and KRX178 ( Fig. 3C) than the differentially expressed genes between 381 and DPG3. Only one KEGG pathway, related to cancer, was significantly enriched by those uncommon genes (Fig. 3D, Table S10).
Functional analysis for genes differentially expressed from AoSMCs infected with wild type P. gingivalis. The venn diagrams for the differentially expressed genes from each wild type P. gingivalis strain treated groups were showed in Fig. 4. To find the key functions that are targeted by P. gingivalis infection, we did the gene ontology analysis for the common genes with significantly up-regulated and down-regulated genes, respectively, using Integrated Discovery (DAVID) bioinformatics tool. The top gene ontology (GO) annotation cluster associate to common up-regulated genes was related to blood vessel development, vasculature development, blood vessel morphogenesis, and the angiogenesiss ( Table 1). The top GO annotation cluster coordinates to common down-regulated genes were related to enzyme linked receptor protein signaling pathway, transmembrane receptor protein tyrosine kinase signaling pathway, and cell surface receptor linked signal transduction    (Table 1). These results indicated that P. gingivalis not only modulates the growth and angiogenesis of AoSMCs, but also alter the cell surface receptors of the cells.
To examine the pathways that were enriched by the common regulated genes, the corresponding mean fold change of those genes were inserted into the R platform and analyzed using GeneAnswers package. These genes were enriched in 5 KEGG pathways, which include: pathway in cancer, regulation of actin cytoskeleton, cytokine-cytokine receptor interaction, focal adhesion, and MAPK signaling pathway (Fig. 5).
Analysis of genes associated with atherosclerosis. The DO analysis for the common differentially expressed genes regulated by the three wild types P. gingivalis strains in AoSMCs showed that 27 diseases were significantly enriched and atherosclerosis is identified as the first disease with highest gene ratio (Fig. 6A). 25 genes from the common differentially expressed genes were associated with atherosclerosis (Table S11). The heat map based on the gene expression level from microarray results of those 25 genes revealed that E8 was clustered together with negative control samples and control samples with erythromycin. The fimbriae mutants DPG3 and KRX178 were clustered closer to wild type P. gingivalis compared with gingipain mutant K1A that was clustered close to E8 and control samples (Fig. 6B).
qRT-PCR validation. To validate the results from microarray experiment, we performed qRT-PCR for 8 genes of interest. Among those 8 genes, 7 genes, which include C-C motif chemokine 11 (CCL11), Interleukin 7 (IL-7), Nucleotide-binding oligomerization domain-containing protein 1 (NOD1), Interleukin-1 alpha (IL-1α ), Angiopoietin 2 (Angpt2), Fractalkine (CX3CL1), and Interleukin-8 (CXCL8) were showed correlation to the development of atherosclerosis from the DO analysis. We also checked the gene expression of Notch homolog 1 (NOTCH1), which has been found to play an important role in pathogenesis of atherosclerosis 22,23 . All qRT-PCR results were proved to have similar regulation as the microarray data (Fig. 7).

Discussion
Atherosclerosis begins with damage to the endothelium and different traditional risk factors are considered to affect this pathogenic process, such as hypercholesterolemia, smoking and hypertension. In addition, different virus and bacteria have been suggested to be involved in the progress of atherosclerosis 24 . These microorganisms either infect the vascular cells directly or affect the vascular wall indirectly by stimulating other types of cells to produce cytokines and acute phase proteins. Buhlin K, et al. 25 concluded that several traditional risk factors of atherosclerosis correlate to severe periodontitis. M. Yakob, et al. 26 found that P. gingivalis infection contributes to the development of carotid atherosclerosis. In this study, we have investigated the role of different virulent factors, which include arginine and lysine gingipains and major and minor fimbriae in P. gingivalis-induced inflammation in AoSMCs using microarray technique. Gingipains, which accounts for 85% of the total proteolytic activity of P. gingivalis, play multiple roles in regulation of bacterial biofilm formation and host immune responses 27,28 . In addition to gingipains, the fimbriae facilitate the binding of P. gingivalis to host cells and induce the production of various cytokines 29,30 .
In this study, we found that both gingipains and fimbriae affected the differently expressed genes in AoSMCs induced by P. gingivalis infection. The gingipain mutants, E8 and K1A, compared with its corresponding wild type strain W50, and the fimbriae mutants, DPG3 and KRX178, compared with its corresponding wild type strain 381, showed deceased number of differently expressed genes. These results reveal that gingipains play an important role in P. gingivalis infection of AoSMCs. By utilizing DO analysis, we further explored that how the differentially expressed genes correlated to different diseases. For all P. gingivalis strains, except E8, atherosclerosis is highly enriched. In addition, we also found that rheumatoid arthritis (RA) is enriched, which is consistent with previous studies showing that periodontitis and P. gingivalis infection are associated with RA 31,32 . The DO analysis of the common differentially expressed genes in AoSMCs trigged by wild type P. gingivalis also showed that atherosclerosis and RA are significantly enriched. These findings further support an association between pathogens of periodontitis and increased risk of getting other inflammatory diseases.
The SPIA analysis for uncommon genes regulated by gingipain mutants and their corresponding wild type P. gingivalis strain showed that both Rgp and Kgp affect the P. gingivalis-mediated activation of focal adhesion, ECM-receptor interaction, and actin cytoskeleton pathway. These results are consistent with the pathway analysis for the common differentially expressed genes regulated by wild type P. gingivalis, showing that the focal adhesion and the regulation of actin cytoskeleton pathway are enriched. Studies have demonstrated that changes in the interaction between cell and extra cellular matrix (ECM) lead to modulation of the cytoskeleton, which further  affect the motility of cells 33 . Remodeling of actin filaments, focal contacts and ECM contributes to the switch of smooth muscle cell phenotype and angiogenesis 34,35 . These findings are further consistent with the GO analysis results revealing that the common up-regulated genes regulated by wild type P. gingivalis strains in AoSMCs are linked to blood vessel development and angiogenesis. Furthermore, the remodeling affects both proliferation and migration of the smooth muscle cells, which are involved in the process of atherosclerosis 36,37 . In a recent study, injection of P. gingivalis in rabbit model induces atherosclerosis with the activation of MAPK pathway and the production of cytokines 38 . Accordingly, our pathway analysis for the common differentially expressed genes regulated by wild type P. gingivalis reveals that the MAPK pathway and the cytokine-cytokine receptor interaction pathway are enriched. The activation of MAPK cascades is important for vascular smooth muscle cells in neointima formation after vascular injury 39 . In our previous study, we reported that P. gingivalis inhibits the inflammatory response in T cells through activation of the MAPK signaling pathway 40 . In this study, we found CXCL8 is down regulated by wild type P. gingivalis.
Besides atherosclerosis and rheumatoid arthritis, periodontal disease is also linked to orodigestive cancers 41 . Oral squamous cell carcinoma (OSCC) is one of the most common cancers worldwide and has showed a direct association with P. gingivalis infection 42,43 . Elevated level of this pathogen was found in gingival carcinomas, compared with healthy gingival tissue. In a large European Prospective Investigation in a Cancer cohort, P. gingivalis was associated with more than 2 fold higher risk of pancreatic cancer 44 . In a NHANES cohort study, P. gingivalis was linked to a >2 fold increase in risk of orodigestive cancer mortality even independently of clinically appearance of periodontal disease 45 . In this study, we found the among common differentially expressed genes regulated by wild type P. gingivalis several are enriched in different cancers, including pancreas cancer and oral cancer. In addition, the function analysis for gingipain and fimbriae mutants revealed that gingipains and fimbriae play a role in the association between P. gingivalis infection and orodigestive carcinogenesis. For some cancers, such as OSCC, there is still lack of reliable diagnostic biomarkers and tools, so that the therapies fail to prevent malignant progression 46 . The differentially expressed genes related to cancers in P. gingivalis-infected AoSMCs, reported in this study, may suggest possible biomarkers to be used in identification and diagnosis of specific cancers.
Of specific interest was to analyze P. gingivalis modulated genes involved in atherosclerosis. Based on DO analysis, we found that 25 genes were identified to correlate with atherosclerosis. The heat map based on these 25 genes, showed no clear difference among negative control groups and the E8-treated group. K1A-stimualted AoSMCs was clustered close to these three groups and the fimbriae mutants DPG3-and KRX178-treated group were clustered close to the wild type P. gingivalis, which further proved that gingipains, especially, Rgp are important for P. gingivalis-induced inflammation in AoSMCs. We picked out 7 genes from these 25 genes and validated by qRT-PCR. The qRT-PCR results were in good agreement with microarray data and results from unpublished studies. In our previous study, we found that wild type P. gingivalis ATCC33277, W50, and 381 up-regulate angiopoietin 2 (Angpt2) in AoSMCs through its corresponding transcription factor, v-ets avian erythroblastosis virus E26 oncogene homolog 1 (ETS1) 47 . The specific function of the genes related to atherosclerosis needs further investigation.

Conclusion
In summary, this study suggests that P. gingivalis infection in AoSMCs is related to many diseases, including atherosclerosis, rheumatoid arthritis, and different forms of cancers. Our findings further reveal possible mechanisms involved in the association between periodontitis and atherosclerosis. Gingipains and fimbriae, especially arginine-specific gingipain Rgp produced by P. gingivalis, play a crucial role in P. gingivalis infection of AoSMCs. Thus, inhibition of Rgp may be a preventive and therapeutic approach against periodontitis and its associated systemic diseases. 2). For DPG3 and KRX178, the culture medium supplemented with 1 ug/ml erythromycin in plus. All P. gingivalis strains were cultured using the anaerobic chamber (80% N 2 , 10% CO 2 , and 10% H 2 , 37 °C) (Concept 400 Anaerobic Workstation; Ruskinn Technology Ltd., Leeds, United Kingdom).

Methods
All P. gingivalis strains were allowed to growth for 72 h before harvested by centrifugation for 10 min at 10000 rpm at room temperature, washed with Krebs-Ringer-Glucose (KRG) buffer (120 mM NaCl, 4.9 mM KCl, 1. measuring the optical density (OD) at 600 nm of the bacteria suspension in KRG buffer by a spectrophotometer (BioPhotometer plus); (Eppendorf AG, Hamburg, Germany). P. gingivalis inoculation. AoSMCs from passage 5-10 were dissociated by trypsin/EDTA solution (Gibco, Carlsbad, CA). After add cell culture medium, the suspended cells were centrifuged at 14,000 rpm for 4 min and re-suspended in fresh cell culture medium. 150,000 cells were seeded per well of the 6-well plate coated with Type I collagen (Gibco, Carlsbad, CA). Cells were cultured in DMEM medium (Gibco, Carlsbad, CA) with 0.5% FBS (Sigma, St. Louis, MO), 2 mM L-glutamine (Gibco, Carlsbad, CA) and antibiotics (Gibco, Carlsbad, CA) for 24 h to get starved. Whereafter, AoSMCs were washed and re-suspended with fresh DMEM medium with 2 mM Lglutamine. The AoSMCs were challenged with different strain of P. gingivalis with the concentration of 10 MOI for 24 h. For DPG3 and KRX178 infection, 1 ug/ml of erythromycin were added to each well and AoSMCs, treated with 1 ug/ml erythromycin were served as control.
Microarray gene expression analysis. The RNA from P. gingivalis-infected AoSMCs for 24 h was extracted from the cells using RNeasy Kit (Omega Bio-Tek, Norcross, GA). The integrity of RNA was accessed using Agilent Bioanalyze (Agilent, Santa Clara, CA) and nanodrop 2000 (Thermo, Wilmington, DE). Followed the protocol of Agilent one color microarray, the RNA samples were stained with cy3 fluorescence dye, fragmented, and load to Agilent human whole genome 8 × 60k arrays (Agilent, Santa Clara, CA). After hybridizing the slides for 17 h at 65 °C with rotation at 10× g, the array slides were scanned by Agilent Microarray Scanner (Agilent, Santa Clara, CA). The microarray data for ATCC33277 infected AoSMCs was downloaded from the Arrayexpress database (E-MTAB-1922) uploaded from our previous study 48 . The limma package 49 offered by Bioconductor repository 50 was used to analyze the data from scanned array pictures pre-processed by Feature Extraction software (version 6.1.1, Agilent Technologies). After normalized the data, the linear model from limma package was applied to find the differentially expressed genes between each experiment groups with the cutoff of fold change >1.5 combined with Benjamini-Hochberg false discovery rate (FDR) < 0.05.

Disease ontology, KEGG pathway, and gene ontology enrichment analysis. The lists of Entrez
Gene identifiers (Entrez Gene IDs) for differentially expressed genes from each group were input into clusterProfiler 51 R package. Through enrichDO function, the diseases associated to the interesting genes were picked out according to the moderated t-test adjusted by Benjamini-Hochberg FDR. The DO categories with adjust p-value less than 0.1 were identified as significantly enriched DO categories. For KEGG pathway enrichment analysis, the GeneAnswers and SPIA package from bioconductor was used to find the pathways enriched by different genes lists of interest. The ENTREZ gene ID, fold change, and adjust p-value for each significantly expressed genes were input GeneAnswers or SPIA package. The significant KEGG pathways were identified with false discovery adjusted global p-value less than 0.1 for GeneAnswers package or p-value less than 0.05 for SPIA package. The GO cluster analysis was performed using the functional annotation tools function of Database for Annotation Visualization, and DAVID bioinformatics tool for significantly up-regulated and down-regulated common genes regulated by three wild type P. gingivalis strains, respectively. Quantitative real-time PCR validation. After RNA was isolated from the cells, cDNA were synthesized using equal amounts of RNA by High Capacity cDNA Reverse Transcription Kits (PERkin-Elmer Applied Biosystems, Foster City, CA) according to the manufacturer's protocol. Real-time PCR was performed using SYBR Green PCR kit (Fermentas, Sweden) with an ABI Prism 7900HT Sequence Analyzer (PERkin-Elmer Applied Biosystems, Foster City, CA). The primer sequences of 8 genes based on microarray experiment results were selected for qRT-PCR analysis and were listed in Table S12. Relative quantification of gene expression was determined by using the Δ Δ Ct method and normalized by the Ct value of GAPDH. Statistical Analysis. All experiments were preformed three times. Differentially expressed genes were identified by using Benjamini-Hochberg FDR to correct the multiple hypothesis test for the results preprocessed by limma package.