*Article* **TREM-1 Is Upregulated in Experimental Periodontitis, and Its Blockade Inhibits IL-17A and RANKL Expression and Suppresses Bone Loss**

**Nagihan Bostanci 1,2,\*, Toshiharu Abe 3, Georgios N. Belibasakis 1,2 and George Hajishengallis <sup>3</sup>**


Received: 10 September 2019; Accepted: 24 September 2019; Published: 1 October 2019

**Abstract:** Aim: Triggering receptor expressed on myeloid cells-1 (TREM-1) is a modifier of local and systemic inflammation. There is clinical evidence implicating TREM-1 in the pathogenesis of periodontitis. However, a cause-and-effect relationship has yet to be demonstrated, as is the underlying mechanism. The aim of this study was to elucidate the role of TREM-1 using the murine ligature-induced periodontitis model. Methods: A synthetic antagonistic LP17 peptide or sham control was microinjected locally into the palatal gingiva of the ligated molar teeth. Results: Mice treated with the LP17 inhibitor developed significantly less bone loss as compared to sham-treated mice, although there were no differences in total bacterial load on the ligatures. To elucidate the impact of LP17 on the host response, we analyzed the expression of a number of immune-modulating genes. The LP17 peptide altered the expression of 27/92 genes ≥ two-fold, but only interleukin (IL)-17A was significantly downregulated (4.9-fold). Importantly, LP17 also significantly downregulated the receptor activator of nuclear factor kappa-B-ligand (RANKL) to osteoprotegerin (OPG) ratio that drives osteoclastic bone resorption in periodontitis. Conclusion: Our findings show for the first time that TREM-1 regulates the IL-17A-RANKL/OPG axis and bone loss in experimental periodontitis, and its therapeutic blockade may pave the way to a novel treatment for human periodontitis.

**Keywords:** TREM-1; periodontal disease; intervention; inflammation; LP17; IL-17; RANKL; OPG

#### **1. Introduction**

Periodontitis entails the destruction of the tooth-supporting (periodontal) tissues, as an outcome of their inflammatory response to the juxtaposed microbial biofilm forming on the tooth surface [1,2]. Although oral bacteria are essential for initiation of the disease, the resulting inflammation is what causes collateral damage to the tissues, which may eventually lead to tooth loss. The inflammatory mediators that lead to alveolar bone destruction form an intricate network [3,4], in which the receptor activator of NF-κB ligand (RANKL)/osteoprotegerin (OPG) system plays a crucial role as a terminal regulator of the resulting osteoclastogenesis and bone resorption [5,6]. Recently discovered host molecules, acting between the microbial challenge and the RANKL/OPG system, may lead to better understanding of the pathogenesis of periodontal disease and offer novel targets for therapeutic intervention.

Triggering receptor expressed on myeloid cells 1 (TREM-1), a member of the immunoglobulin superfamily, has been defined as a modifier of local and systemic inflammation, especially in response to bacterial infections [7–9]. Bacterial infection can upregulate the membrane-bound and soluble forms of TREM-1, which in turn amplifies inflammation. This is a particularly crucial response associated with systemic sepsis [10,11]. Blockade of TREM-1 engagement by either soluble forms of TREM-1 or synthetic peptides thereof reduces hyper-inflammatory responses and morbidity [12]. In a TREM-1 knock-out mouse model of viral or parasitic infection, it was demonstrated that the lack of TREM-1 signaling mitigated the severity of inflammation and disease (as compared to the wild-type mice) without, however, affecting pathogen clearance [13]. The study by Weber et al. [13] suggested that TREM-1 regulates inflammation, and that its therapeutic targeting may be beneficial in infection-driven inflammatory diseases without compromising pathogen clearance.

There is also correlative evidence to suggest that TREM-1 might modify periodontal inflammation. Specifically, the presence or expression of TREM-1 is increased in saliva, serum [14,15], gingival crevicular fluid [16–18], and gingival tissues [19] of patients with periodontitis as compared to individuals with periodontal health. TREM-1 levels also positively correlate with the levels of putative periodontal pathogens present in subgingival biofilms or lysed gingival tissue [16,19]. In this respect, multispecies biofilms [19] or *Porphyromonas gingivalis* alone induce TREM-1 gene expression in monocytes [20], whereas sub-antimicrobial doses of doxycycline can abolish this upregulatory effect [21].

The studies discussed above collectively suggest that TREM-1 expression is upregulated in periodontitis as a result of microbial stimulation. However, there are as-yet no interventional studies in preclinical models to conclusively demonstrate TREM-1 involvement in periodontitis. Hence, this in vivo study in a validated model of murine ligature-induced periodontitis [22] was designed to investigate the effect of local TREM-1 inhibition on the induction of experimental periodontitis, as well as on the expression of inflammation- and osteoclastogenesis-associated molecules in the gingival tissue. Our results described below implicate for the first time TREM-1 in the pathogenesis of periodontitis in a preclinical model and suggest a novel therapeutic approach for the treatment of this oral inflammatory disease.

#### **2. Materials and Methods**

#### *2.1. Ligature-Induced Periodontitis Model in Mice*

The well-established ligature-induced periodontitis model in specific pathogen-free C57BL/6 mice was used as described earlier [22]. All animal procedures were performed according to protocols approved by the Institutional Animal Care and Use Committee of the University of Pennsylvania, and adequate measures were taken to minimize pain or discomfort. To induce experimental periodontitis, a 5-0 silk ligature was tied around the maxillary left second molar for up to 8 days (*n* = 4–5 mice/group). The unligated contralateral molar in each mouse was used as baseline control (unligated control). A synthetic peptide derived from the extracellular domain of TREM-1 (LP17; LQVTDSGLYRCVIYHPP, Pepscan, Lelystad, Netherlands) was used as described earlier [7]. The LP17 blocking peptide is considered as a competitive antagonist of membrane-bound TREM-1 for its natural ligand, therefore acting as a decoy receptor for TREM signaling [23]. For the intervention experiments performed in this study, 5 μg of LP17 peptide or PBS were injected into the palatal gingiva of the ligated second maxillary molar 1 days before placing the ligature and every day thereafter until the day before sacrifice (day 5).

The measurements on the alveolar bone height were done on defleshed maxillae under a Nikon SMZ800 microscope (Nikon Instruments, Melville, NY, USA), and images of the maxillae were captured using a Nikon Digital Sight DS-U3 camera controller. The distance between the cemento-enamel junction (CEJ) and alveolar bone crest (ABC) was measured at six predetermined points on the ligated molar and adjacent regions using NIS Elements software (Nikon Instruments, Melville, NY, USA) [22]. To calculate bone loss, the six-site total CEJ–ABC distance for the ligated side of each mouse was subtracted from the six-site total CEJ–ABC distance of the contralateral unligated side. The results are presented in millimeters, and negative values indicate bone loss relative to the unligated control.

#### *2.2. Bacterial Counts on Silk Sutures*

The ligated silk sutures obtained from LP17-treated or PBS-treated mice at day 5 were collected (*n* = 5 mice/group). These were suspended individually in sterile PBS, and adherent bacteria were disassociated from the sutures via high-speed vortexing for 2 min. Serial dilutions of the samples were plated onto blood agar plates (BD Difco Laboratories, Detroit, MI, USA), and the plates were incubated anaerobically at 37 ◦C for 7 days. Results are reported as the mean number of colony forming units (CFUs) per millimeter length of silk suture ± the standard error of the mean (SEM). Anaerobic CFUs were preferred over aerobic ones because of the stronger etiological association of anaerobic organisms with periodontitis.

#### *2.3. Antimicrobial E*ff*ects of the Synthetic Peptides in Vitro*

A 6-species oral biofilm model was used to investigate the potential antimicrobial effects of LP17. The biofilm consisted of *Actinomyces oris* OMZ 745, *Veillonella dispar* OMZ 493 (ATCC 17748T), *Fusobacterium nucleatum* OMZ 598 (KP-F2), *Streptococcus mutans* OMZ 918 (UA159), *Streptococcus oralis* OMZ 607 (SK 248), and *Candida albicans* OMZ 110. In brief, biofilms were grown according to the standard protocol in 24-well cell culture plates on sintered hydroxyapatite (HA) discs, which were pre-conditioned for 4 h with pooled human saliva, for pellicle formation. Throughout the following experimentation period, the biofilms were grown in the presence of LP17 or 0.9% NaCl (sham control). After 5 days of biofilm growth under anaerobic conditions, the HA discs were vortexed vigorously for 1 min in 1 mL of 0.9% NaCl and then sonicated (Branson Sonic Power Company, Danbury, CT, USA) for 5 s to harvest the adherent biofilms. Then, to determine the total CFUs, the bacterial suspensions were serially diluted in 0.9% NaCl and 50 μL aliquots were plated on agar plates supplemented with 5% whole human blood at 37 ◦C for 72 h.

#### *2.4. Quantitative TaqMan Real-Time PCR and TaqMan Array Analysis*

The TaqMan Array 96-well Mouse Immune Response kit (Applied Biosystems) was used to assess the expression of 92 predetermined genes mediating the immune response and four endogenous control genes including *GAPDH, HPRT, GUSB and 18S RNA*. For this analysis, gingival tissues were collected at day 5 (*n* = 3 mice/group). Total RNA was extracted from these tissues by Qiagen Fibrous Tissue Extraction kit. According to the manufacturer's protocol, cDNA was mixed with 2× TaqMan Universal Master Mix and H2O to a total volume of 2160 μL. Subsequently, 20 μL of the mixture was placed into each well of the PCR array. The three steps of the cycling program were 95 ◦C for 10 min for 1 cycle, then 95 ◦C for 15 s, and 60 ◦C for 1 min for 40 cycles, using a Step One Plus Real-Time PCR System (Applied Biosystems). In addition, the transcription levels of TREM-1, interleukin (IL)-1β, RANKL, OPG, COX-2, and IL-6 were assessed by individual TaqMan Gene Expression Assays (Applied Biosytems). For TaqMan qPCR analysis, mouse ACTB (β-actin) was used as an endogenous control.

#### *2.5. Statistical Analysis*

All statistical analyses were performed using Prism v.6.0 software (GraphPad Software, La Jolla, CA, USA). One-way ANOVA was used to determine differences between three or more groups, whereas an unpaired, two-tailed Student *t* test was used to determine the statistical significance of differences between two groups. Differences were considered significant at *p* < 0.05.

#### **3. Results**

#### *3.1. Local Tissue Kinetics of TREM-1 Expression in Ligature-Induced Periodontitis*

Using the ligature-induced murine periodontitis model, we first investigated the regulation of TREM-1 expression in the gingival tissue. TREM-1 gene expression was significantly upregulated at the ligated sites in a time-dependent manner, peaking at day 8 compared to the unligated control sites forming the baseline (Figure 1A). Compared to the unligated control sites, TREM-1 mRNA levels in the ligated sites were approximately 16-fold and 17-fold higher at day 5 and day 8, respectively (*p* < 0.01). Since TREM-1 activation is involved in the upregulation of a number of key proinflammatory cytokines [20], including IL-1β, which is crucial in periodontal pathogenesis, IL-1β gene expression levels were also assessed. A similar expression pattern was observed for IL-1β. In particular, compared to the baseline control, IL-1β gene expression at ligated sites was approximately 13-fold, 21-fold, and 27-fold higher at days 3, 5, and 8, respectively (*p* < 0.01) (Figure 1B).

**Figure 1.** Kinetics of gingival tissue expression of TREM-1 and IL-1 beta in ligature-induced periodontitis. TREM-1 mRNA **(A)** and IL-1 beta mRNA **(B)** levels were examined in unligated control gingiva and ligature-induced periodontitis gingival tissues at 1 to 8 days. The gene expression levels were detected by TaqMan real-time qPCR and calibrated against the expression of housekeeping gene β*-actin*. Results are means ± SEM (*n* = 4 mice/group). \* *p* < 0.05 and \*\* *p* < 0.01 between the indicated groups.

#### *3.2. Role of TREM-1 in Alveolar Bone Loss*

The kinetics of TREM-1 gene expression followed a pattern similar to those of ligature-induced bone loss seen in our previous publication [22]. To determine whether there was a cause-and-effect relationship between gingival TREM-1 expression and alveolar bone loss, we subjected groups of mice to ligature-induced periodontitis with local administration of the LP17 or with PBS sham control. Five days after placement of the ligatures, mice treated with LP17 developed significantly less alveolar bone loss as compared to the sham-treated mice (*p* < 0.05) (Figure 2A,B), indicating that TREM-1 signaling contributes to induction of alveolar bone loss.

**Figure 2.** Inhibition of ligature-induced bone loss by LP17. Ligatures were placed on the left maxillary molars of C57BL/6 mice and then locally microinjected with 5 μg of TREM-1 blocking peptide (LP17) or with PBS sham 1 day before placing the ligature and every day thereafter until day 5. The distance between the cemento-enamel junction (CEJ) and alveolar bone crest (ABC) was measured at six predetermined points on the ligated side. Representative images of PBS sham- and LP17-treated maxillae exhibiting differential bone loss **(A).** To calculate bone loss, the six-site total CEJ–ABC distance for the ligated (L) side of each mouse was subtracted from the six-site total CEJ–ABC distance of the contralateral unligated (U) side. The results are presented in millimeters, and negative values indicate bone loss relative to the unligated control **(B)**. Data are means ± SEM (*n* = 4–5 mice/group). \* *p* < 0.05 and \*\* *p* < 0.01 between the indicated groups.

#### *3.3. Investigation of Potential LP17 Antimicrobial Activity*

To determine whether the protective effects of LP17 in ligature-induced periodontitis could, in part, be attributed to potential antimicrobial effects, we determined the microbial load of the treated mice from the above-described in vivo experiment (Figure 2). To this end, bacteria were extracted from the recovered ligatures (day 5) and cultivated anaerobically for 7 days on blood agar plates. To normalize the data, the counted CFUs were divided by the length of corresponding suture, and the results revealed that sutures from the LP17-treated mice yielded similar CFUs, as compared to the PBS sham-treated group (*p* > 0.05) (Figure 3A). Furthermore, the potential antimicrobial impact of LP17 was tested in vitro on a 6-species biofilm model for 5 days. A 1.3-fold reduction in total CFUs was observed, compared to the saline sham-treated group (*p* > 0.05). These results suggested that the LP17 peptide preferentially acted by regulating the host response rather than bacterial growth. Hence, we next investigated the host-modulating activity of LP17.

**Figure 3.** LP17 does not affect the microbial load in vivo. (**A)** Detached material from the recovered ligatures at day 5 from mice used in Figure 2 were cultivated anaerobically for 7 days on blood agar plates, followed by total colony forming unit (CFU) determination. To normalize the data, the counted CFUs were divided by the length of corresponding suture. Data are means ± SEM (*n* = 5 mice/group). NS: Not significant, *p* > 0.05. (**B**) LP17 does not affect the microbial load in vitro. The in vitro biofilms were grown in the presence of LP17 or 0.9% NaCl (sham). After 5 days of biofilm growth under anaerobic conditions, biofilm bacteria were harvested from the discs and cultivated anaerobically for 3 days on blood agar plates, followed by total CFU determination. The CFUs are given per hydroxyapatite (HA) disc. Data are means ± SEM (*n* = 3 disc /group). NS: Not significant, *p* > 0.05.

#### *3.4. Modulation of Immunoregulatory Genes by TREM-1*

To understand how TREM-1 signaling regulates the host periodontal response, the defleshed gingival tissues were analyzed for the expression of a number of immunoregulatory genes at day 5, by using a mouse immune response qPCR Array profiling for 92 individual genes (Table S1). Differential expression analysis was done by the following pair-wise comparisons: (a) unligated sites versus ligated sites and (b) PBS sham-treated ligated sites versus LP17-treated ligated sites. Although a basal expression level was detected for all studied genes at unligated control sites, a total of 38 genes were differentially transcribed by more than two-fold during the experimental infection period (Table S1). Among those, 27 genes were induced in the ligature-induced gingival sites, 7 of which reached statistical significance (*p* < 0.05). Another 11 genes were repressed more than two-fold, 7 of which also reached statistical significance (*p* < 0.05). The significantly upregulated genes were *IL-17A, IL-1*β*, CD80, CCR4, HMOX1, VEGFA,*and *CD68*, whereas the significantly downregulated ones were *SKI, SMAD 7, IL-7, NFATC 3, FAS, IL-15,* and *SMAD 3* (Figure 4A).

**Figure 4.** Modulation of immunoregulatory genes by TREM-1. Dissected gingiva from unligated control sites (UL) and ligated shame treated sites (Sham) for 5 days. The mRNA expression of 92 key genes mediating the immune response and four endogenous control genes including *GAPDH, HPRT, GUSB and 18S RNA mRNA* were assessed by qPCR. The gene expression levels were calibrated against the expression of housekeeping genes (detailed list provided in Table S2). The significantly regulated genes are presented (fold-change ≥ 2 and \* *p* < 0.05). **(A).** Dissected gingiva from unligated control sites (UL), or ligated sites from PBS sham-treated sites (Sham) or sites treated with 5 μg synthetic TREM-1 inhibitor (LP17) for 5 days. The mRNA expression of IL-17 is presented (fold-change ≥ 2 and \* *p* < 0.05). **(B)**. Data are means ± SEM (*n* = 3 mice/group).

Treatment with the LP17 peptide altered the expression of 27 genes by more than two-fold (23 downregulated, 4 upregulated) (Supplement Table S2). Although the expression of proinflammatory cytokines associated with periodontal disease pathogenesis (such as, IL-1β, IL-6, IL-17A, and TNF) was inhibited, statistical significance was reached only for IL-17A, which was downregulated by 4.9-fold (Figure 4B). Taken together, these data indicate that ligature-induced periodontitis is associated with upregulation of a number of proinflammatory genes that seem to be inhibited by LP17, which predominantly targets IL-17A expression, a signature cytokine of Th17 cells that were shown to drive inflammatory bone loss in mice and humans [24].

#### *3.5. Regulation of the RANKL*/*OPG Axis by TREM-1*

The upregulation of IL-17A expression, a cytokine associated with chronic inflammatory tissue destruction and alveolar bone loss [25,26], prompted us to investigate further the involvement of TREM-1 in the molecular regulatory mechanisms of bone resorption, particularly the RANKL/OPG system. RANKL was significantly induced at the sham-treated ligated sites (39-fold), whereas administration of LP17 inhibited this upregulatory effect by 8.9-fold (Figure 5A). The expression of OPG also increased at the sham-treated ligated sites (2.7-fold) but was not significantly affected by administration of LP17 (Figure 5B). As a result, the relative RANKL/OPG ratio, a molecular determinant of bone resorption, was significantly reduced in response to LP17 treatment by five-fold as compared to PBS sham treatment (Figure 5C).

As IL-6 and COX-2 that are produced in high levels during inflammation are considered as key regulators of RANKL expression [4,27,28], we assessed their regulation by TREM-1. Interestingly, the expressions of COX-2 and IL-6 were not significantly affected by LP17 treatment, indicating that the regulation of RANKL in this model may not be dependent on COX-2 (Figure 6A) or IL-6 (Figure 6B).

**Figure 5.** Inhibition of receptor activator of nuclear factor kappa-B-ligand (RANKL)/osteoprotegerin (OPG) ratio by LP17. Gingival tissue samples were dissected at day 5 from mice used in Figure 2 and were processed for gene expression of RANKL **(A)** and OPG **(B)** by qPCR. The relative RANKL/OPG ratio was also calculated **(C)**. The expression of the indicated molecules was determined in unligated (UL) control gingiva and in ligated gingival tissues treated with 5 μg synthetic TREM-1 inhibitor (LP17) or PBS sham. The gene expression levels were detected by TaqMan real-time qPCR and calibrated against the expression of the housekeeping gene β*-actin*. Results are means ± SEM (*n* = 4 mice/group). \* *p* < 0.05 and \*\* *p* < 0.01 between the indicated groups. NS: Not significant, *p* > 0.05.

**Figure 6.** LP17 does not affect COX-2 and IL-6 levels. Gingival tissue samples were dissected at day 5 from mice used in Figure 2 and were processed for gene expression of COX-2 **(A)** and IL-6 **(B)** by qPCR. The expression of the indicated molecules was determined in unligated (UL) control gingiva and in ligated gingival tissues treated with 5 μg synthetic TREM-1 inhibitor (LP17) or PBS sham. The gene expression levels were detected by TaqMan real-time qPCR and calibrated against the expression of the housekeeping gene β*-actin*. Results are means ± SEM (*n* = 4 mice/group). NS: Not significant, *p* > 0.05.

#### **4. Discussion**

Our present study shows for the first time that TREM-1 regulates alveolar bone loss in experimental periodontitis and paves the way for a novel approach to treat human periodontitis. In line with earlier observations in humans demonstrating upregulated TREM-1 gingival expression in periodontitis patients, as compared to healthy controls [19], our study showed progressively increased induction of TREM-1 gingival expression during the course of experimental periodontitis in response to biofilm accumulation. TREM-1 propagates proinflammatory cytokine expression, representatively demonstrated by IL-1β in the present study. These findings are in accordance with our previous studies showing a positive correlation between subgingival biofilms and TREM-1 levels in gingival tissue or gingival crevicular fluid of individuals with periodontitis [16,19]. Although the cellular distribution of TREM-1 in gingival tissue was not investigated in the present experimental setting, monocytes/resident macrophages and polymorphonuclear neutrophilic leukocytes (PMNs) are known to be a major source of TREM-1 in inflammation [9,29–31]. In this respect, our earlier work presented that multispecies oral biofilms [19] or the keystone pathogen *Porphyromonas gingivalis* is able to induce TREM-1 gene expression in monocytes or in PMNs in the tissue culture systems [20,32].

The major novel finding of this in vivo study is that local (gingival) injection of a TREM-1 blocking peptide, namely LP17, substantially reduced the RANKL/OPG osteoclastogenic ratio and alveolar bone loss, thus providing preclinical support for a new therapeutic target for periodontitis. Clinical studies have demonstrated that the RANKL/OPG ratio as well as IL-17 gingival tissue expression are upregulated in human periodontitis [33–36]. Intriguingly, LP17 selectively downregulated IL-17 expression among all studied immune response markers. Given that IL-17 regulates the expression of RANKL [37], it is possible that the capacity of LP17 to downregulate the RANKL/OPG ratio may be mediated through its ability to inhibit IL-17. Similarly, inhibition of IL-17 by its antagonist Del-1 has been shown to efficiently block osteoclastogenesis and subsequent periodontitis [38,39]. Thus, our study lends further support to the concept that IL-17 is a key driver of periodontal bone loss, although this is the first time that TREM-1 signaling is linked to IL-17 in the context of periodontitis.

Our present data are also consistent with a recent in vitro study, demonstrating that the LR12 TREM-1 inhibitor LR12 could prevent monocytic activation by *P. gingivalis* LPS [40], as well as our earlier in vitro findings demonstrating that LP17 can reduce cytokine release by monocytes in response to *P. gingivalis* whole bacteria [20,21]. Moreover, an earlier study in a psoriasis model showed that TREM-1 blockade in vitro and ex vivo significantly reduced the number of Th17 cells and decreased the secretion of IL-17, suggesting that TREM-1 positively regulates Th17 responses [41]. The COX-2 pathway and IL-6 are also important regulators of the RANKL/OPG ratio [27]. However, it is unlikely that TREM-1 regulates the RANKL/OPG ratio via COX-2 or IL-6 since LP17 failed to affect the expression of either gene. Thus, it is concluded that the alveolar bone resorptive effects of TREM-1 are, at least in part, mediated through activation of the IL-17-RANKL axis.

Although the impact of TREM-1 signaling on microbial control has been controversial in several bacterial challenge models [13], in the experimental periodontitis model, LP17 did not show a significant effect on oral bacterial load. This finding suggests that TREM-1 inhibition protects against periodontitis predominantly through host-modulation effects and is in line with earlier work indicating that LP17 did not alter the in vitro levels of *P. gingivalis* [20,32,42].

Moreover, TREM-1-deficient mice used in colitis and other experimental models of infection-driven inflammatory diseases exhibited no alterations in microbial clearance efficiency [13]. On the other hand, studies on lung infection models (e.g., *Pseudomonas aeruginosa*-induced pneumonia) indicated that administration of LP17 peptide reduced the bacterial load at an early stage of infection while increasing it at later stages; these effects, however, were attributed to indirect antimicrobial effects of TREM-1 related to early enhancement of neutrophil influx and consequent increase in phagocytic activity [43]. These observations are in line with the main function of TREM-1 as an inflammation fine-tuner [44], rather than a direct eliminator of infection, as is the case, for instance, for TNF-alpha. Yet, the use of anti-TNF antibody may be complicated because of the increased risk for reactivation of latent infection [45].

#### **5. Conclusions**

The present study conclusively demonstrated the involvement of TREM-1 in alveolar bone resorption during the course of experimental periodontitis in mice. Mechanistically, TREM-1 reduced the RANKL/OPG osteoclastogenic ratio, presumably via the inhibition of IL-17. Importantly, our findings also reveal a previously unidentified TREM-1-driven axis for inflammatory bone loss that could be targeted via small-molecule antagonists for therapeutic intervention in human periodontitis.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2077-0383/8/10/1579/s1, Table S1: Immune gene expression profile in ligated gingiva vs healthy gingiva, Table S2: immune regulatory genes in mouse gingiva by LP17.

**Author Contributions:** Conceptualization, N.B. and G.H.; Methodology, N.B., T.A,. G.N.B., and N.B.; Validation, N.B., G.N.B. and G.H.; Formal Analysis, T.A., N.B.; Resources, N.B., G.N.B., G.H.; Writing—Original Draft Preparation, N.B.; Writing—Review & Editing, G.H., G.N.B.; Visualization, T.A.; Supervision, G.H.; Funding Acquisition, N.B., G.N.B. and G.H.

**Funding:** This research was funded by grants from the National Institutes of Health (DE015254, DE024716, DE024153, and DE026152 to G.H.), the Swedish Research Council funds (2017-01198 to N.B.), and the APC by strategic funds from Karolinska Institutet (G.N.B.).

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Saliva and Serum Immune Responses in Apical Periodontitis**

#### **Milla Pietiäinen 1,\* , John M. Liljestrand 1,\* , Ramin Akhi 2,3,4,\*, Kåre Buhlin 1,5, Anders Johansson <sup>6</sup> , Susanna Paju 1, Aino Salminen 1, Päivi Mäntylä 7,8, Juha Sinisalo <sup>9</sup> , Leo Tjäderhane 1, Sohvi Hörkkö 2,3,4 and Pirkko J. Pussinen <sup>1</sup>**


Received: 10 June 2019; Accepted: 17 June 2019; Published: 21 June 2019

**Abstract:** Apical periodontitis is an inflammatory reaction at the apex of an infected tooth. Its microbiota resembles that of marginal periodontitis and may induce local and systemic antibodies binding to bacteria- and host-derived epitopes. Our aim was to investigate the features of the adaptive immune response in apical periodontitis. The present Parogene cohort (*n* = 453) comprises patients with cardiac symptoms. Clinical and radiographic oral examination was performed to diagnose apical and marginal periodontitis. A three-category endodontic lesion score was designed. Antibodies binding to the bacteria- and host-derived epitopes were determined from saliva and serum, and bacterial compositions were examined from saliva and subgingival samples. The significant ORs (95% CI) for the highest endodontic scores were observed for saliva IgA and IgG to bacterial antigens (2.90 (1.01–8.33) and 4.91 (2.48–9.71)/log10 unit), saliva cross-reacting IgG (2.10 (1.48–2.97)), serum IgG to bacterial antigens (4.66 (1.22–10.1)), and Gram-negative subgingival species (1.98 (1.16–3.37)). In a subgroup without marginal periodontitis, only saliva IgG against bacterial antigens associated with untreated apical periodontitis (4.77 (1.05–21.7)). Apical periodontitis associates with versatile adaptive immune responses against both bacterial- and host-derived epitopes independently of marginal periodontitis. Saliva immunoglobulins could be useful biomarkers of oral infections including apical periodontitis—a putative risk factor for systemic diseases.

**Keywords:** apical periodontitis; adaptive immunity; saliva; serum; antibody

#### **1. Introduction**

Apical periodontitis (AP) is an inflammatory disease that affects the tissues surrounding the apex of the tooth. It is triggered by oral pathogens infecting root canal. Both acute (abscess) and chronic inflammatory reaction (periapical granuloma and radicular cyst) can develop depending on the intensity of the bacterial infection and the host immune responses. Primary apical periodontitis usually develops when the bacteria in a caries lesion enter through enamel and dentin and cause microbial colonization of the pulp and eventually necrosis of the pulp tissue. Secondary apical periodontitis arises from a persistent infection of previously treated root canals or leakage of the filling in a root canal-treated tooth. Apical periodontitis is diagnosed from radiographs as an evident radiolucent area (referred to as endodontic lesion) at the tip of the root. Even slight radiographically evident widening of the periapical space is associated with an infection in the tooth [1]. AP is treated with root canal treatment where infection is eliminated chemomechanically and the root canal is filled.

Apical periodontitis is a highly common and underdiagnosed disease. It is estimated that approximately 10% of all teeth are endodontically treated, 5% have periapical radiolucencies [2], and the prevalence of apical periodontitis varies between 24 and 86% in different populations [3]. Up to 78% of endodontically treated teeth have root canal fillings with poor quality and ~36% of the root canal-treated teeth present apical periodontitis [2], suggesting that recurrent or persistent endodontic infections are common. Apical periodontitis is usually symptomless, and it can be diagnosed only by radiography.

Endodontic infections are polymicrobial and the structure of the intracanal biofilm may evolve toward obligate aerobes and Gram-negative anaerobes as the infection progresses. More than 400 different microbial taxa have been identified in endodontic samples from teeth with different forms of apical periodontitis [4]. Several studies have also shown that distinct bacterial communities are found in primary and secondary AP [5–8]. Despite the high interindividual variability in endodontic microbial community composition, the most often encountered phyla in the intracanal samples include Firmicutes, Actinobacteria, Bacteroidetes, Proteobacteria, and Fusobacteria. Genera such as Prevotella, Fusobacterium, Parvimonas, Lactobacillus, Streptococcus, and Porphyromonas are highly prevalent in intracanal samples [9]. Several members of these genera are also considered etiological pathogens for marginal periodontitis and the microbial profiles of these two conditions resemble each other [10].

Microbial antigens stimulate innate immune responses in periapical tissue aiming to restrict the infection. The expression of proinflammatory cytokines, prostaglandins, and proteolytic enzymes are markedly increased in the areas of tissue destruction [11]. As one antimicrobial strategy, apical periodontitis is also associated with oxidative stress [12]. Some studies suggest a modest contribution of endodontic infections to the plasmatic inflammatory markers [13,14], while a recent study found a significant association between endodontic lesions and systemic inflammatory burden in young adults [15].

Additionally, adaptive immune responses are activated to prevent the microbial invasion into the tissues surrounding teeth or into circulation. High concentrations of local immunoglobulins IgG and IgA and lesser amounts of IgM and secretory IgA are present in the inflamed tissues [16–19]. The levels of systemic immunoglobulins, including total IgA, IgG, and IgM, are increased in patients with AP [13]. We recently showed that subgingival *Porphyromonas endodontalis* levels and serum IgG against it were associated with a higher endodontic lesion score [20]. Several oral pathogens are also known to be able to induce cross-reactive antibodies, which may influence inflammatory responses. The cross-reactive antibodies are part of an immunological process called molecular mimicry, in which bacterial antigens sufficiently resembling human proteins are able to induce the production of antibodies reacting with human epitopes. The most studied epitopes include those present in the heat shock proteins (HSPs) and in oxidized low-density lipoproteins (oxLDL) [21].

The association of marginal periodontitis with several systemic conditions such as cardiovascular diseases (CVDs) is well established [22]. Due to similarities in the inflammatory and microbial profiles between marginal periodontitis and AP, it is also suggested that there could be a link between AP and CVDs [23,24]. Even though the possible association of apical periodontitis with systemic diseases has been of high interest, the adaptive immune response against the disease has not been investigated in detail. In this study we aimed to investigate serum and saliva antibodies against several oral pathogens associated with apical periodontitis and the role of cross-reactive antibodies in the disease.

#### **2. Experimental Section**

#### *2.1. Population*

The Corogene is a prospective cohort of Finnish patients who had an indication to coronary angiography between June 2006 and March 2008 at the Helsinki University Hospital [25]. The present study comprises the Parogene, which is a substudy of 508 patients with clinical and radiographic oral health examinations. The details of the examinations have been described elsewhere [26]. The information of smoking habits was collected with a questionnaire before the oral examination. The presence of diabetes (type I and II) was obtained from medical records. All subjects signed an informed consent and the study was approved by the Helsinki University Hospital ethics committee (approval reference number 106/2007). Patients with antibody measurements from serum and saliva samples were included (*n* = 453, 89.2% of the whole cohort). The number of dentate patients and subgingival samples was 426 (n of edentulous 27, 6.0%).

#### *2.2. Oral Diagnosis*

Endodontic lesions were diagnosed from the radiographs as described in detail earlier [20]. The recorded findings included root canal fillings, widened periapical space indicating irreversible pulpitis or precursors for endodontic lesions [1], and apical periodontitis seen as periradicular destruction in the tip of the root. An endodontic lesion score was defined to describe the severity of apical periodontitis [20]. Score I included patients without endodontic lesions (*n* = 162, 38.2%); score II, patients with ≥1 widened periapical space and/or 1 tooth with apical periodontitis (*n* = 194, 45.2%); and score III, patients with ≥2 teeth with apical periodontitis (*n* = 68, 16.0%). In addition, another subgrouping—the endodontic treatment score—was designed according to treated/untreated apical periodontitis: I, no endodontic lesions (*n* = 352, 77.7%); II, teeth with apical periodontitis, all with root canal fillings (*n* = 51, 11.3%); and III, apical periodontitis in tooth/teeth without root canal fillings (*n* = 50, 11.0%). Number of teeth and implants, presence of carious teeth, and inadequate root canal fillings were also recorded from the radiographs.

Diagnosis for marginal periodontitis was based on alveolar bone loss (ABL) detected in the radiographs and bleeding on probing (BOP) registered in the clinical examination from four sites of each tooth. Patient was considered periodontally healthy, when no ABL and <25% BOP was present; with gingivitis, when no ABL but ≥25% BOP; and with periodontitis, when ABL was present [27].

#### *2.3. Bacterial Analyses*

Subgingival plaque samples were collected from the deepest pathological periodontal pocket (≥ 4 mm) in each dentate quadrant as described earlier [28]. The microbiome analysis including 79 taxa was performed by using the checkerboard DNA-DNA hybridization assay [29] and the data was analyzed as described in our earlier article [28]. In the present work, we summed up the results of Gram-positive taxa (*n* = 45) and Gram-negative taxa (*n* = 34), which are presented in Supplementary Table S1.

Saliva samples were collected after stimulation by chewing for 5 min, and a minimum of 2 mL of saliva was collected by expectoration. The methods for sample processing and quantitative real-time PCR have been described in detail earlier [30]. Saliva concentration of four bacterial species associated with periodontitis was analyzed: *Aggregatibacter actinomycetemcomitans*, *Porphyromonas gingivalis*, *Tannerella forsythia*, and *Prevotella intermedia*.

#### *2.4. Antibody Determinations*

Serum IgA- and IgG-class antibody levels against seven bacterial species—*A. actinomycetemcomitans*, *P. gingivalis*, *T. forsythia*, *P. intermedia*, *Campylobacter rectus*, *Fusobacterium nucleatum*, and *P. endodontalis*—were determined with ELISA as described earlier [31]. The antigens were composed of formalin-killed whole cells and two dilutions in duplicate were measured [32]. After all antibody levels were determined, the absorbances were normalized according to the reference applied on each plate and the results were expressed as continuous ELISA-units (EU). The list of the antigens, sample dilutions, and coefficients of interassay variations are presented earlier [31].

Saliva IgA- and IgG-class antibody levels against five species—*A. actinomycetemcomitans*, *P. gingivalis*, *T. forsythia*, *P. intermedia*, *and P. endodontalis*—were determined from saliva supernatants obtained after centrifugation at 9300× *g* for 3 min. The target antigens used in the assays were either heat-killed whole bacterial cells or oxidized LDL epitope malondialdehyde acetaldehyde modification (MAA-LDL), copper-oxidized LDL (CuOx-LDL) [33], recombinant *P. gingivalis* virulence factor gingipain (Rgp44) [34], and 60-kDa *A. actinomycetemcomitans* heat shock protein (Aa-HSP60) [35]. Levels of salivary IgA and IgG antibodies to oxidized LDL and bacterial epitopes were determined by chemiluminescence immunoassay as previously described in detail [36,37]. The saliva samples were diluted accordingly: 1:250 for total IgA and IgG, 1:50 for IgA to oxidized antigens, 1:20 for Aa-HSP60, and 1:10 for bacterial antigens. For IgG measurements, saliva samples were diluted 1:10 for all antigens. Each saliva sample was measured as triplicates. Immunoassay results were presented as relative light units (RLU) per 100 milliseconds (ms).

#### *2.5. Calculations of Cross-Reactive Antibodies and Antibodies Binding to Bacterial Antigens*

In addition to the mean levels of antibodies against each specific antigen, the combined antibody levels of saliva and serum IgA and IgG were calculated. The bacterial antigens included *A. actinomycetemcomitans*, *P. gingivalis*, *P. intermedia*, *P. endodontalis*, and *T. forsythia* (referred as IgA/IgG against bacteria). The epitopes recognized in *P. gingivalis* and *A. actinomycetemcomitans* giving rise to cross-reactive antibodies with MAA-LDL included Rgp-44 and Aa-HSP60 (referred as cross-reacting IgA/IgG).

#### *2.6. Statistical Methods*

The characteristics are presented as mean values with standard deviations (SD) or 95% confidence intervals. For clarity, standard error is displayed in the figures as error bars. In the supplementary tables, the bacterial levels are presented as medians and interquartile ranges (IQR). Before statistical comparisons, the antibody and bacterial levels were transformed with 10-base logarithm. The significance of the differences was tested by using *t*-test, ANOVA, Chi-square, or Mann–Whitney, when appropriate. The weighted linear terms were examined with ANOVA and Jonckheere–Terpstra test for normally distributed and skewed data, respectively. The associations were analyzed by using linear and logistic regression models adjusted for age, sex, marginal periodontitis (healthy, gingivitis, and periodontitis), number of teeth, and smoking (never/ever). When the dependent variable was composed of several subgroups, multinomial regression was used. When the associations were examined in the subgroup of patients without marginal periodontitis, the confounders were limited to age, sex, and smoking (never/ever).

#### **3. Results**

Characteristics of the dentate population are presented in Table 1. The mean (SD) age was 62.9 (9.1) years and 67% were males. The mean number of teeth was 21.4 (7.5), and caries and apical periodontitis were common findings, in 47.4% and 23.8% of the population, respectively. Also, marginal periodontitis ranging from mild to severe was present in most patients (75.5%).



The endodontic findings registered included root canal fillings, widened periapical space, and apical periodontitis. Mean antibody levels in serum and saliva against specific antigens, as well as the saliva and subgingival bacterial levels according to the endodontic findings are presented in supplementary tables (Supplementary Tables S2 and S3).

Among serum or saliva IgA-class antibody levels only sporadic significant differences were observed between patients with and without endodontic findings, whereas among IgG-class antibodies several significant differences were found. The antigens producing these differences included *A. actinomycetemcomitans*, *P. gingivalis*, *P. intermedia*, *P. endodontalis*, *C. rectus*, *F. nucleatum*, and *T. forsythia*, as well as Aa-HSP60, rgp44, MAA-LDL, and CuOx-LDL (Table S2). Among the salivary or subgingival bacterial concentrations, significant differences were mostly found between patients with and without widened periapical spaces (Table S3).

For further analyses, the microbial biomarkers were combined, and the mean levels are presented in Figures 1 and 2. The combinations included antibody level against bacteria, cross-reactive antibodies, salivary bacteria, and subgingival bacteria. Similarly as above, the mean saliva IgG-class antibody levels against bacteria and the cross-reactive antibodies as well as saliva and subgingival bacterial levels were higher in patients with endodontic findings. From the serum antibody levels, the IgG against bacteria were higher only in patients with widened periapical spaces (*p* = 0.015). In these patients, the increase of subgingival bacterial levels was due to both Gram-positive (*p* = 0.022) and Gram-negative (*p* = 0.005) species.

**Figure 1.** Saliva and serum antibody levels according to endodontic findings. The patients were divided into groups according to the presence of root canal fillings, widened periapical spaces, and apical periodontitis. Saliva (A, B) and serum (C) IgA- and IgG-class antibodies were determined. The bacterial antigens included *A. actinomycetemcomitans*, *P. gingivalis*, *P. intermedia*, *P. endodontalis*, and *T. forsythia*. The antigens giving rise to cross-reactive antibodies included MAA-LDL, Rgp-44, and Aa-HSP60. White columns depict the absence of the endodontic finding and black columns depict the presence of the endodontic finding. Means and standard errors are shown. The asterisks depict statistical significance between the groups defined by the *t*-test after logarithmic transformation: \* *p* < 0.05, \*\* *p* < 0.01. \*\*\* *p* < 0.001.

**Figure 2.** Saliva and subgingival bacteria according to endodontic findings. The patients were divided into groups according to the presence of root canal fillings, widened periapical spaces, and apical periodontitis. Salivary bacterial concentrations of *A. actinomycetemcomitans*, *P. gingivalis*, *P. intermedia*, and *T. forsythia* were determined by qPCR, and subgingival *A. actinomycetemcomitans*, *P. gingivalis*, *P. intermedia*, *P. endodontalis*, and *T. forsythia* by checkerboard DNA–DNA hybridization (A). This method was also used to examine subgingival 79 taxa, which were divided into Gram-positive (*n* = 45) and Gram-negative (*n* = 34) (B). The white columns depict the absence, and black columns presence of the endodontic finding. Means and standard errors are shown. The asterisks depict statistical significance between the groups defined by the *t*-test after logarithmic transformation: \* *p* < 0.05, \*\* *p* < 0.01. \*\*\* *p* < 0.001.

The associations of antibody and bacterial levels with endodontic findings are presented in Table 2 for the whole population calculated by using linear and logistic regression models adjusted for age, sex, number of teeth, smoking, and status of marginal periodontitis. The estimates are presented for a 10-fold increase in the antibody or bacterial levels, number of root canal-treated teeth associated with saliva IgA and IgG against bacteria, and cross-reacting IgG. Among these, only saliva IgG against bacteria associated with the presence of root canal-treated teeth with an OR (95% CI) 2.52 (1.43–4.43). Number of widened periapical spaces associated with saliva cross-reactive IgA and IgG, saliva IgG against bacteria, and subgingival bacteria in linear regression models. The presence of widened periapical spaces associated with saliva IgA and IgG against bacteria with ORs 2.09 (1.01–4.34) and 2.25 (1.40–3.61), respectively, and with cross-reacting IgG, 1.56 (1.22–1.99). Significant associations were also observed between the presence of widened periapical space and Gram-positive and Gram-negative subgingival bacteria with ORs 1.40 (1.02–1.92) and 1.45 (1.05–2.00). Number of teeth with apical periodontitis associated with saliva IgG against bacteria and cross-reacting IgG, which also presented significant ORs with the presence of apical periodontitis (2.90 (1.71–4.92) and 1.62 (1.24–2.11)).

A 3-category endodontic lesion score was designed for the severity of apical periodontitis. In addition, a 3-category endodontic treatment score was designed to investigate the contribution of treatment (root canal fillings) on the associations. The characteristics of the population are presented according to these scores in Table 3. Number of teeth, carious teeth, teeth with root canal fillings, and inadequate root canal fillings increased significantly with increasing scores. Also marginal periodontitis was more prevalent with high endodontic lesion (*p* < 0.001) or endodontic treatment (*p* = 0.287) score. Mean antibody levels in serum and saliva against specific antigens, as well as the saliva and subgingival bacterial levels according to these scores are presented in supplementary tables (Table S4 and S5). The association of the scores with the combined microbial biomarkers was analyzed by multinomial regression models for the log10-transforemed units (Figure 3). All measured parameters displayed positive trends with the increasing endodontic lesion score. Statistically significant associations (OR (95% CI)) with the highest endodontic scores were observed for saliva IgA (2.90 (1.01–8.33)) and IgG (4.91 (2.48–9.71)) against bacteria, saliva cross-reacting IgG (2.10 (1.48–2.97)), serum IgG against bacteria (4.66 (1.22–10.1)), subgingival species (1.15 (1.07–1.25)), and Gram-negative subgingival species (1.98 (1.16–3.37)). Regarding the treatment, only saliva IgG against bacteria, cross-reacting IgG, and serum IgA and IgG displayed increasing trends. Significant odds (OR (95%CI)) for untreated apical periodontitis were observed for saliva IgG against bacteria (5.32 (2.61–10.8)) and for cross-reacting IgG (2.04 (1.44–2.88)) (Figure 3).

The main results were reanalyzed in the subgroup of patients without marginal periodontitis (*n* = 132). There were no significant differences in the bacterial levels between groups divided according to the endodontic findings. Saliva IgG antibodies against bacteria and cross-reacting antibodies were higher in subjects with root canal fillings (*p* = 0.003 and 0.004), widened periapical spaces (*p* = 0.008 and 0.008), and apical periodontitis (*p* = 0.012 and 0.385). Both antibody levels increased in groups of patients with increasing endodontic scores (p for linear trend 0.009 and 0.020). The antibodies against bacteria (*p* for linear trend 0.007), but not the cross-reacting antibodies (*p* = 0.569), increased in patients with greater endodontic treatment scores. The associations of the saliva IgG class antibodies with endodontic lesion score and endodontic treatment score are presented in Table 4. Increasing trends were observed clearly only for saliva IgG against bacteria; the multivariate odds (OR (95% CI)) for having multiple teeth with apical periodontitis and for having teeth with untreated apical periodontitis were 3.45 (0.83–14.3) and 4.77 (1.05–21.7), respectively.



The dependent variable in the linear regression was the number of findings and in the logistic regression the presence of findings. Adjusted for age, sex, marginal periodontitis (healthy, gingivitis, periodontitis), number of teeth, and smoking (never/ever). \* Antibodies against *A. actinomycetemcomitans*, *P. gingivalis*, *P. intermedia*, *P. endodontalis*, and *T. forsythia*; \*\*Cross-reactiveantibodies,antigensPg-Rgp44,Aa-HSP60,andMAA-LDL.



ANOVA; Chi-square test; Endodontic lesion score: score I, no endodontic lesions; score II, patients with ≥1 widened periapical space and/or 1 tooth with apical periodontitis; and score III, patients with ≥2 teeth with apical periodontitis. Endodontic treatment score: score I, no endodontic lesions; score II, apical periodontitis only in teeth with root canal fillings; score III, apical periodontitis in teeth without root canal fillings.



1 Adjusted for age, sex, and smoking (never/ever). Multinomial logistic regression. \* Antibodies against *A. actinomycetemcomitans*, *P. gingivalis*, *P. intermedia*, *P. endodontalis*, and *T. forsythia*; \*\*Cross-reactiveantibodies,antigensPg-Rgp44,Aa-HSP60,andMAA-LDL.

**Figure 3.** Associations of the antibody and bacterial levels with endodontic lesion score and endodontic treatment score. Endodontic lesion score (A): score I, no endodontic lesions; score II, patients with ≥1 widened periapical space and/or 1 tooth with apical periodontitis; and score III, patients with ≥2 teeth with apical periodontitis. Endodontic treatment score (B): score I, no endodontic lesions; score II, apical periodontitis only in teeth with root canal fillings; score III, apical periodontitis in teeth without root canal fillings. The associations were investigated by using multinomial regression models adjusted for age, sex, marginal periodontitis (healthy, gingivitis, periodontitis), number of teeth, and smoking (never/ever). The estimates for lowest (reference), middle, and highest scores are depicted with white, gray, and black diamonds, respectively.

#### **4. Discussion**

We showed that apical periodontitis is associated with elevated levels of saliva IgA and IgG and serum IgG against bacterial antigens and saliva cross-reacting IgG, which recognise both bacterial and host epitopes. The associations were independent of marginal periodontitis. The local antibody response may contribute to the systemic IgG levels, which associate with the severity of apical periodontitis and arise mainly from untreated apical infections. High salivary IgA was associated with the number of widened periapical spaces, most likely indicating early endodontic infection.

Elevated levels of salivary total IgG associated with endodontic findings including root canal treatments, widened periapical spaces and radiographically diagnosed apical periodontitis. Both the presence and number of endodontic findings were significantly associated with total salivary IgG levels. In the case of total salivary IgA, the presence of root canal fillings and the number of widened periapical spaces, but neither the presence nor the number of teeth with apical periodontitis, associated with higher antibody levels. In health, the saliva IgGs mainly derive from the circulation by transudation through the gingival crevice. They comprise less than 15 percent of the total salivary immunoglobulins, as the major salivary immunoglobulin is secretory IgA produced by the salivary glands in mucosal plasma cells [38,39]. However, high concentrations of IgG and IgA and smaller amounts of IgM and secretory IgA have been detected within the periapical granulomas, in periapical cysts, as well as in root canal exudates with periapically affected teeth [40]. In addition, the total IgG and IgA levels detected from the periapical exudate were shown to correlate with clinical findings of the infected teeth [41]. As the half-life of IgA-class antibodies is only a few days, they are considered to reflect either recent or repeated exposure to the pathogen, while IgG is more stable, thus indicating a past, and maybe chronic, infection. Widened periapical spaces reflect either symptomatic teeth with irreversible pulpitis or precursors for established AP in necrotic teeth [1]. Since all determined antibody levels and bacterial concentrations correlated with the endodontic score, our results support the suggestion that the widened periapical spaces are likely to reflect early endodontic lesions [20].

When antibody response was studied in more detail, it was observed that salivary IgG levels against all studied species (*A. actinomycetemcomitans*, *P. gingivalis*, *P. intermedia*, *P. endodontalis*, and *T. forsythia)* were significantly higher in the groups with endodontic findings. In addition, patients with widened

periapical spaces had higher saliva IgA-antibodies against *P. endodontalis*. It is widely accepted that endodontic infections have a multimicrobial etiology [4], and several pathogens associated with marginal periodontitis, such as *P. intermedia*, *P. gingivalis*, *T. denticola* and *P. endodontalis*, are frequently detected in teeth with necrotic pulps [42–44]. As apical periodontitis is often restricted to the periapical tissues, it is not surprising that the amount of studied salivary and subgingival bacteria were not consistently associated with the endodontic findings. On the other hand, it is reported that in the case of combined endodontic-periodontal lesions, where apical periodontitis can be initiated either in the pulp or in the periodontium, the microbial profiles of apical lesions and periodontal pockets resemble each other [10]. In such cases, it is probable that bacteria enter the root canal from the periodontium via the apical foramen, dentinal tubules and accessory root canals [45].

Two major pathogens in marginal periodontitis, *A. actinomycetemcomitans* and *P. gingivalis*, express several virulence factors including *P. gingivalis*-specific gingipains degrading the extracellular matrix and bioactive peptides [46], as well as heat shock proteins (HSPs) produced by both species [47]. These proteins elicit strong antibody production and are also able to induce a variety of cross-reactive antibodies recognizing human epitopes such as HSPs and oxidized low-density lipoproteins (oxLDL). These cross-reactions are considered potential links between periodontitis and an increased risk of cardiovascular diseases [21]. The oxidation of LDL gives rise to various epitopes and a frequently used model of oxLDL include the immunodominant epitopes malondialdehyde (MDA) and malondialdehyde acetaldehyde (MAA). It is reported that the presence of antibodies binding to MDA-LDL is associated with both the progression of atherosclerosis and with the presence and severity of periodontitis [37,48–50]. A monoclonal IgM antibody to MDA-LDL recognizes *P. gingivalis* virulence factor gingipain (Rgp44) as an antigen [34] and *A. actinomycetemcomitans* heat shock protein 60 (Aa-HSP60) cross-reacts with MAA-LDL [35]. To our knowledge, this study is the first to show the association between apical periodontitis and salivary cross-reactive antibodies. Especially cross-reactive antibodies representing IgG-class were strongly associated with different endodontic conditions.

All measured parameters displayed positive trends with increasing number of endodontic findings as both the salivary IgA and IgG against bacterial antigens, as well as the cross-reacting IgG, were significantly associated with the highest endodontic score. In addition, the effect of endodontic treatment on antibody response was evident indicating that the levels of both saliva IgG against bacterial antigens and cross-reacting IgG were significantly higher in the patients with primary apical periodontitis compared to those who had apical periodontitis in teeth with root canal fillings. The aim of root canal treatment is to eradicate the biofilm from the infected root canal and prevent the recurrent infection. However, if the treatment is inadequately performed, some bacteria may survive and cause secondary infection. The microbiome of endodontically treated root canals consists of fewer bacterial species, and some of the species are more resilient to endodontic treatment [10]. This phenomenon may reflect to the levels of antibodies measured in this study.

The production of local IgGs is also enhanced in advanced marginal periodontitis by local plasma cells of the gingiva [51]. We repeated our main analyses in a subgroup of subjects without marginal periodontitis. Although the number of patients in this subgroup was low, the association between saliva IgG against bacterial antigens and primary apical periodontitis remained significant, suggesting further that IgG antibody response is independent of marginal periodontitis.

The main limitation of this study is that our study population consists of middle-aged and elderly participants, and thus the oral infections are very common. In addition, all participants had an initial indication for coronary angiography. Another restriction is the lack of intracanal bacterial samples; hence the bacterial analyses were only conducted from saliva and subgingival samples. Different methods were used for the detection of the antibody levels in serum and saliva, and not the same antibody panels were available. For instance, we did not have information on the serum cross-reactive antibodies, which will be an aim for future investigations. Also different methodologies were used for the bacterial analyses, since the subgingival samples were examined by checkerboard DNA–DNA hybridization and the saliva samples by qPCR.

Although apical periodontitis has been considered as a potential risk factor for systemic diseases such as coronary artery disease (CAD) [52], only a few studies have attempted to draw conclusions on the associations between apical periodontitis and systemic diseases. Evidence for the association between AP and CVDs, such as endothelial dysfunction [53], atherosclerosis [54], and coronary heart disease [55], has been reported in separate studies. However, recent systematic reviews suggest only modest participation of endodontic infection on the systemic levels of biomarkers and a moderate or low correlation between some systemic diseases and apical periodontitis [14,23,24,56]. In our recent study, we demonstrated a confounder-adjusted association between apical periodontitis and CAD [20]. In the present study, we showed that apical periodontitis may contribute to the levels of IgG in serum which link oral bacteria to CAD risk [31]. These serum antibodies have been repeatedly associated with prevalent and incident CVD as well as with subclinical atherosclerosis [57–60]. Also the saliva cross-reacting antibodies and immunoglobulins against bacterial antigens have been associated with increased risk for CAD [37].

Salivary immunoglobulins are potential biomarkers of oral infectious diseases, but the specific antigens should be selected carefully. This would be especially beneficial in the case of apical periodontitis, as the disease is often asymptomatic and remains undiagnosed. This study represents a limited set of antibodies against selected bacterial targets, and further research is needed to investigate the levels of antibodies against other bacterial species commonly found in infected root canals.

#### **5. Conclusions**

Our results suggest that the inflammatory condition caused by endodontic infections could be identified by the increased salivary IgG levels independently of marginal periodontitis. The levels of saliva IgG may have a small, but significant effect on the systemic levels of biomarkers, indicating the potential link between apical periodontitis and systemic diseases.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2077-0383/8/6/889/s1, Table S1: Bacterial species determined from subgingival plaque, Table S2: Saliva and serum antibody levels according to endodontic findings, Table S3: Saliva and subgingival bacterial levels according to endodontic findings, Table S4: Saliva and serum antibody levels according to endodontic scores, Table S5: Saliva and subgingival bacterial levels according to endodontic scores.

**Author Contributions:** Conceptualization, M.P. and P.J.P.; investigation—laboratory analyses, M.P., J.M.L., R.A., S.H., and P.J.P.; investigation—clinical examination, K.B., S.P., P.M., and J.S.; investigation—radiographic examination, J.M.L., K.B., S.P., A.S., P.M., and L.T.; formal analysis, M.P., A.S., and P.J.P.; writing—original draft preparation, M.P.; writing—review and editing, A.J., A.S., S.P., J.M.L., R.A., and P.J.P.; project administration, P.J.P; funding acquisition, S.P. and P.J.P.; supervision, P.J.P.

**Funding:** This research was funded by Academy of Finland [grant number 1266053 (P.J.P.), 1296541 (S.P.), and 1316777 (S.P.)], Paulo Foundation (P.J.P.), Finnish Dental Society Apollonia (P.J.P. and R.A.), University of Oulu Scholarship Foundation (R.A.) European Endodontic Society (P.J.P.), and Sigrid Juselius Foundation (P.J.P.).

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Review*
