Community Research and Development Information Service - CORDIS


TRIGGER Report Summary

Project ID: 306029
Funded under: FP7-HEALTH
Country: Norway

Final Report Summary - TRIGGER (King of hearts, joints and lungs; periodontal pathogens as etiologic factor in RA, CVD and COPD and their impact on treatment strategies.)

Executive Summary:
Extensive clinical and epidemiological data clearly shows that chronic periodontal disease (PD), the most prevalent infectious inflammatory disease of mankind, is strongly linked to systemic inflammatory diseases such as cardiovascular diseases (CVD) , rheumatoid arthritis (RA) , and chronic obstructive pulmonary disease (COPD). Taking into account that up to 30% of the adult population worldwide suffers from severe periodontitis , the impact of this disease on human health is immense and has been recognised by World Health Organisation. Nevertheless, in many EU countries periodontal disease (PD) is a neglected disease, both by the population in general and health-care personnel. Often this negligence comes to the point that, like a hair-loss, the tooth-loss due to periodontitis is still considered as a normal inevitable event associated with ageing. To combat this misconception and conceive novel approaches to prevent and/or treat CVD, RA, and COPD we will explore the highly innovative idea that these non-communicable diseases are at least aggravated, if not initiated, by periodontal infection. Results emanating from our project will: i) elucidate a relationship between the presence of specific periodontal pathogens and severity of systemic diseases; ii) show that extensive periodontal treatment improves clinical parameters of investigated systemic diseases; iii) reveal the impact of eradication of specific periodontal pathogen on the level of inflammatory markers; iv) develop novel, periodontal-pathogen specific bactericidal compounds based on periodontal glutaminyl cyclase (QC), the enzyme essential for these pathogens vitality. This may reduce mortality and increase the quality of life of CVD, RA, and COPD patients. All of these may be possible based on the knowledge of mechanisms beyond the causative links between specific pathogen driven periodontal disease and CVD, RA, and COPD revealed by the research programme outlined in this project.
Project Context and Objectives:
Extensive clinical and epidemiological data clearly shows that chronic periodontal disease (PD), the most prevalent infectious inflammatory disease of mankind, is strongly linked to systemic inflammatory diseases such as cardiovascular diseases (CVD), rheumatoid arthritis (RA), and chronic obstructive pulmonary disease (COPD). Taking into account that up to 30% of the adult population worldwide suffers from severe periodontitis, the impact of this disease on human health is immense and has been recognized by World Health Organization. Nevertheless, in many EU countries PD is a neglected disease, both by population in general and health-care personnel. Often this negligence comes to the point that, like hair-loss, tooth-loss due to periodontitis is still considered a normal inevitable event associated with aging. To combat this misconception and conceive novel approaches to prevent and/or treat CVD, RA, and COPD we will explore highly innovative idea that these non-communicable diseases are at least aggravated, if not initiated, by periodontal infection. To this end, in this proposal we expect to elucidate and clarify causative links between chronic inflammatory diseases and PD initiated and driven by P. gingivalis infections. Implementation of the knowledge gain in this project should improve management of CVD, RA, and COPD.
Periodontitis, the chronic inflammatory disease caused by specific pathogens: The recent ground-braking paper showed that P. gingivalis is a key microorganism which initiates a shift in the composition of tooth-surface microflora from commensal to pathogenic. In humans, infection with P. gingivalis invites T. denticola and T. forsythia to colonize the subgingival bacterial biofilm. This small consortium of pathogens is described as the “red complex”. Proliferation of “red complex” species, especially P. gingivalis on the subgingival tooth surface initiates a chronic inflammation of the periodontium, which, if left untreated, can last for years causing erosion of tooth supporting structures. It is now understood that periodontal tissue damage is a consequence of a futile attempt by the host’s innate immune response to eradicate microbial invaders. The signs of an on-going “battle” is visible not only locally in the periodontal pockets but also systemically as elevated levels of inflammatory molecules present in the blood of periodontitis patients. This apparently may have a direct bearing on development of atherosclerotic plaque and aggravation of other chronic inflammatory diseases, including RA and COPD.
Rheumatoid arthritis (RA), an autoimmune disease triggered by citrullinated proteins and clinical link between RA and PD: Rheumatoid arthritis (RA), affecting 0.5-1% worldwide, is a severe systemic autoimmune disease, where patients suffer chronic joint inflammation, causing pain and disability, comorbidities and increased mortality. RA is fuelled by disease-specific autoantibodies to citrullinated proteins, products of physiological post-translational modification of proteins by endogenous peptidylarginine deiminases (PADs). Factors, which trigger the breakdown of tolerance to citrullinated proteins, are unknown. Interestingly, clinical studies of RA and periodontitis provide strong evidence for a significant association between the two diseases. Patients with longstanding, active RA have a substantially increased frequency of periodontitis as compared with healthy subjects. Conversely, patients with periodontal disease have higher prevalence of RA than patients without periodontitis2. The expression of PAD by P. gingivalis (PPAD) (unique for this bacterium) may constitute a mechanistic link explaining the clinical correlation between RA and periodontitis, since this bacterial enzyme is able to citrullinate proteins that are considered essential autoantigens in RA, including fibrinogen and enolase.
Cardiovascular diseases (CVD) and the correlation between CVD and PD: Cardiovascular diseases, including atherosclerosis, are now considered to be driven by chronic inflammation. Meta-analysis of cohort studies revealed that PD patients have a significantly increased risk of CVD (34%, p<0.0001)1. The mechanism responsible for this link is still unknown. One hypothesis suggests that periodontal inflammation by keeping circulating proinflammatory cytokines and factors at the elevated level contributes to atherogenic process. Smoking is also known as a common risk factor for both diseases and lately myeloperoxidase (MPO) driven protein carbamylation emerged as an important factor in atherogenesis. As MPO also serves as an inflammatory marker in periodontitis and is abundant in gingival tissue it may lead to high rate of lipoprotein carbamylation and hence may directly influence atherogenesis. The second hypothesis postulates that periodontopathogens can penetrate into the vascular lesion and stimulate immune cell influx into atherosclerotic plaque expediting local inflammation. Most likely, both pathways contribute to CVD pathology affecting the endothelial function. Importantly, endothelium dysfunction in patients with severe periodontitis was rescued by treatment of periodontitis.
Chronic obstructive pulmonary disease (COPD) and association with PD: COPD is a chronic disorder with substantial comorbidity contributing to high morbidity and mortality rates. COPD is associated with abnormal inflammatory immune responses of the lung to noxious particles and gases, which activate innate immune cells such as epithelial cells and macrophages by triggering pattern recognition receptors. Importantly, viral and bacterial infections not only cause acute exacerbations of COPD, but also amplify and perpetuate chronic inflammation in stable COPD. In common with periodontitis and vascular disease, the neutrophil is believed to be a clear effector cell releasing damaging proteolytic enzymes including serine proteinases and MPO. Important recent studies have confirmed that the neutrophil function is abnormal in usual COPD being less accurate in migration whilst more destructive to the tissues. This potential is greatly increased in patients with the genetic deficiency of Alpha-1-Antitrypsin Deficiency (AATD). Systemic inflammation is a recognised feature of COPD as is chronic colonisation of the airways, which influences exacerbation rate, and hence health status and lung function decline.
As the autoimmune component of COPD became recognised as an important factor, it was also recognized that protein citrullination also occurs in lungs and that may also serve as a marker of lung diseases. It had been also shown that citrullination of bactericidal peptide LL-37 by PPAD may affect its bactericidal and anti-inflammatory activities, thereby negatively affecting outcome of exacerbations. The clinical effect would be to increase colonisation of the airways, the frequency of bacterial infections and the systemic contribution to inflammation. In keeping, recently, several studies concluded that the risk for COPD is significantly elevated in patients with severe periodontitis suggesting a cause and effect.
Inflammation as a link between PD and systemic chronic diseases: Periodontitis is associated with increased plasma levels of CRP, IL-6 and fibrinogen, inflammatory markers linked to an acute exacerbation of COPD, atherosclerosis, ischaemic heart disease, stroke, and cardiovascular mortality. Significantly, circulating CRP is an independent predictor of vascular health. Patients with COPD have elevated plasma fibrinogen levels, which are associated with an increased prothrombotic risk. Exacerbations are associated with a rise in serum IL-6 levels leading to a further rise in plasma fibrinogen and increasing CRP production. Thus, periodontal infection fuelling a low-level but persistent systemic inflammation may have a role in predisposing to COPD colonisation and exacerbations, coronary heart disease or stroke.
Hypothesis: Based on available evidence we postulate that detailed evaluation of the mechanistic link between periodontal disease and chronic inflammatory diseases (RA, CVD, and COPD) will open novel opportunities to prevent or treat systemic diseases. Moreover, it is conceivable that oral interventions that improve oral health status will lower systemic inflammation and hence the severity and frequency of lung infection, RA, and CVD in susceptible populations. To this end we hypothesize that specific eradication of P. gingivalis by blocking glutaminyl cyclase (QC), the activity vital for bacterium proliferation or inhibition gingipain complex assembly/secretion at an early step of zymogens processing will resolve the chronic inflammation in the periodontium driven by virulence factors produced by this microorganism. In this context it is important to emphasize that a QC inhibitor should interfere with growth of two other important periodontopathogens (Prevotella intermedia and Tannerella forsythia), which also use QC to modify their secreted proteins.
In this proposal we seek to determine whether the characterization and exploitation of specific virulence factors of P. gingivalis and proteins at the interface between PD bacterial infection and human non-infectious inflammatory diseases has innovative potentials to pave the way to novel diagnostic, preventive and therapeutic strategies to combat CVD, COPD, and RA. If the hypothesis is sustained then some innovative routs for improved and effective management of these crippling and often mortal systemic diseases will be clarified, confirmed and hopefully implemented in routine medical procedures.
Therefore, overall objectives of this project are as follow:
• Investigate how the presence and abundance of P. gingivalis and other periodontal pathogens in patients affects clinical manifestations of CVD, COPD, and RA
• Study effect of extensive periodontal treatment and eradication of P. gingivalis and other periodontal pathogens on clinical manifestations of CVD, COPD, and RA
• Elucidate the impact of eradication of specific periodontal pathogen on the level of inflammatory markers known to be associated with severity and/or progression of CVD, COPD, and RA
• Employ relevant animal models to provide proof of concept that oral infection with P. gingivalis (experimental PD) impacts (induces and/or accelerates and/or aggravates) CVD, RA and COPD
• Study mechanisms how bacterial virulence factors (known and newly identified) affect CVD, RA and COPD at the functional, and molecular levels
• Develop novel, periodontal pathogen-specific antibacterial compounds based on the P. gingivalis target, glutaminyl cyclase (QC) enzyme and innovative approach to interfere with the assembly of virulence factors and test if they can prevent adverse effects of oral infection on atherosclerosis, RA and COPD
• Initiate training courses for young scientists and scientific exchanges
• Disseminate the knowledge and intellectual proprietary rights management (IPR)
These objectives will be accomplished by:
• Epidemiological approach by studying large patient cohorts
• Analysis of human clinical samples
• Interventional and progressive studies in humans
• Animal model studies
• Preclinical studies using proprietary inhibitors of glutaminyl cyclase (QC)
Project Results:
During 48 months of its existence the TRIGGER consortium made a steady progress in investigation of the prevalence of periodontitis in the Swedish Epidemiological Investigation of RA (EIRA) cohort. Additionally centres in Bern and Birmingham were able to proceed with samples collection and analysis. Continuous analysis of the RA and PD cohorts lead us to the few conclusions:
1) our combined data from RA and pre-RA cohorts, suggests that P. gingivalis may be a causative factor in the development of RA, in particular ACPA+ RA, and that this oral pathogen acts in concert with classical life style- and genetic risk factors in increasing the risk for RA.
2) our study shows no support for an association between established RA and prevalence and/or severity of periodontitis. Important to note though, we have not been able to investigate PD as a risk factor for RA, since most EIRA RA cases identified in DHR would have had RA for a number of years when the linking of registers was performed.
3) the risk for periodontitis increased significantly by age and current smoking status in both RA cases and controls, confirming smoking and ageing as risk factors for PD. We could not demonstrate increased prevalence of PD in patients with established RA compared to controls, and we could not detect any differences between ACPA+ and ACPA- RA, or between RF+ and RF- RA, with respect to periodontal disease
We have also been able to make significant discoveries on the front of detecting citrullinated and carbamylated peptides/proteins in biological fluids and established highly sensitive and reliable HPLC fluorometric method with N-dansyl-glycyl-L-arginine as the substrate for detection of the PAD activity in the biological fluid.
We also designed and developed two generations of the pgQC inhibitors throughout the lifetime of the project. The structural modification of the first generation compounds led to inhibitors exhibiting Ki values in the lower nM-range. Selected compounds were applied in growth experiments in vivo to determine the minimal inhibitor concentration (MIC) and minimal bactericidal concentration (MBC) of Oral pathogens P. gingivalis ATCC 33277, P. gingivalis M5-1-2, T. forsythia ATCC 42077, P. intermedia ATCC 25611 and as control S. gordonii ATCC 10558 and A. actinomycetemcomitans ATCC 33384. 2 of our preliminary QC inhibitors influenced bacterial growth in a different manner, whereas inhibitor MWT-S-00072 inhibited growth of clinical isolate P. gingivalis M5-1-2 as well as T. forsythia ATCC 42077 and P. intermedia ATCC 25611 significantly but not control strain S. gordonii ATCC 10558. These experiments underline the first proof of principle given in the last report (test of MWT-S-00001 in bacteria culture) concerning the grow inhibition of P. gingivalis. For the first time the exact values for the relevant biological constants on different species could be measured. However these values are not in a sufficient window for a animal/human trials. WE continue our work on perfecting the inhibitor design and efficacy.
During the 48 months of the total duration of the project, the following crystal structures were obtained, published and deposited with the Protein Data Bank: (1) Porphyromonas gingivalis Kgp in complex with inhibitor TCLK (PDB 4RBM), (2) P. gingivalis RgpB C-terminal domain (PDB 5AG8, 5AG9 and 5HFS), (3) P. gingivalis RagB (PDB 5CX8), (4) P. gingivalis PPAD, isolated, as substrate complex and as substrate-mimic complex (PDB 4YT9, 4YTG and 4YTB), (4) P. gingivalis PPAD, isolated, as substrate complex and as substrate-mimic complex (PDB 4YT9, 4YTG and 4YTB), (6) Tannerella forsythia prokarilysin (PDB 4R3V), (7) P. gingivalis PorZ (PDB 5M11),( 8) Pro-domain of P. gingivalis Kgp (PDB 5MUN), 8) Pro-domain of P. gingivalis Kgp (PDB 5MUN) as well as our primary target : pgQC.
Network-based mathematical models have been built around PD, RA, OA (osteoarthritis) as a negative control, COPD and ATH. The complex network behaviour was modelled by using artificial intelligence, machine learning, graph theory and statistical pattern recognition technologies. A virtual biological system able to faithfully simulate at molecular level the observed biology was obtained, allowing effective, mechanistically-driven analyses. Details of the analysis results are included into the progress reports. Major consortium achievements in relation to the project objectives are presented below:

Autoimmunity, genetics and environmental risk factors in RA, CVD and COPD, in relation to P. gingivalis infection and periodontitis – prevention based on new etiological models.
Eighty six per cent of EIRA participants were identified in the Swedish dental health registry (DHR). Approximately 70% of these had at least one diagnostic code relating to periodontal disease (increased risk for periodontitis; increased risk for peri-implantitis; gingivitis; periodontitis or peri-implantitis). Gingivitis (33% of RA / 35% of controls) and periodontitis (33% of RA / 32% of controls) were the most common diagnoses. The percentage of current smokers was significantly higher among RA cases (25%) compared to controls (18%) (p < 0.0001). The risk for periodontitis increased significantly by age and current smoking status in both RA cases and controls, confirming smoking and ageing as risk factors for PD. We could not demonstrate increased prevalence of PD in patients with established RA compared to controls, and we could not detect any differences between ACPA+ and ACPA- RA, or between RF+ and RF- RA, with respect to periodontal disease. The study shows no support for an association between established RA and prevalence and/or severity of periodontitis. Important to note though, we have not been able to investigate PD as a risk factor for RA, since most EIRA RA cases identified in DHR would have had RA for a number of years when the linking of registers was performed.
Partner 9 in collaboration with 1, 2 and 11 analysed oral microbiota in patient plaque samples. The preliminary results suggest that the composition of bacterial sequences at phylum level is quite similar between the two types of plaque samples, before and 12 months after scaling and root planning (SRP) In total, 11 phyla were identified and the sequence abundance of the 3 major phyla Bacteroidetes, Firmicutes, and Fusobacteria were only slightly changed between the sampling time points. However, the sequence abundance for Proteobacteria and Actinobacteria changed noticeably from 10% to 23 % and from 12% to 5%, respectively. The complexity and diversity of the human plaque microbiome were increased at genus level. Among the 22 most abundant genera (genera represented with a sequence abundance of 1-17%) Fusobacterium, Prevotella, and Streptococcus were presented with highest percentage and frequency for both types of samples (before and after treatment). Bacteria belonging to the red-complex, which includes important pathogens in adult periodontal dis- ease, such as Porphyromonas, Tannerella and Treponema, were also present among the top 22 genera for both sample types (before and after treatment). The sequence abundance for Porphyromonas was 7-9 %, for Tannerella 4-7 % and for Treponema only 1%.
We have completed another, pilot study, where we investigated the severity of periodontitis in patients with established RA. Moderate to severe PD was present in 75% of RA patients, while 25% had no or mild PD. Interestingly, ACPA positivity was significantly (p=0.032) more frequent (86%) in moderate/severe PD as compared to no/mild PD (50%). A number of pro-inflammatory mediators, including sCD30 (TNFRSF8), IL-19, IL-26, MMP1, gp130 (sIL-6Rb) and sTNF-R1 were significantly elevated in serum, saliva or GCF samples of RA patients with moderate/severe PD. Notably, the B cell survival and maturation factor APRIL (TNFSF13) was significantly increased in both serum and saliva in RA patients with moderate/severe PD, compared to RA patients with no/mild PD. This study somewhat contradicts our previous study. However, in the former study, we were unable to investigate severity of PD. Of note, the frequency of PD was also higher in the present study, compared to the previous study (75% versus 30%), as was the frequency of ACPA-positivity (75% versus 63%), factors which may explain the somewhat discrepant results.
From our new study, we conclude that patients with ACPA+ RA have more severe forms of PD, irrespective of DMARD/biological DMARD therapy, compared to ACPA- RA patients. Moreover, patients with moderate/severe PD have increased serum and salivary levels of the B cell survival and maturation factor APRIL, potentially pinpointing this cytokine as a key mediator involved in the association between severe forms of PD and ACPA+ RA.

In the cross-sectional study of Partner 11 (UB) both 62 study participants with periodontal disease (7 with aggressive (AP) and 55 with chronic periodontitis (CP), 26 with RA and 10 periodontally and systemic healthy controls were included. The objective of the analysis was to analyze the citrullination in gingival region in association with the presence of bacteria being associated with periodontitis. Number of teeth was lower in patients with RA, but number of sites with probing depths (PD) ≥5 mm was higher in study participants without RA. Whereas in volunteers without RA, presence of P. gingivalis was highly associated with periodontal disease, the species was detectable in 5 of the 14 RA patients without diagnosed periodontitis. The findings were similar for T. forsythia and T. denticola. As expected, in serum positive Anti-CCP levels were found mainly in RA patients (about 80%). Antibodies against P. gingivalis (whole bacterium) did not show any difference between groups and subgoups, antibodies against A. actinomycetemcomitans (leukotoxin) were highest in AP. Citrullination in GCF was correlated weakly with Anti-P. gingivalis (R=0.275; p=0.025) and moderately with GCF IL-1β levels in Non RA and with MCP-1 levels in RA (R=0.607; p=0.001). Univariate linear models confirm a dependency from these variables. Inclusion of smoking showed an influence when presence of P. gingivalis in biofilm was considered in addition.

Two separate ELISAs based on P. gingivalis-derived antigens have been developed, the RgpB protein ELISA and the citrullinated P.PAD-peptide 3 (CPP3) ELISA. These ELISAs have been used to screen a number of RA and pre-RA cohorts for presence of anti-P. gingivalis antibodies, including the large EIRA cohort of nearly 2000 RA cases and 400 matched controls, and the pre-RA cohort from northern Sweden of 251 RA cases and 198 controls, with n=422 blood samples donated before the onset of RA symptoms, by 5.2 (6.2) years [median (Interquartile range)]. Results from these screens show that elevated antibody levels to P. gingivalis are more common in RA patients than in controls, especially in ACPA+ RA. Moreover, our recent data show that these antibodies can be detected years before clinical onset of RA. In addition, Partner 11 found that PAD and bacterial peptidylarginine deiminase activity (PPAD) in the periodontium in the context of serum levels of antibodies against citrullinated epitopes and P. gingivalis were assessed in rheumatoid arthritis and periodontitis. Elevated PAD and PPAD activities within the periodontium were found in RA and non- RA patients with periodontitis. Further, the results suggest that PPAD secreted by P. gingivalis residing in epithelial cells may exert its citrullinating activity in distant regions of the periodontium or even distant tissues.
The RgpB and CPP3 ELISA have further been used to investigate presence of these antibodies in rats, following pristine-induced arthritis (PIA) in animals subjected to P. gingivalis infection by repeating swabs and ligature placement. Briefly, all rats infected by P. gingivalis developed PD; 89% mounted a strong anti-RgpB antibody response; and anti-CPP3 antibody levels were significantly increased (p<0.05) in infected PIA rats, compared no non-infected PIA rats. However, arthritic scores were not affected by P. gingivalis infection. This new experimental study demonstrates that infection by P. gingivalis induces a strong antibody response to RgpB in a majority of animals, and that a subset of animals also responded with elevated anti-CPP3 IgG levels, demonstrating a direct link between P. gingivalis and the production of anti-citrullinated protein (i.e. anti-CPP3) antibodies.

Using the EIRA cohort we have analyzed HLA-DRB1 SE alleles, the PTPN22 risk allele and smoking habits, as well as presence of ACPAs. Since we had no data on PD-status prior to RA development in EIRA, and since our study linking EIRA with DHR showed no evidence of an increased prevalence of PD in established RA, compared to controls, but we had data on anti-P. gingivalis antibodies, we focused on “anti-P. gingivalis antibody-associated” RA rather than on “PD- associated” RA, as stated in Objective 1.2. Our data show a significant association between anti-P. gingivalis (anti-RgpB) antibodies and RA, which was even stronger than the well-known association between smoking and RA. In ACPA-positive RA we could show significant interactions between anti-P. gingivalis antibodies and smoking, as well as HLA-DRB1 SE alleles, but not with the risk allele of PTPN22.
In our analyses of anti-RgpB and anti-CPP3 IgG in samples collected from before the clinical onset of RA, we could see that these antibodies were present years prior to the RA diagnosis. Anti- RgpB IgG levels were significantly increased in pre-symptomatic individuals and in RA-patients, compared to controls. Anti- CPP3 antibodies were found in 5% of pre-symptomatic individuals and in 8% of RA-patients, with elevated levels in both subsets compared to controls. Anti-CPP3 antibodies resembled the ACPA response, with increasing antibody levels over time, whilst anti-RgpB antibodies were elevated and stable in pre-symptomatic study subjects, with a trend towards lower levels after RA diagnosis.
We have also screened 2.859 serum samples from the Swedish population-based early RA cohort EIRA (Epidemiological Investigation of RA) for presence of antibodies to CPP3/RPP3, using the multiplex peptide array, containing CPP3/RPP3, as well as eight other citrullinated peptides derived from human protein sequences. With a cut-off for positivity set at the 98th percentile among 370 EIRA controls, 11% of EIRA RA cases were anti-CPP3 IgG positive. The great majority of these patients (91%) were found in the CCP2 positive subset, similar to the classical ACPA response directed against citrullinated epitopes on human proteins. There was no antibody response to the arginine-containing control peptide RPP3.
When analysing co-occurrence of different ACPA fine-specificities, anti-CPP3 antibodies were shown to cluster outside the main group of ACPA, and in gene-environment analyses, anti-CPP3 IgG showed no specific association with PTPN22 polymorphism, while classical ACPAs did, and when consideration was taken to CCP2 antibody status, no specific association was identified between anti-CPP3 antibodies and the best known genetic risk factor for ACPA+ RA, namely HLA-DRB1 SE. Smoking on the other hand, associated specifically with CPP3+ RA, beyond the well-described association between smoking and anti-CCP2 antibodies. Collectively these data may suggest a different aetiological pathway for the generation of anti-CPP3 antibodies, as compared to the classical ACPA response. We propose that this pathway involves the actions of P. gingivalis, specifically P. PAD.
In this study, we also investigated the antibody response to CPP3/RPP3 in 65 patients with chronic PD and 59 periodontally healthy controls, and a subset of PD patients (10%) had elevated anti-CPP3 and anti-RPP3 IgG levels, in the absence of an anti-CCP2 antibody response, while these antibodies were not present in periodontally healthy individuals. Moreover, we have identified two human monoclonal antibodies, generated by recombinant expression of immunoglobulins from B cells isolated from RA patients, with CPP3-reactivity. Interestingly, one of these monoclonal antibodies binds to both CPP3 and RPP3, but not to CCP2, while the other monoclonal antibody binds to CPP3 and to a citrullinated peptide derived from human vimentin. This latter antibody is also reactive with CCP2, but not with RPP3, and show extensive CDR mutations, indicative of antigen-driven clonal selection and affinity maturation.
Based on these new data, we hypothesise that the antibody response to CPP3/RPP3 (as seen in 10% of PD patients) is a physiological immune response to an exogenous (bacterial) antigen, and that autoimmunity characterised by the classical ACPA response may arise as a result of epitope-spreading to endogenous (human) citrullinated autoantigens.

A questionnaire of respiratory symptoms typical of COPD was administered to patients referred to the Dental Hospital for assessment of PD. This numbered 420, that have been analysed to date (average age 52.4 years) of whom 315 had some evidence of periodontitis and 105 served as dental controls. Females accounted for 64.6% of the subjects and 36.7% were either current or ex smokers. Localised PD was diagnosed in 34% of patients. Generalised PD was described in 56.6% ranging from mild (10.7%) to moderate (23.2%) and severe (22.6%). Aggressive PD was diagnosed in 9.4% of patients. This same group of subjects were asked to fill in COPD directed questionnaire including a Family history of relatives with a diagnosis of COPD/bronchitis or emphysema (3.8%/5.2% and 3.6% respectively). The same questions were asked of the patients of whom 0.24% had a diagnosis of COPD, 0.95% had bronchitis and 0.2% had emphysema. However, when asked about symptoms 4.0% had daily cough, 2.9% produced sputum regularly , 2.1% complained of symptoms consistent with chronic bronchitis and 6.9% complained of persistent or intermittent breathlessness on exertion indicating an increased incidence of COPD symptoms in patients with PD.. The results confirm a smoking related association and increased COPD related features although symptoms were greater than diagnosed terms and thus consistent with some under diagnosis of COPD in the community as a whole.
Because the Alpha-1 Antitrypsin Deficiency (AATD) patients are widely scattered throughout the United Kingdom a PD questionnaire was sent to more than 200 patients to determine the incidence and severity of PD symptoms and 180 have been returned for analysis. A similar questionnaire was distributed to COPD patients attending the hospital out-patients clinics aiming to recruit a further total of 100 in the 48 months of the project and we have this data on 188 patients. The data indicates the AATD patients were younger, were less likely to be smokers and had better dental hygiene although features of PD were similar between both groups although the AATD patients reported less tooth loss than usual COPD

Gingival tissue biopsies from PD patients with or without RA / ACPA were collected during dental surgery to treat PD. Monoclonal antibodies were generated using an in-house single B cell–based cloning technology for the isolation and expression of immunoglobulin (Ig) genes (Amara et al, JEM, 2013 Mar 11;210(3):445-55). To this date, we have successfully isolated 480 CD19+ B cells from an ACPA+ RA patient with chronic PD, and generated 48 monoclonal antibodies, which are currently being assessed for reactivity with citrullinated and bacterial antigens. We have analyzed 110 variable heavy chain Ig genes (IGHV) for mutational patterns, and compared to healthy controls. RA/PD B cells showed significant decrease in VH3, IGHV4–31 was overrepresented compared to controls, while IGHV1-2, IGHV3-23 and IGHV4-34 were under-represented. Antibodies with positively charged IGHV CDR3 regions, a feature associated with auto-reactivity, were enriched in gingival tissue. Alignment of VH sequences to their closest germ line counterparts revealed that RA/PD B cells exhibited extensive mutations in the IGH CDR regions and higher levels of somatic mutations in the V gene segments compared to controls, suggesting an antigen-driven B cell response.
We have used the EIRA database, which holds information on more than 6000 RA cases and matched controls regarding genetic risk factors, smoking habits and serology, for calculating RA risk in different subsets of patients (i.e. PD-positive versus PD-negative; RgpB-positive versus RgpB-negative; CPP3-positive versus CPP3-negative).

Based on the data generated within WP1, we can draw the following conclusions:
Data on self-reported periodontitis (from questionnaires) cannot be trusted
Antibodies to P. gingivalis virulence factor RgpB cannot be used as a surrogate marker for periodontitis
The antibody response to P. gingivalis virulence factor RgpB is significantly elevated in RA
(specifically ACPA+ RA) compared to controls
There is an interaction between anti-RgpB IgG exposure and HLA-DRB1 SE, as well as
smoking, in the development of ACPA+ RA, but not ACPA- RA
Severe forms of PD are more common in ACPA+, compared to ACPA- RA patients, suggesting a link between severity of PD and ACPA+ RA.
Serum & saliva levels of APRIL are elevated in RA patients with moderate/severe PD, compared to RA patients with no/mild PD. Since ACPA are enriched in moderate/severe PD, our data may suggest that APRIL (a B cell survival/maturation factor) plays a role in the generation of ACPA, and potentially this takes place at mucosal sites, such as the gingival tissue.
Antibodies to RgpB and CPP3 precede the onset of RA by several years, indicating that infection by P. gingivalis occurs before the onset of clinical RA (these data confirms our previous pre-RA study).
P. gingivalis infection in rats induces PD, and an antibody response to RgpB and CPP3, demonstrating a direct link between P. gingivalis and ACPA (i.e. anti-CPP3 antibody) production.
Antibodies to CPP3 and RPP3 are present (at low levels) in a subset (10%) of CCP2- non-RA patients with chronic PD, suggesting that only a subset of PD patients develops anti-CPP3 antibodies, and we propose that it is this group that has an increased risk of developing ACPA-positive RA in the future.
Compared to PD patients, RA patients (11%) have high levels of anti-CPP3 antibodies; these patients are mainly CCP2-positive/RPP3-negative, which may suggest that specific CPP3/RPP3 B cells (present in PD patients) produce antibodies that cross-react with human citrullinated antigens, and these B cells are subsequently selected and expanded by auto-reactive T cells (in RA).
Cigarette smoking associates specifically (beyond the effect of CCP2) with the presence of anti-CPP3 antibodies, while RA risk factors PTPN22 and HLA-DRB1 SE do not. Anti-CPP3 IgG also cluster outside the main group of classical ACPAs. Collectively, these data may suggest a different pathway for the generation of anti-CPP3 antibodies, and we suggest that this pathway involves the actions of P. gingivalis, and specifically P. PAD.
B cells with (weak) CPP3/RPP3 cross-reactivity and B cells with (strong) CPP3/Cit-vimentin cross-reactivity can both be detected in RA patients, and antibodies with CPP3/Cit-vimentin cross-reactivity show extensive CDR mutations, indicating antigen-driven clonal selection and affinity maturation.
Gingival tissue B cells from an ACPA+ RA patient with severe PD show extensive CDR mutations (suggesting local antigen-driven clonal selection and affinity maturation) and specific features associated with auto-reactivity, supporting the hypothesis that the gingival tissue could be a site for break of immune tolerance to citrullinated autoantigens.

Taken together, our novel data strengthens the support for the oral pathogen Porphyromonas gingivalis in the etiology of ACPA+ RA.

Impact of post-translationally modified proteins (PTM) on immune system, animal models to elucidate causative link between periodontitis and COPD, CVD, RA and effects of experimental treatment.

Due to the nature and amount of the modification, it is much more difficult to detect citrullination as compared to carbamylation in complex samples such as GCF, airway secretions, atherosclerosis plaque, synovial fluid, serum or plasma without further enrichment/purification. In our screening of synovial fluid we were able to 121 citrullinated proteins and 64 carbamylated proteins. As shown on the figure below only small percentage of the proteins (9 citrullinated and 4 carbamylated) overlapped with the previously reported by J van Beers (2012). We discovered that citrullination of fibrinogen prevents thrombin from cleaving of FibA and FibB fragments which is pivotal step in clot formation. Interestingly, in case of carbamylated fibrinogen, although thrombin is fully active and we can detect FibA and B in the supernatants using mass spectrometry clot formation is still suppressed. We also showed that citrullination of kininogen greatly affected its ability to release bradykinin in in vitro experiments. That active 9 aa peptide is formed by proteolytic cleavage between Lys362-Arg363 and Arg371-Ser372 of HMW kininogen.

A growing number of reports link RA with periodontitis which led us to assess human and bacterial peptidylarginine deiminase (PAD) activity in the gingival crevicular fluid (GCF) in the context of serum levels of antibodies against citrullinated epitopes in rheumatoid arthritis and periodontitis. To this end human PAD and P. gingivalis-derived enzyme (PPAD) activities were measured in the GCF of 52 rheumatoid arthritis (RA) patients (48 with periodontitis and four without) and 44 non-RA controls (28 with periodontitis and 16 without). Collectively, our data indicate that PAD and PPAD activities within the periodontium are elevated in RA and non-RA patients with periodontitis. Therefore it is apparent that PPAD secreted by P. gingivalis residing in epithelial cells may exert its citrullinating activity in distant regions of the periodontium or even distant tissues. This allows us to conclude that in periodontitis, the citrullination of proteins/peptides by human and bacterial peptidylarginine deiminases may generate antibodies after breaching immunotolerance in susceptible individuals.
One of the most abundant proteins on the P. gingivalis surface is FimA, which is assembled into long fimbriae covering like hairs the bacterium. FimA was purified from wild-type P. gingivalis strain ATCC 33277 and from the isogenic PPAD-null mutant. All analyzed samples of FimA derived from the wild-type P. gingivalis ATCC33277 were citrullinated on position 10 of the amino acid sequence. This is in the stark contrast to FimA obtained from the PPAD-null strain which shown no trace of citrullination of Arg on position 10. Together these results reveal beyond any doubt that FimA is citrullinated by PPAD.

The human LL-37 peptide does not only possess bactericidal activity, but is also a potent regulator of the immune response facilitating the delivery of oligonucleotides to intracellular TLR-9. We found that citrullination diminished the efficiency of the peptide to bind oligonucleotides in a manner directly proportional to the number of deiminated Arg residues (five in total). Moreover, citrullination of LL-37 hindered peptide-dependent DNA uptake and sensing by pDCs as manifested by decreased secretion of IFNα. Together, we showed that citrullination of LL-37 alters the immunomodulatory function of the peptide plausibly representing a novel mechanism contributing to the prevention of the breakdown of the immunotolerance dependent on recognition of self-DNA by antigen-presenting cells and thus development of autoimmunity.
The presence of LL-37 in NETs was confirmed by a targeted MS/MS analysis. Several different sample preparations including in gel digest, precipitation, ultrafiltration and combinations hereof were unsuccessful in identifying citrullinated LL-37.
Since we proved that PTMs in NETs can bring a high immunomodulatory impact, the subsequent part of studies was focused on the role of P. gingivalis in that process. Our studies have shown that many factors are able to induce NETs, such as gingipains secreted by P. gingivalis. Moreover, we revealed that NETs induced by gingipains do not exert antibactericidal activity against this pathogen and this paradox has stimulated us to conduct further investigation of that phenomenon. We found that the biological activity of NETs induced by gingipains is different from NETs generated by phorbol myristate acetate, a strong inducer of NETs. The formation of NETs is dependent on gingipains proteolytic activity, which was proved by using specific gingipain inhibitors, Kyt-1 and Kyt-36. Further studies were conducted to compare the bactericidal function of NETs against other bacteria of the oral cavity, such as Streptococcus salivarius. Obtained data has shown that NETs induced by gingipains are ineffective comparing to NETs-induced by PMA. Therefore, we started molecular analysis to verify the changes in NETs proteome, which can be responsible for that observation. Those studies are going to be conducted in the future, however our data already suggest that NETs induced by gingipains are ineffective and it could be the mechanism of P. gingivalis self-protection.

During the project we set up a highly sensitive and reliable HPLC fluorometric method with N-dansyl-glycyl-L-arginine as the substrate for detection of the PAD activity in the serum (and subsequently its citrullination capacity). We evaluated 229 patient from the ARCTIC cohort as well as 90 sera from healthy controls. Interestingly approximately 37% of the RA patients had a significant stimulatory effect on PAD4 activity. Also, of this later group, only 24% were ACPA negative, which strongly suggest PAD4’s involvement in regulation of inflammatory events preceding or independent of processes related ACPA formation. Statistical analyses is underway and will focus primarily on i) the differences between clinical manifestations and severity of RA between patient’s with high vs low PAD-activity ii) If biomarker signatures derived from the molecular data available, e.g. cytokine levels, indicate differences in the molecular basis of RA-pathogenesis for PAD-low and PAD-high patients? If so, what is the degree of resemblance and what are the key differences between stratifying patients according to ACPA and PAD-status? iii) If therapeutic intake of statins significantly alter PAD-activity in serum. If yes, are quality and quantity of alterations in PAD-activity associated with specific disease phenotypes of RA, e.g. disease severity, ACPA-status or treatment response profile?
As alluded to above, for all cohorts focusing on treatment of RA (ARCTIC, ULRA and VBfRS), each patient’s clinical response to a specific drug regimen will be evaluated on a regular basis in the course of follow-up visits. Thus, in a first step and according to the criteria commonly used to define treatment success, we will be able to categorize these patients into responders and non-responders. Importantly, this information will enable us to group clinical and biological data obtained from these patients at the time-point of diagnosis into data representing future responders to treatment and data representing future non-responders to treatment. On the basis of this grouping we will be able to ask: i) Is PAD-status determined at diagnosis predictive of if a patient will respond successfully to a specific treatment regimen? ii) Is PAD-status in conjunction with other biological parameters determined at diagnosis predictive of if a patient will respond successfully to a specific treatment regimen?
Collectively, these results provide strong evidence that PAD-activity, and thereof level of citrullination, is heavily impacted in the sera of RA patients. We think (based on preliminary results) that the increased serum PAD-activity might may the constitute risk factor for the RA development via i) excessive citrullination, which contributes to that breakdown of tolerance subsequently lading to the development of RA and ii) the role of PADs in gene expression and subsequently inflammatory events. PAD-activity profile will allow us to stratify the patients that might in the future benefit form a targeted treatment with novel PAD inhibitors, which are currently under development by various companies.

During work on the secreted proteins of the periodontal pathogen P. gingivalis, we noticed that many secreted proteins were predicted to have a glutamine residue at the N-terminal end after cleavage of the signal peptide. In order to check whether this observation had a systematic basis, we used the batch version of SignalP (for gram-negative bacteria) to predict secreted proteins and signal peptide cleavage sites, and then determined the fraction of secreted proteins with a Q residue immediately downstream of the signal peptide cleavage site. Using the batch version of LipoP to eliminate predicted lipoproteins from the set, the fraction of secreted proteins with Q residue downstream of the signal peptide sequence constituted well over 60%. We demonstrated that signal peptidase I client proteins in Bacteroidetes, but not in the entire FCB group or other bacterial phyla, frequently have a glutamine residue immediately downstream of the signal peptidase I cleavage site. We have further shown that Bacteroidetes typically contain an animal- or plant type QC, but not both. The QC is predicted to be an inner membrane anchored lipoprotein (as we confirmed experimentally for the P. gingivalis enzyme) and predicted to be active based on the conservation of key active site residues. Based on these findings, we suggest that many signal peptidase I client proteins are processed on the periplasmic surface of the inner membrane by QC, perhaps immediately after exposure of an N-terminal glutamine residue by signal peptidase I, which is itself a lipoprotein. Proteins modified in this way may remain inserted in the inner membrane, reside in the periplasm, become inserted into the outer membrane, or be secreted to the extracellular milieu. Pyroglutamylation could simply serve to increase protein stability (against degradation by aminopeptidases), particularly in the extracellular milieu. Alternatively, it could facilitate or even be required for protein secretion by T9SSs. Finally, taking into account that human-type glutaminyl cyclase activity is essential for growth of periodontal pathogens, targeting this activity with specific inhibitors should have limited effect on oral and gut commensal bacterial flora which either escape the Q-rule or use plant-type QC. In this way bactericidal activity of developed in this project QC inhibitors should target predominantly pathogenic bacteria implicated in development of periodontitis.

P. gingivalis in aetiology and pathogenesis of CVD, RA and COPD – clinical samples and interventional studies.
We evaluated the presence of P. gingivalis in plaque and immune tissues using an established mouse model of atherosclerosis, the low density lipoprotein knock-out (LDL-/-) mice. We found more bone regression in the P. gingivalis exposed mice. However, the control mice had also some bone regression, causing the absolute difference between the groups to be less evident. These results suggest that in this model of LDL- KO mice on a high cholesterol diet, this alone may induce periodontitis. Analysis of atherosclerotic plaques in the aortic sinuses and found that P. gingivalis ribosomal16S was expressed by real time RT-PCR in the majority of mice exposed to P. gingivalis (4 out of 7). This would suggest that P. gingivalis is present in the atherosclerotic plaque when orally exposed. Despite this there was little difference in plaque size between mice exposed to P. gingivalis compared to control mice exposed only to cholesterol. Both groups had significant atherosclerosis and differences were minimal.
We further analyzed the immune responses towards P. gingivalis by in-vitro studies of cells isolated from these mice. We isolated CD4+ T-cell from spleen and draining lymph nodes of mice exposed to P. gingivalis 8 weeks after exposure. Minimal adaptive T-cell immune responses were evident towards various antigens including live or heat-killed P. gingivalis at different concentrations as well as towards oxidized LDL. These experiments suggest that the adaptive immune response towards P. gingivalis does not play a significant role in the induction of atherosclerosis in this model.
In vitro and ex vivo studies of tolerance breakdown in periodontitis/CVD/COPD and clinical samples analysis.

Neutrophil function in COPD is consistently abnormal compared to age matched healthy controls and showed a wide inter patient variability. We studied 18 COPD patient with and 18 without PD. The data showed no relationship of PD to neutrophil dysfunction in the usual COPD group. Neutrophils from AATD patients with lung disease do not show the same neutrophil defect as usual COPD but are less activated and chemotaxis and has also been assessed in AATD patients with and without PD and again no differences were observed. Assessment of NET production did not relate to PD and bacterial phagocytosis of common lung pathogens in vitro using neutrophils from COPD and AATD patients was not impaired. However, studies of phagocytosis in the presence of opsonising plasma from AATD patients showed impairment for Haemophilus influenza. This effect was not present using plasma from healthy subjects and neutrophils from patients with AATD. The data suggests that AATD plasma contains some blocking agent to opsonophagocytosis.

A total of 2,638 sequences from vascular disease (VD) patients with PD and 859 sequences from VD patients without PD were analysed. A total of 38 out of 40 (95%) vascular samples with bacterial DNA were subjected to further analysis. The final fraction of bacterial sequences, reflecting overall bacterial load, obtained from the vascular biopsies in patients with PD was in average 68 ± 23% at an individual level. In patients without PD, this fraction was significantly lower with an average of 25 ± 26%. The mean diversity of bacterial taxa detected from the vascular biopsies was significantly higher in patients with PD. P gingivalis was detected in the vascular biopsy from only one VD patients with PD. No other bacterial members of the red complex were detected from the vascular biopsies in either group, while checkerboard analysis revealed presence of one to three members of the red complex in seven out of 10 subgingival plaque samples from the PD patients. Patients with PD displayed a wide variety of bacterial DNA identified at the genus level from their vascular biopsies, including some known oral bacterial taxa (e.g. Streptococcus spp., Prevotella sp., and Capnocytophaga sp.) and members of the Enterobacteriaceae family (e.g. Serratia and Klebsiella spp.). Pseudomonas sp. and Propionibacterium acnes were abundant in both groups of patients. Streptococcus, Staphylococcus, and Acinetobacter spp. were also detected in both groups. Not-yet-cultured and environmental bacterial species (e.g. Phenylobacterium and Bradyrhizobium spp.) were also identified from the vascular biopsies. Bacterial cells were observed at the surface of the intravascular plaque lining the arterial lumen, most often co-aggregated as micro-colonies at a distance from each other. The co-aggregated bacteria in each location were most often of uniform morphology. Apparent bacterial cell division with invagination of the cell membrane was seen in some of the SEM images. In conclusion VD biopsies from patients with CP had a higher load of bacteria and higher bacterial diversity compared to the biopsies from patients without CP. Both oral bacteria and bacteria belonging particularly to the gut were abundant among the identified species, while putative periodontopathogens were rarely identified.

In line with Objective 4.2 Partner 1 studied effects of PPAD and consequently, the citrullination of proteins by P. gingivalis on the inflammatory gene expression of fibroblasts and epithelial cells, oral keratinocytes that were infected with wild type P. gingivalis (ATCC 33277) and its isogenic mutant (deltaPAD). Preliminary data pointed to an up-regulation of IL36γ in the presence of PPAD. IL36γ is one of three cytokines referred to as IL36, which are all members of the IL1 family, and like the other IL1 family cytokines IL36 promotes the inflammatory response. IL36 is expressed predominantly but not exclusively by epithelial tissues and keratinocytes in the skin. Microarray analysis revealed that IL36G was the most up-regulated following PG wild type infection. In contrast, it was found that IL36G was the most down-regulated gene following infection with the PPAD knock-out (KO) versus competent bacteria. The lack of PPAD enzyme during infection of fibroblast further resulted in a depletion two IL-2 related genes: positive regulation of IL-2 production and positive regulation of IL-2 biosynthesis process.

A pilot study determining the expression of human and bacterial QCs related to periodontitis and rheumatoid arthritis (RA) was finished and data were analyzed. In a total 30 study participants were included. The patients with RA (20.6±6.4) had statistically significantly (p< 0.05) fewer teeth than the healthy controls (28±0) and CP patients (27.1±3). P. gingivalis was most prevalent in CP. When the GCF levels of biomarkers for inflammation were analysed there was a significant difference for IL-1β levels between the groups.
Quantitatively, there were statistically significant differences of human QC mRNA expression (EDTA blood) between the groups, further analysis revealed an increased expression in RA patients compared to CP patients and healthy controls. Five of all analyzed 30 GCF samples were tested positively for human QC expression. Except for each one sample all samples were obtained from RA patients. In eight of the 13 P. gingivalis positive biofilm samples, PgQC expression was detected.
Additionally the association between dementia and periodontitis was of interest of Partner 11. First, a systemic review incl. meta-analysis was performed in order to assess a difference in clinical periodontal measures between dementia patients and non-demented individuals. From 64 abstracts 43 articles remained for full text reading. Finally, six articles met the inclusion criteria and only four studies provided data suitable for meta-analysis. Each chosen parameter periodontal probing depth (PPD), bleeding on probing (BOP), gingival bleeding index (GBI), clinical attachment level (CAL) and plaque index (PI) revealed a statistically significant difference between dementia patients and non-demented individuals. Although the current findings indicate that demented patients show significantly higher clinical periodontal indices in comparison to systemically healthy individuals, more epidemiological studies including a high numbers of participants, the using of exact definitions both for dementia and chronic periodontitis and adjusted for cofounders are warranted.
Within a collaboration, samples (GCF, subgingival biofilm, serum, liquor) obtained from 20 younger (30-70 years old) patients diagnosed with Alzheimer disease and 20 respective controls were analyzed. The hypothesis of this study was that there is a local production (due to an infection) of antibodies against P. gingivalis in the brain. Not any difference related to clinical periodontal indices, bacteria in subgingival biofilm was found. Moreover, all serum and liquor samples were tested negatively for the major periodontopathogens. This result was confirmed by partner 9. Analysis of antibody levels against P. gingivalis. A. actinomycetemcomitans and Treponema sp. did not reveal any hint for a local antibody production. Grouping the study participants related to the severity of dementia showed no differences of bacteria and laboratory variables but a worst oral hygiene combined with more inflammation and tooth loss in dependence of severity of dementia.

Characterization of periodontal pathogen-derived QC, inhibitor identification, design, and synthesis.

The QC from P. gingivalis was successfully cloned, expressed and purified. The activity of the en- zyme was proven and it was used for the characterization of the inhibitory potency of potential pgQC inhibitors.
Significant progress in the design of pgQC inhibitors was made. Novel MBGs could be discovered. The derivatization of the novel imidazo[4,5-b]pyridine led to a novel class of inhibitors and the results gained earlier in the benzimidazole series could be transferred to this type of inhibitors. The synthesis of inhibitors based on other MBGs is ongoing and focus of the current work. The structural modification of the imidazo[4,5-b]pyridines led to inhibitors exhibiting Ki values in the lower nM-range. Selected compounds have already been evaluated in bacterial culture. In co-operation with partner 11, it was possible to establish an assay for the determination of the micro-biological relevant values MIC and MBC and first compounds could be investigated.
After the successful expression and purification of the three target enzymes further investigations concerning the characterization of the protein were made. Further characterization of recombinant bacterial QCs were performed by applying CD spectroscopy. The CD spectra were virtually identical for all bacterial QCs, which supports the strong similarity of the globular domains between PgQC, PiQC and TfQC. Spectrum of proteins with high α helical content exhibits two typical minima at 208 nm and 222 nm in their secondary structures which were also observed for PgQC, PiQC and TfQC. Additionally a calculation of quantities of α -helix, β-sheet, turn, and random structure according to the method of Yang et al. indicating similar folding patterns of all three proteins and suggesting an α / β topology.
Furthermore CD spectra can be used for analyzing of altered secondary structure in dependency of different environmental conditions. Therefore changes in secondary structure of recombinant proteins were observed using CD with increasing temperature from 20°C to 80°C. The spectra of PgQC and TfQC changed at a temperature at about 40°C indicating the starting point of denaturation. The spectrum of PiQC changed at a temperature at about 45°C. All three proteins were fully denatured at 50°C. A four parameter fit was apply to elucidate the exact temperature transition and indicates a similar thermal stability of bacterial QCs; PgQC and TfQC at 43°C and 41°C and for PiQC at 47°C.
For the drug development process the knowledge of the localization of the target enzyme is essential to optimize the physico-chemical features of the molecule. Therefore the following investigations were done. The generated polyclonal antibody exhibits high affinity and specificity against PgQC, whereas PiQC and TfQC could also been detected with presumably lower affinity.
In co-operation with partner 11 (UNIBE, Prof. S Eick) the antibody serum was then used for Cryo-EM investigations, where the antibodies were labeled with silver. P. gingivalis was used after freeze fracture and the bacterial QCs could be detected (Figure 4). Unfortunately the results do not fully clarify the localization of the target enzyme, whereas they strongly support the hypothesis of being a periplasmic enzyme. In silico analysis of the open reading frame of PgQC revealed a putative signal sequence. This indicates a localization of PgQC outside the cytoplasm in P. gingivalis. C-terminal fusions of PgQC inclusive “native” signal sequences with alkaline phosphatase (PhoA) were constructed to investigate the localization of PgQC. Thereby fusions with alkaline phosphatases are exclusively active if it is localized in the periplasm. For expression of seqPgQC-´PhoA fusion protein, a recombinant vector was transformed into CC118 along with a helper plasmid. Expression of seqPgQC-´PhoA protein was induced and finally PhoA activity was determined in permeabilized cells. The expression of seqPgQC-`PhoA, resulted in a high phosphatase activity. Data given above strongly suggests a periplasmic localization of the bacterial QC in P. gingivalis.
The last period of the project was focused on the design of inhibitors. Different strategies were investigated where uptake systems of the bacteria could be used as targets. For instance porins, cell penetrating peptides or iron transporters are possible systems to enhance the uptake of the inhibitor. Thereby we used our established co-operation with the University of Bern (partner 11, UNIBE, Prof. S. Eick). Here we tested the in vivo activity of the compounds in bacterial models and we were able to correlate this data with in vitro data (Ki-values). One approach is the targeting of bacterial porins. For this a different logD of the compounds is suggested compared to the values that are normally wanted. So, a logD-range of -1 to -5 for compounds is proposed to have a positive effect on gram negative bacteria. Based on known active inhibitors more hydrophilic compounds were synthesized and the in vitro activity for bacQCs could be kept also for the modified compounds. Unfortunately the approach does not work concerning the translation of the enzymatic activity to the activity against the pathogens. An alternative approach is the conjugation of the inhibitors to peptides, which are known to be able to penetrate the membranes and that are bactericidal by itself (BioMed Res. Int. (2015), Article ID 909870). Both effects, that of the QC-inhibitor and that of the bactericidal peptide, together should strongly enhance the impact of the compounds. The compounds were synthesized by using “click” chemistry and the QC-inhibitor part and the peptides were successfully conjugated by the formation of an triazole via a so called Copper(I)-Catalyzed Azide-Alkyne Cycloaddition (CuAAC). Also in that case it was possible to keep the in vitro activity of the resulting compounds compared with the original small molecule. The approach does not lead to compounds with a clear selectivity against the QC-bearing bacteria strains. Obviously the effect of the bactericidal acting peptides alone has the most prominent effect to most of the tested strains.

P. gingivalis as the major pathogen has a unique recognition system of the essential iron. In contrast to other strains, P. gingivalis recognizes a heme group that is not necessarily occupied by an iron ion, whereas in other species, the iron itself is working as the recognition pattern (Org. Biomol. Chem. 2009, 7, 2855-2863; Org. Biomol. Chem. 2015, 13, 98–109.) Conjugates of a active glutaminyl cyclase inhibitor, a solubility enhancing substructure and a porphyrin system were synthesized and tested. The QC-activity was lowered by one order of magnitude compared to the non-conjugated QC-inhibitor, but compounds of this scaffold were able to selectively inhibit the growing of the targeted bacteria (Table 3). Therefor this approach is very promising for further development of potential drug molecules. The resulting MICs of the conjugates are comparable or even enhanced, compared with literature values for Metronidazole that was also linked to a porphyrin system.
Several approaches were tested to overcome the lacking in vitro – in vivo correlation of the compound activity. Thereby it was possible to create compounds that are able to selectively kill only bacteria with a type II Glutaminyl Cyclase.

Crystal structures for glutaminyl cyclases of P. gingivalis and T. forsythia were obtained with reasonable resolutions. These crystal structures give a first insight into the binding mode of the inhibitors are are suitable for further structure based approaches.
Compounds were tested in several levels of a typical drug screening cascade that is established at IZI-MWT. For the tested compounds no cytotoxicity could be observed and an acceptable bioavailability could be determined for 2 tool compounds. A test in a reliable animal model was not possible because of the back draws in the drug development process. Meanwhile the compounds are ready to use in POC-tests that might be done in co-operation with partner 2 after the project period is finished.

Structure-function studies at the molecular level.
Two peptidases, gingipain K (Kgp) and R (RgpA and RgpB), which differ in their selectivity after lysines and arginines, respectively, collectively account for 85% of the extracellular proteolytic activity of P. gingivalis at the site of infection. Therefore, they are promising targets for the design of specific inhibitors. Partner 5 structurally characterized Kgp in its complexes with TCLK and KYT-36, as well as characterized structure and function of RagB outer membrane protein.
It has been suggest bacterial PAD as the mechanistic link between P. gingivalis periodontal infection and rheumatoid arthritis and we have solved three structures of PPAD, which have contributed to shed light on the working mechanism of this enzyme.
We determined the crystal structures of the zymogens of two of these (P. abyssi proabylysin and M. jannaschii projannalysin), which are soluble and, with ~100 residues, constitute the shortest structurally characterized MPs to date.
Karilysin, from the human periodontopathogen T. forsythia, is the only bacterial MMP to have been characterized biochemically to date. It shares the catalytic domain with eukaryotic forms, but none of the flanking domains. Instead of the consensus MMP pro-domain, it features a 14-residue pro-peptide—the shortest reported for a metallopeptidase—, which lacks cysteines. We have unveiled the activation mechanism of karilysin based on crystal structures.

P. gingivalis, a dysbiotic member of the human oral microbiome, employs a newly desccribed type-IX secretion system (T9SS) for protein secretion. This is a multi-component translocon encompassing at least 12 Por proteins whose recognition signal is a conserved C-terminal protein domain (CTD), which is removed from the cargos by sortase PorU during translocation. We identified a novel component of T9SS, PorZ, which is loosely associated with the cell surface and is essential for surface exposure of PorU. We further solved the crystal structure of full-length PorZ, which revealed two seven-fold β-propeller domains. PorZ also encompasses a C-terminal seven-stranded β-sandwich, which conforms to the canonical T9SS CTD architecture. Consistently, we further found that PorZ is itself transported to the cell surface via T9SS as a full-length protein with its CTD intact, independently of the presence or activity of PorU sortase. Analysis of mutants expressing PorZ with oligohistidine insertions/substitutions within the C-terminus and the inter-domain linker preceding CTD generally revealed plasticity to accommodate modifications without affecting the secretory phenotype. Taken together, the present results shed light on the architecture and possible function of a novel component of T9SS only found in Bacteroidetes.

To counteract endopeptidases from the host or microbial competitors, T. forsythia possesses a serpin-type proteinase inhibitor called miropin. While serpins from animals, plants, and viruses have been widely studied, those from prokaryotes have received only limited attention. Here we show that miropin uses the serpin-type suicidal mechanism. We found that similarly to a snap trap, the protein transits from a metastable native form to a relaxed triggered or induced form after cleavage of a reactive-site target bond in an exposed reactive-center loop. The prey peptidase becomes covalently attached to the inhibitor, is dragged 75Å apart, and is irreversibly inhibited. This coincides with a large conformational rearrangement of miropin, which inserts the segment upstream of the cleavage site as an extra β-strand in a central β-sheet. Standard serpins possess a single target bond and inhibit selected endopeptidases of particular specificity and class. In contrast, miropin uniquely blocked many serine and cysteine endopeptidases of disparate architecture and substrate specificity owing to several potential target bonds within the reactive-center loop and to plasticity in accommodating extra β-strands of variable length. Phylogenetic studies revealed a patchy distribution of bacterial serpins, incompatible with a vertical descent model. This finding suggests that miropin was acquired from the host through horizontal gene transfer, perhaps facilitated by T. forsythia's long and intimate association with the human gingiva.

Together with genes of two component signaling (TCS) system regulating expression of structural proteins of T9SS we have cloned (or synthetized) 14 genes and expressed them from several different expression systems. Up to date we have succeeded to obtain large amounts of high purity recombinant PorU (PG0026, sortase), PorV (LptO, PG0027), PorL (PG0289), PorN (PG0291), PorX (PG1019) and PG1604. Already, diffracting crystals of PorU and PG1604 have been obtained.
In addition to expression, purification and crystallization of T9SS components we have studied subcellular localization of PG1604 and reciprocal interactions between different components. Interestingly we have found that PorU and PorV form stable heterotrimmers and only in the presence of PorU, the recombinant PorV folds correctly. Because PorV is the integral OM beta-barrel protein while PorU is exposed on the bacterial surface this data suggest functional organization of these two components of T9SS. Alike PorU, also PG1604 is expressed on the P. gingivalis surface. Such location of these components is compatible with the presence of the CTD domain in their structure. However, in contrast to T9SS cargo proteins CTD is not cleaved off during apparent translocation of PorU and PG1604 across the OM.
Inter-alpha-trypsin inhibitor (IαI) is a proteoglycan composed of the protease inhibitor bikunin and two heavy chains. The proteins are present in plasma in relatively high concentrations, although their function here remains unclear. IαI interacts with tumor necrosis factor stimulated gene-6 protein (TSG-6) and in concert these proteins play a role at different physiological conditions including infertility (mice studies) and inflammatory diseases (e.g. arthritis). Previous studies have found that heavy chain 2 in IaI is citrullinated in synovial fluid from RA patients. We have shown that carbamylation and citrullination of IaI reduce the affinity for heparin. Moreover the PAD mediated citrullination reduce the interaction with TSG-6 and prevents the stabilization of the extracellular matrix.

Bioinformatics analysis, prospective study design and appraisal of future value of knowledge.
Topological analysis using standard measures in in the field of biological networks has shown a close relationship between TRIGGER pathologies. The enrichment analysis in pathological conditions showed as expected a close relation between the 4 pathologies. The closest relation, that is the higher number of common proteins, is between RA and PD. COPD and ATH appear less related with PD and RA and also between them, being ATH the one sharing less number of common proteins. Noteworthy that in RA, COPD and ATH network analysis the closest relation is seen with PD. Noteworthy that in RA, COPD and ATH networks the closest relation with any TRIGGER pathology is always with PD. Distance analysis showed a similar result as the enrichment analysis. PD and RA are closer than COPD and ATH.
Network-based mathematical models have been built for the different networks (PD, RA, COPD and ATH) developed with the purpose of identifying key proteins linking the different pathologies. Models developed are accurate in terms of describing the current molecular knowledge. Some models present an elevated complexity, thus need to be built from different starting points and with a higher number of iterations to be able to cover the whole spectrum of the disease. Network-based mathematical models built on the bases of the networks of (PD, RA, COPD and ATH) are accurate in terms of describing the current molecular knowledge and new analytical measures to evaluate network-based mathematical models have been developed, such as: mean MoA representation, MoA comparison, and model reversion.

To further analyse the links between the pathologies and try to identify possible key links between them network-based mathematical models for PD, RA, COPD and ATH were built based on available molecular information D 7.1. Here we describe the outcome of the analysis of those models together with a second modelling strategy based on artificial neural networks (ANNs) used to model complex relationships between inputs and outputs.
The key PD proteins identified with higher potential of contributing to ATH, based on a mathematical network model, highlight different possible connections. Most of the proteins are linked to the induction of innate immunity response and inflammation, like: NOD-like receptor signalling pathway and Toll-like receptor signaling pathway.
The key proteins APOB (more active) and APOA1 (less active) and suggest that a possible linkage could be through periodontitis contributing to dyslipidemia [PMID: 20835400] which in turn have an important role in in atherogenesis. Increased expression of apoE, apoM, apoC and SAA, and decreased expression of apoA1- effects that are associated with an increased uptake of cholesterol and triglycerides into vessel walls [2011, Infection&Immunity 79:1489-1497].
Interestingly, it appears that the prooxidant and inflammatory microenvironment in the vicinity to the atherosclerotic lesion, ApoA1/HDL are subjected to modification resulting in neo-epitopes inducing the production of proatherogenic ApoA1-specific IgG antibodies. In patients with periodontitis the anti-ApoA1 antibodies levels showed association with pathological ankle-brachial index (a parameter to predict the severity of peripheral arterial disease) in subjects younger than 50 years [doi:10.1038/labinvest.2016.56]. Such a process would resemble the citrulination of certain proteins seen in RA and the use of ACPAs as diagnostic measures. In fact, citrulinated APOA1has been identified in some instances, citrullinated ApoA1 is present in ulcerative colitis serum samples. There is data providing evidence that ApoA1 citrullination strongly correlates with disease severity in a mouse model of ulcerative colitis and could serve as a potential biomarker for this and potentially other diseases [Patent: US 9347948 B2]. In fact the use of APOA1 for PD is under patent [Patent: WO2014037924 A2].
There are some initial studies examining the role of cit-apoA1 in RA. They could show a higher proportion of cit-apoA1 in the serum associated with impaired HDL efflux capacity (p=0.039) but they could not show an association between antibodies against cit-ApoA1 with HDL efflux capacity (p=0.136) [Abstract Number: 1510 Arthritis & Rheumatology. 67():1898–1899, OCT 2015].
The key proteins LPL and APOE suggest that a possible linkage could be through periodontitis contributing to dyslipidemia [PMID: 20835400] which in turn has been suggested as an upstream cause of inflammation and oxidative stress contributing to lung pathologies such as those associated with COPD [PMID: 19752857].
NOX1 and RAC1 are some of the key proteins identified that could link PD and COPD through increased ROS response, but interestingly also dyslipidemia increases the risk for overproduction of reactive oxygen species (ROS), thus we could be seeing here additive contributions to COPD pathogenesis.
Another possible connection would be through the toll-like receptor signalling, many of the key proteins identified like Myd88 have a role in this pathway. A potential additional contribution to this linkage could be through ROS production increasing blood levels of oxidized low-density lipoprotein (OxLDL), which some studies have demonstrated that modulates toll-like receptor expression and signalling.
Proteins were selected when the probability of affecting the target disease is above the mean+STD (calculated from the probability of affecting any of the 200 pathologies contained in our data base). In this way proteins that have a lot of links and relate with many pathologies are discarded. From this set of proteins we also discarded those that also show a relation with the pathologies selected as negative control). That is for RA, spondylitis (AS) and osteoarthritis (OS), for COPD asthma and for ATH, atrial fibrillation, cardiac valve disease and cardiac myopathy.
The toll-like receptor pathway is again represented (TLR4 and TLR2), innate immunity (LBP). TGFB1 has a role in TH17 differentiation pathway, cytokines secreted by Th17 cells play an important role in eliminating harmful microbes but unregulated expression of these cytokines may also contribute to the pathogenesis of a number of autoimmune diseases. But there is only one that appears in the 3 measurements, NCF1. NCF1 (Neutrophil Cytosolic Factor 1) is a cytosolic subunit of neutrophil NADPH oxidase which in turn is a multicomponent enzyme that is activated to produce superoxide anion. This protein is strongly related with the production of ROS and innate immunity [Am J Pathol. 2016 Mar;186(3):659-70], it has also been related with RA pathogenesis [J Immunol. 2009;183(2):874-81], emphysema [The Journal of Pathology 235(1)] and atherosclerosis [J. Clin. Invest. 108(10):1513–1522; 2001]. But most significant probably is the association of a missense variant in NCF1 with susceptibility to multiple autoimmune diseases [Nature Genetics (2017). DOI: 10.1038/ng.3782].
We have i) identified key common and complementary pathways that could contribute or have a triggering effect linking PD with RA, ATH and COPD pathologies. The conjunction of effects identified could contribute cumulatively influencing the host immune response. Our results highlight more atherogenic lipid profile / dyslipidemia and innate immunity thought toll-like receptor signalling pathway as one of the possible associations.
ii) Highlighted PD related proteins with potential contributions to RA, ATH and COPD pathologies: enhanced ApoB/ApoA1 ratio, inhibition of APOE and LPL and induction of proteins of the Toll-like receptor pathway (TLR4, TLR2, Myd88). Noteworthy, NCF1 (strongly related with the production of ROS and innate immunity) a missense variant has been recently associated with susceptibility to multiple autoimmune diseases.

Potential Impact:
In our opinion it is impossible to overestimate the impact of this project if it is carried out successfully. The morbidity and mortality of CVD, RA, and COPD individually makes each of the diseases the major factor debilitating human health after cancer. The patients’ sufferings are beyond imagination of a healthy human being and only matches the costs of treatment of these devastating chronic diseases. Therefore any, even incremental, improvement in prevention and treatment of CVD, RA, and COPD will save millions of human beings and contribute to quality of life of patients. Such chance is clearly described in this project aimed to find and characterizing the causative relation between PD, which is an infectious disease and these non-communicable chronic inflammatory diseases. They have a lot of in common, but PD is neglected, both by patients and health authorities. Of course, losing teeth does not seem as bad as dying of a heart attack, or being forced into vegetative state by a stroke or crippled by RA or COPD. Clinical studies clearly showed that PD correlates with CVD, RA, and COPD but did not answer the critical question if such co-morbidities are reason of shared risk factor or the chronic diseases are “fuelled” by bacterial infections responsible for development and progression of PD. By answering this question this project will impact on (i) the European research environment and novel, innovative approaches to (ii) diagnosis, (iii) prevention, (iv) treatment, and (v) and management of diseases. This will reduce mortality and ameliorate quality of life of CVD, RA, and COPD patients. All of these will be achieved based on the knowledge of mechanisms beyond the causative links between specific pathogen driven periodontal disease and CVD, RA, and COPD revealed by research program outlined in this project. This is described in more details below:
Impact on the European research environment and novel, innovative approaches.
The project itself is based on a very innovative idea that to reveal and understand, in mechanical terms, the causative relation between PD and CVD/RA/COPD we need to focus on P. gingivalis, the main and essential pathogen in human PD. Furthermore, we intend to find which combination of P. gingivalis virulence factors contributes to pathogenicity of CVD/RA/COPD. This novel, ground-braking approach will yield European scientists and biotech companies a cutting-edge in future research of causative relation and co-morbidity between disease which is treatable (PD) and diseases which elude medical intervention (CVD and COPD) or are associated with extremely high costs of treatment (e.g. TNFα inhibitors in RA).
Impact on diagnosis.
PD is arguably the most prevalent chronic infection-driven chronic inflammatory disease. The disease is relatively easy to diagnose and characterize with respect to microbial pathogenic flora. Therefore finding the bacteria-derived factor(s), which predisposes PD patients to develop or aggravate CVD/RA/COPD will allow providing earlier appropriately aimed care to prevent or slow down the life-threatening systemic diseases.
Impact on prevention.
The knowledge that PD is causatively linked to CVD/RA/COPD should have a tremendous impact on strategies to prevent morbidity and mortality associated with the systemic diseases. Here, it must be strongly reiterated that PD is a treatable disease. In addition, it is firmly established that appropriate oral hygiene in conjunction with professional dental plaque removal every 6 months can prevent PD development. Therefore, it can be easily envisioned that appropriate encouragement from dentists backed by a media-stimulated awareness program and support from insurance companies should reduce the prevalence of PD in the general population. This should significantly reduce incidence of CVD/RA/COPD or at least decrease severity of these diseases saving patients life and suffering. It is tempting to speculate that this preventive strategy will have comparable impact on CVD management as drugs lowering cholesterol level and controlling blood pressure. In addition, a nice smile and eliminated oral odour can be expected as a positive “side effect” of such preventive strategy.
Impact on guideline development.
Self-administered plaque control programs without professional supervision were shown to be inconsistent in providing long-term inhibition of gingivitis. By providing the knowledge of causative relationship between PD and RA, COPD and CVD development we will impact on clinical guideline/recommendations development for healthcare professionals (dentist, oral surgeons). As mentioned in the dissemination section (p. 56) dentists, periodontists, and oral surgeons will be one of the TRIGGER most important target groups.
Impact on treatment.
One significant outcome of this project will illuminate on the beneficial effects of periodontal treatment on clinical parameters of CVD/RA/COPD. As described above periodontal disease is the treatable disease and it has already been shown that successful periodontal treatment improves management of CVD and RA. Unfortunately, PD treatment can be experienced as painful, time consuming and expensive since it requires mechanical removal of subgingival bacterial plaque by scrubbing with curate. This treatment in conjunction with systematically administrated antibiotics needs to be repeated. It must be noted that antibiotics are only partially effective against periodontal pathogens and repeated treatment with antibiotics can contribute to rise in antibiotic resistance in other human pathogens like S. aureus, enterpatogenic E. coli and Salmonella species. Therefore, development of an alternative to antibiotics based on inhibitors of the QC enzyme or compounds interfering with assembly of virulence factors on P. gingivalis may revolutionize PD treatment. This is especially important facing the fact that up to 30% of the adult population suffers from moderate to severe periodontitis. As discussed above in Impact on prevention such new treatment will benefit not only periodontitis patients but also patients suffering from CVD or RA or COPD.
Impact on aging related problems.
EU has to currently cope with aging of significant part of its population. Life expectancy increased by 8 years between 1960-2006 and is expected to continue to grow another 5 years by year 2050 (Commission Communication of 12 October 2006 "The demographic future of Europe – From challenge to opportunity). Elderly people are very often in very delicate health conditions and have multiple co-existing diseases. They are particularly at risk of adverse drug reactions and drug-drug interactions that are becoming major cause of death. Prevention or decreasing morbidity of CVD, RA and COPD by early and efficient treatment of PD will not only prevent mortality due to the drug-drug interaction but also relief the pressure on hospital care significantly reducing cost of long term treatment and hospitalization. Finally, the efficient treatment of PD preventing tooth mobility and lost, by itself, will significantly increase the quality-of-life of aging population.
Impact on accepting PD as a serious disease, which can be treated
Very high prevalence of PD, slow progress of disease and relatively mild consequences (until tooth loss is inevitable) seems to desensitize patients, public opinion and health authorities to morbidity caused by PD. Occupied with heart diseases, cancer and stroke, leading causes of death in industrialized nations, we tend to ignore the reality that PD is the most prevalent infection-driven chronic inflammatory disease of mankind. Just by publicizing, especially via media releases the causative link between PD and life-threatening diseases the project should encourage people to seek evaluation of gingival health, take preventive measures (oral health and regular dental plaque removal) and subject to periodontal treatment if necessary. If we succeed in this single point the social and economic impact of this project will be tremendous.
Impact on the costs of CVD, RA, and COPD treatment.
We need to remember that the average annual cost of treatment of RA with biologics is around 15 000€, and both COPD and CVD are major causes of morbidity and mortality in Europe. Low cost prevention of all after mentioned diseases by treatment of periodontitis using either known procedures or developing novel, more effective drugs (QC inhibitor) might have profound effect on epidemiology and health economics across Europe.
To sum up, the successful projects will increase our knowledge of the causative links between specific pathogen driven periodontal diseases and non-communicable diseases, including CVD, RA and COPD. Obtained results will contribute to better prevention and management of patients suffering from such co-morbidities. Prevention of periodontal diseases at early stage of life may significantly decrease number of new cases of CVD, RA and COPD. On the other hand treatment of chronic periodontitis in patients that already show symptoms of the diseases should have positive impact on their quality of life, decrease number of health complications and significantly lower amount of exacerbation, reducing overall cost of treatment. The expected impact includes design of novel treatment based on inhibition of an essential enzyme for periodontal pathogen vitality. This will reduce mortality and morbidity associated with these diseases and significantly improves quality of life of patients.

List of Websites:
Web page of the project can be found online under

Reported by

Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top