Ratio of Pro-Resolving and Pro-Inflammatory Lipid Mediator Precursors as Potential Markers for Aggressive Periodontitis

Aggressive periodontitis (AgP) is a rapidly progressing type of periodontal disease in otherwise healthy individuals which causes destruction of the supporting tissues of the teeth. The disease is initiated by pathogenic bacteria in the dental biofilm, and the severity of inflammation and attachment loss varies with the host response. Recently, there has been an increased interest in determining the role of lipid mediators in inflammatory events and the concept of pro-inflammatory and pro-resolution lipid mediators has been brought into focus also in periodontal disease. The present study aimed to determine the profile of omega-3 or n3- as well as omega-6 or n6- polyunsaturated fatty acids (PUFAs) and PUFA-metabolites of linoleic acid, arachidonic acid (AA), eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) in gingival crevicular fluid (GCF), saliva and serum in AgP patients and healthy controls. In total, 60 selected n3- and n6-PUFAs and various PUFA metabolites were measured using high performance liquid chromatography-tandem electrospray ionisation mass spectrometry (HPLC-ESI-MS-MS). Of these, 51 could be quantified in this study. The concentrations of the majority were low in saliva samples compared with serum and GCF, but were mainly higher in AgP patients compared with healthy controls in all three kinds of sample. Ratios of n3- to n6-PUFAs (DHA + EPA)/AA were significantly lower in the GCF of AgP patients than in the healthy controls. Furthermore, various ratios of the direct precursors of the pro-resolution lipid mediators (precursors of resolvins and protectins) were calculated against the precursors of mainly pro-inflammatory lipid mediators. These ratios were mainly lower in GCF and saliva of AgP patients, compared with healthy controls, but only reached significance in GCF (P<0.05). To conclude, the ratios of precursors of pro-resolution/pro-inflammatory lipid mediators seem to be more relevant for describing the disease status of AgP than the concentration of specific lipid mediators.


Introduction
Aggressive periodontitis (AgP) is a rapidly progressive form of inflammatory periodontal disease. If not treated, AgP may result in tooth loss even in young people [1]. The initiating factors of periodontal disease are bacteria and bacterial products which form a biofilm covering the tooth surface in the subgingival area. In susceptible individuals, the host-mediated response to the pathogenic flora is subsequently responsible for the periodontal tissue destruction [2,3]. The course of the disease is modified by different environmental factors, and the patient's genetic make-up contributes to patient susceptibility, in particular in the case of aggressive forms of periodontitis [4][5][6].
Persistence of inflammation or the failure of tissue to return to the normal state favours tissue destruction [7]. Thus, resolution of inflammation is essential to the process of regaining tissue homeostasis after infection, and it has become clear that this is not a passively, but an actively coordinated process involving several biochemical pathways, enzymes and mediators [8,9]. Over recent years, there has been an increased interest in determining the role of lipid mediators in anti-inflammatory events, and the concept of pro-inflammatory and pro-resolution lipid mediators has also been brought into focus in periodontal disease pathogenesis [10].
Host factors, including both immune components and resident cells, contribute to tissue destruction via synthesis and release of different pro-inflammatory molecules [11]. The pathogens trigger the cell membrane of the immune cells, leading to the release of free fatty acid, mainly arachidonic acid (AA). Pro-inflammatory lipid mediators such as thromboxanes (TX), prostaglandins (PG) and leukotrienes (LT), as well as anti-inflammatory lipid mediators such as lipoxins (LX), are among the products generated, mainly via AA-metabolism.
Arachidonic acid is metabolized to biologically active eicosanoids through the cyclooxygenase pathway (COX-1 and COX-2), which results in production of various PG-like PGE 2 and PGD 2 [12,13]. Other mediators such as LT, particularly LTB 4, arise through the conversion of hydroxyeicosatetraenoic acid (HETE) by 5-lipoxygenase (5-LOX). The AA-derived prostanoids and LT are associated with the destruction of collagen and bone resorption occurring in periodontal disease [14,15]. Lipoxygenases involved in AA, EPA and DHA metabolism are the 5-LOX, 12-LOX and 15-LOX. The human 12-LOX can be classified into two types, the platelet type and the epidermis type [16,17] while the human type can be classified into the reticulocyte type (15-LOX-1) [18], leukocyte type [19], and the epidermis type   [20,21]. Prostaglandins are biological mediators involved in inflammation and have been implicated as stimulators of bone loss [8]. Levels of PGE 2 are significantly elevated in the gingival crevicular fluid (GCF) of patients with periodontal disease. High levels of PGE 2 are associated with disease aggressiveness and constitute a reliable indicator of current clinical periodontal destruction [14,15]. High levels of LTB 4 have also been found in periodontal inflamed tissues and are considered to be an indicator of periodontal inflammation [14].
Omega-6 (n6) and omega-3 (n3) polyunsaturated fatty acids (PUFAs) are substrates for the production of various eicosanoids and docosanoids (Figure 1). The mono-hydroxylated PUFA metabolites of AA, 15-HETE and 5-HETE are direct precursors for LXs, which, in contrast to PGs and LTs, attenuate the inflammatory effect [22,23]. The major n3-PUFAs are eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), which are the precursors of various novel pro-resolution lipid mediators like resolvins (Rv), protectins (PD) and maresins (MaR) which are responsible for the active resolution of inflammation [24][25][26][27]. The PUFAs EPA and DHA are substrates for various classes of Rv of the E (originating from EPA) and D (originating from DHA) series, respectively [28]. Interestingly, local application of RvE1, which is one of the bioactive pro-resolution mediators, reduced inflammation and promoted regeneration of pathologically lost tissues in a rabbit model of periodontal disease [29]. In humans, the daily dietary supplementation of n3-PUFAs in combination with aspirin was reported to reduce the frequency of deep periodontal pockets and inflammatory lipid mediators in saliva [30].
Little is known about the presence and the amount of different bioactive lipid mediators in oral fluids during periodontal inflammation and its resolution. In particular, a comparison between their levels locally near the inflamed site (in GCF and saliva) and systemically (in serum) has, to our knowledge, not been done. In this report, we analysed eicosanoids and docosanoids in GCF, saliva and serum in a group of AgP patients and healthy controls in an attempt to understand better the roles and relationships of these lipid mediators.

Results
In the present study the levels of various PUFAs and monohydroxylated PUFA metabolites and pro-resolution and proinflammatory lipid mediators in GCF, saliva and serum samples from AgP patients and healthy controls were measured by high performance liquid chromatography -electrospray ionization
The GCF samples were additionally analysed by ELISA for PGE 2 and LXA 4. The levels of PGE 2 were significantly increased in patients (mean 5.661.3 ng/ml) compared with healthy controls (mean 4.760.7 ng/ml) (P = 0.002), which was in line with the HPLC-ESI-MS-MS results. By use of ELISA, the LXA 4 concentrations between patients and healthy controls was slightly lower in GCF of AgP compared with healthy controls (P = 0.049) ( Figure 2).

Saliva analysis
From the PUFA levels of LA, AA, EPA and DHA were found at slightly higher concentration in saliva samples of AgP patients compared with controls ( Table 2). Lipid derivatives originating from the 5-LOX pathways, in particular LTB 4 (P = 0.002), 20-OH-LTB 4 (P = 0.048), 5-HETE (P = 0.017), 5-oxoETE and from the 15-LOX pathways, 15-HETE (P = 0.004), 15-oxoETE and 17-HDHA, were found in higher concentrations in AgP patients ( Table 2). In contrast, 12-oxoETE was present in significantly lower concentrations in the patient group. The concentrations of PGE2 in saliva in AgP patients and healthy controls were not significantly different.

Serum analysis
As with GCF, the levels of the fatty acids in serum including AA, EPA, DHA and LA, were higher in AgP patients (Table 3) Table 3). The concentration of the lipid mediator d15d12PGD2, a COX metabolite, was increased in the serum samples of AgP patients, as well as 20-COOH-AA, 20-HDHA and 11-HETE (Table 3). In general, concentrations of various PUFA metabolites were much higher in serum than in saliva and GCF.

Ratios of precursors of pro-resolution/pro-inflammatory lipid mediators and LOX/COX-pathway analysis
There were considerable variations of the concentrations of DHA, EPA and AA between groups in GCF, saliva and serum analysed in this study (Table 1, 2 and 3). The ratio of the n3-PUFAs (DHA and EPA) concentration vs. the concentration the n6-PUFA (AA) in GCF was significantly lower in AgP compared with the control group (P = 0.004) ( Figure 3A). By contrast, there was no significant difference in these ratios in saliva and serum ( Figure 3B and C).
In addition the ratio of the direct precursors of pro-resolution lipid mediators (14-HDHA, 17-HDHA and 18-HEPE), which are precursors of resolvins and protectins, was calculated against various precursors of the AA-derived and mainly pro-inflammatory derivatives (5-HETE, 12-HETE and 15-HETE) as well as ratios of sum of HEPEs and sum of HDHAs vs. the sum of HETEs were calculated ( Table 4). The ratio of precursors of pro-resolution to pro-inflammatory lipid markers (HEPE/HETE and HDHA/ HETE) were mainly reduced in the GCF, saliva and serum of AgP patients in comparison with healthy controls but was just significant in GCF samples (P,0.05) ( Table 4).
The sum of 5-, 12-, 15-LOX and COX pathway originating metabolites were mainly upregulated in serum, GCF and saliva samples from AgP compared with healthy controls (Table 4).

Discussion
To draw a general picture of the contribution of bioactive lipid mediators in periodontal health and disease, lipid mediators were profiled using the HPLC-ESI-MS-MS methodology for serum, saliva and GCF samples from AgP patients and healthy controls. The precise concentration of PUFAs, PUFA metabolites and further lipid mediators determined using lipidomic technology leads to a deeper understanding of the role of lipid mediators in the mechanism of periodontal disease [31]. In this study, we report   [32] also reported of significantly higher levels of EPA and DHA in serum of patients with generalized chronic periodontitis compared with that of patients with gingivitis. Whereas there were significantly lower EPA/AA and DHA/AA ratios in GCF of the AgP patients compared to healthy controls, this was not the case in serum and saliva samples in which approximately similar ratios in AgP patients and controls were demonstrated. In a previous study it was reported that in periodontitis patients with bone loss, the serum levels of fatty acids from the n6-PUFA pathways, which were obtained after saponification, were higher than in healthy individuals, whereas the opposite was observed with fatty acids of the n3-PUFA pathway [22]. The bone loss was linked to an imbalance between the two pathways with mainly an increase in AA and a decrease in EPA and DHA.
Recently the active resolution of inflammation was recognized as a process driven by lipoxins originating from AA and novel compounds from EPA and DHA known as resolvins, which have a potential to dampen or resolve the damaging aspects of the inflammatory response. LXA 4 is an endogenous lipid mediator which plays an important role in the local resolution processes [7,33]. In the present study, we found LXA 4 concentration to be increased in patients compared with healthy controls when measured by HPLC-ESI-MS-MS, but this was significant only in serum samples. LXA 4 in GCF was also determined by ELISA. Using this technique, LXA 4 was identified at slightly higher concentrations in healthy controls (P = 0.049), but the difference was minor. The pro-resolution lipid mediators such as RvD1, MaR and PD1 were detected either at very low concentrations or were below the detection limit. This can be explained either by their instability during sample collection and storage or simply by very low and undetectable endogenous concentrations with current analytical technologies [34,35]. Recently, the ''Serum Metabolome Project'' identified 65 lipids by LC-MS [36]. RvE1 and RvD1 were among the lipid mediators identified in plasma of systemically healthy individuals with unknown periodontal status. Whereas these molecules have regulator functions in inflammation, the importance of circulating levels of these molecules is debated [36].
In the present study, RvD1, RvD2 and PD1 were determined in serum samples of AgP patients and healthy controls, with comparable RvD1 level. Recently, RvD1 and RvD2 were detected in the serum samples of healthy individuals having oral n3-fatty acid supplementation, while PD1 was below the detection limit [37]. Notably, in the present study, there was no n3-PUFAs supplementation of AgP patients or healthy controls. The precursor of RvE1 18-HEPE may represent the EPA-derived resolving capacity [37].
AA was present in higher concentrations in GCF samples of AgP patients than in controls. The main AA-derived proinflammatory lipid mediators are prostaglandins and leukotrienes. They are important biological mediators involved in the inflammatory response, as well as in regulation of bone formation and bone resorption. PGE2 has been proposed as a measure of disease status in GCF [32]. In our study, PGE2 was present at higher concentrations in GCF samples of AgP patients than in controls, which is in line with a previous study in which PGE2 concentration was significantly increased in GCF during the active phase of the disease [14]. Relatively small differences in PGE 2 levels (ng/ml) were observed when comparing AgP patients with healthy controls in the current analysis. The mean PGE 2 concentration in GCF was 2.2 ng/ml in AgP patients and 0.7 ng/ ml in healthy controls as determined by LC-MS.
In the present study, the concentrations of the mono-hydroxylated PUFA-metabolites; 5-and 12-HETE were mainly found in higher levels in GCF and serum, while 15-HETE was found in significantly higher levels in all samples of the AgP patients. A positive association with high levels of HETEs has also been reported in other types of disease with an inflammatory component [38,39]. Here we report that LTB4 was markedly increased in saliva samples of AgP patients, and its potential as an important marker for diagnosing periodontal disease activity should be further investigated. High concentrations of PGs, LTB4 and cysteinyl-leukotrienes have been reported to promote the accumulation of inflammatory cells such as polymorphonuclear leukocytes (PMNs) by increasing the blood flow and vascular permeability [40].
While initial fluid accumulation in the periodontal pocket seems to represent a transudate of interstitial fluid produced by an osmotic gradient leading to similar concentrations of for instance proteins in GCF and interstitial fluid, GCF is regarded as a serum exudate during inflammatory processes, with similar protein concentrations as in serum. This may also apply to the similarities in concentrations of lipid molecules in GCF and serum. Based on these findings, and with the techniques used in this study, GCF seems to be the most reliable biological fluid to reflect the local destruction (reviewed by Griffiths [41]).
The PMNs underlying the junctional epithelium in the periodontium represent the first line of defense against periodontal pathogens [42]. Transmigration of PMNs across epithelial surfaces represents a shared phenomenon among inflammatory mucosal  conditions. In order to be recruited to the site of the battle, the cells depend on chemotactic signals [43]. Hepoxilin A 3 (HXA 3 ) has been shown to be a potent chemoattractant for PMNs. In our study, the concentration of HXA 3 in GCF was not significant different between patients and controls. In serum, however, the concentration of HXA 3 was higher in patients compared to controls (P = 0.005).
A balanced ratio between n3-and n6-PUFAs is suggested by World Health Organization (WHO) to decrease risk of coronary heart disease (WHO 1995). Recently, resolution actions in periodontal disease have been credited to products from the n3-PUFAs [30]. The finding in the present study further demonstrated a possible effect of n3-to n6-PUFAs ratios in AgP patients. Future studies are needed to confirm the effect of pro-resolution vs. pro-inflammatory imbalance in different types of periodontal diseases.

Conclusions
In the present study, we report on generally elevated levels of eicosanoids and docosanoids and various n6-and n3-PUFAs in the GCF, saliva and serum of AgP patients compared with healthy controls. In addition, there were significantly increased concentrations of PGE 2 in GCF of AgP patients compared with healthy controls determined via two independent methodological approaches. We demonstrated a lower ratio of precursors of proresolution to pro-inflammatory lipid mediators in the GCF of AgP patients compared with healthy controls. GCF seemed to be the most reliable biological fluid for the assessment of the periodontal condition. Findings from this study suggest that the ratio of the precursor of pro-resolution vs. pro-inflammatory derivatives might be appreciated as markers of local destruction in aggressive periodontitis. Further studies focussing on various markers for diseases severity in bigger cohorts of patients and correlating these values with various lipid markers and ratios are in progress.

Study population
Nineteen AgP patients (4 males and 15 females) and 19 controls   number of teeth in each patient was from 21 and 28. The diagnosis was confirmed by radiographs in addition to a fullmouth clinical examination performed by the same dentist. Before the examinations, intra-individual calibration of the examiner was undertaken. Medical history was recorded for each patient, and subjects who had received antibiotics or periodontal treatment within the last three months before the examination, or were pregnant or had any systemic disease that could affect the progression of periodontal disease, were excluded. The 19 control subjects were systemically healthy employees and students at UST, with PPD #3 mm and CAL #2 mm for all teeth, and in most cases all teeth were present.

Ethics Statement
Ethical approval was obtained from the Research Ethics Committee at UST, Omdurman, Sudan, and the Regional Committee for Medical Research Ethics (REK) Western Norway (REK 177.04) (Bio bank no. 2355). Each subject signed an informed consent written in their native language (Arabic) prior to participation. On behalf of the young participants involved in the study, the informed consents were assigned by their guardians, all in accordance with the policy of REK, University of Bergen, and UST.

Collection of biological materials
Saliva. Patients and controls were asked to rinse the mouth with water before saliva was collected in a 10 ml sterile plastic tube. The unstimulated whole saliva (UWS) samples were processed the same day by centrifuging for 10 min at 30006g (SUPERFIT centrifuge, Mumbai, India). Each supernatant was dispersed into 4 tubes and stored in liquid nitrogen.
Blood. Peripheral blood was collected in vacutainer tubes; for serum. The whole tubes were preserved at bench-side for not more than two hours, and centrifuged for 10 min at 30006g (SUPER-FIT centrifuge, Mumbai|, India). The tubes with serum were then stored in liquid nitrogen.
Gingival crevicular fluid. A perio-paper (PERIOPAPERH Gingival Fluid Collection Strips, Oraflow Inc, New York, USA) was used for collection of GCF from the mesio-buccal site of posterior teeth (mean pocket depth of sampled teeth was 5.9 mm). Prior to the GCF collection, the area was isolated with cotton rolls and exposed to a gentle air stream for five seconds. Then the perio-paper was placed subgingivally for 10 seconds, before being removed and stored in an empty tube in liquid nitrogen.
HPLC-ESI-MS-MS Analysis of free fatty acids, eicosanoids and docosanoids in GCF, saliva and serum The previously described procedure for the HPLC-ESI-MS-MS method was followed with minor modifications [44], but including complete quantification of the detected lipid derivatives.
Sample preparation. The whole analytical sample preparation procedure is based on an established method used for retinoid quantification. Saliva samples were much lower in analyte concentration and were dried in the Eppendorf concentrator 5301 (Eppendorf, Germany) at 45uC from 1.5 ml to 50 ml prior to extraction. In summary, 50 ml of serum, saliva extract or GCF and 150 ml acetonitrile were shaken for 3 min, the precipitated protein was centrifuged at 13.000 rpm, 4uC for 6 min, 130 mL of the resulting supernatant was spiked with 10 ml isotope supernatant mix, evaporated in Eppendorf reaction vials with an Eppendorf concentrator at 30uC for ,60 min until the sample volume was ,10 ml. The Eppendorf concentrator was vented with argon to prevent degradation of eicosanoids and docosanoids. The dried extract was resuspended with approximately 25 ml of HPLC solvent A [64.3% water (water, Chromasolv Plus from Sigma-Aldrich, Hungary) and 35.5% acetonitrile (Merck KGaA, Germany) and 0.2% formic acid (Fluka, Hungary)] to yield 35 ml, then vortexed (15 sec), shaken (3 min) and transferred into micro injection inserts vials (Waters, Hungary). These glass vials with the 35 ml extract were transferred into brown screw top vials with PTFE/silicone septa sample (Waters, Hungary) and placed in the pre-cooled (15uC) autosampler of the Waters 2695XE separation module.   Hungary).
HPLC conditions. The eluents were degassed in the Waters 2695XE separation module prior to mixing, then passed through an in-line filter (1-2 mm; Knauer, Germany) before reaching the analytical column (LiChroCART, 125 X 2 mm; Superspher 100, RP-18, endcapped) from Merck KgaA (Germany) embedded in the column compartment. A multilinear gradient was formed from solvent A (see above) and solvent B (methanol; Merck KGaA Germany). The gradient consisted of the following steps: 0.0 min 20% B, 3.0 min 20%B, 5.0 min 60%B, 15.0 min 100%B, 15.9 min 100%B and 16.0 min 5%B. The flow rate was adjusted to 0.4 ml/min and the column was heated to 40uC. From the same biological extract, 10 ml for each HPLC analysis was used. This step was performed twice using the same HPLC conditions and two different MS-MS analysis options for better resolution and quantification of the various analytes.
MS options. The Micromass Quattro Ultima PT was controlled via the MassLynx software. Argon with an inlet pressure of 0.8 bar was used. ESI (electro spray ionization source, Waters, Hungary) was vented by nitrogen continuously produced by the nitrogen generator (Peak Scientific NM30 Nitrogen generator) including compressor (Waters, Hungary) with the inlet flow set at 3.6 e-3 mbar.
Multiple reaction monitoring settings: ESI, with a negative ESI -setting, was performed with the HPLC eluent following the ion source temperature of 85uC. The desolvation gas flow was 780 l/h, the desolvation temperature was 400uC, the cone gas flow was 10 l/h, the capillary current was 3 mA, and the cone voltage was 50 V. Aperture voltage was set at 0 V and the RF lens voltage was set at 35 V (for 1) and 0.2 V (for 2). The analyzer settings were LM1 resolution 14.5; HM1 resolution 14.5; Ion energy 1 0.7; entrance -1; collision 0 (collision parameters are set for each substance at the MS -method parameters); exit 2; LM2 resolution 14.5; HM2 resolution 14.  the Curve'' (AUC) and compared with the AUC of standard compounds. To ensure optimal extraction isotope labelled standard compounds were used. This analytical procedure was established for liquids and tissue analysis [45].

ELISA
Levels of PGE 2 and LXA 4 in GCF. Perio-papers saturated with GCF from AgP patients and controls were immersed in 12 mM (pH 7.6) Tris buffer. The lipids recovered from the GCF samples were diluted in the assay buffer of the ELISA kit; 10 times for the analysis of PGE 2 and 5 times for the LXA 4. The PGE 2 and LXA 4 were quantified by the competitive ELISA (Neogen Corporation, Nandino Blvd., Lexington KY, USA) according to instructions provided by the manufacturer. All GCF samples were run in duplicate and measured by FLUOstar OPTIMA, and data were reported in ng/ml.

Statistical analysis
Values which were determined under the detection limit were replaced with values which present half of our quantification limit: 0.02 ng/ml for saliva and 0.05 ng/ml for GCF and serum. The difference in significance between AgP and healthy controls was determined by Predictive Analytics Software (PASW) version 18, using the Mann Whitney test. P values ,0.05 were considered statistically significant.