*Aggregatibacter actinomycetemcomitans* **Biofilm Reduces Gingival Epithelial Cell Keratin Expression in an Organotypic Gingival Tissue Culture Model**

#### **Arzu Beklen 1, Annamari Torittu 1, Riikka Ihalin <sup>1</sup> and Marja Pöllänen 2,\***


Received: 11 November 2019; Accepted: 29 November 2019; Published: 1 December 2019

**Abstract:** Epithelial cells express keratins, which are essential for the structural integrity and mechanical strength of the cells. In the junctional epithelium (JE) of the tooth, keratins such as K16, K18, and K19, are expressed, which is typical for non-differentiated and rapidly dividing cells. The expression of K17, K4, and K13 keratins can be induced by injury, bacterial irritation, smoking, and inflammation. In addition, these keratins can be found in the sulcular epithelium and in the JE. Our aim was to estimate the changes in K4, K13, K17, and K19 expression in gingival epithelial cells exposed to *Aggregatibacter actinomycetemcomitans.* An organotypic gingival mucosa and biofilm co-culture was used as a model system. The effect of the biofilm after 24 h was assessed using immunohistochemistry. The structure of the epithelium was also studied with transmission electron microscopy (TEM). The expression of K17 and K19, as well as total keratin expression, decreased in the suprabasal layers of epithelium, which were in close contact with the *A. actinomycetemcomitans* biofilm. The effect on keratin expression was biofilm specific. The expression of K4 and K13 was low in all of the tested conditions. When stimulated with the *A. actinomycetemcomitans* biofilm, the epithelial contact site displayed a thick necrotic layer on the top of the epithelium. The *A. actinomycetemcomitans* biofilm released vesicles, which were found in close contact with the epithelium. After *A. actinomycetemcomitans* irritation, gingival epithelial cells may lose their resistance and become more vulnerable to bacterial infection.

**Keywords:** *Aggregatibacter actinomycetemcomitans*; keratins; organotypic gingival mucosa

#### **1. Introduction**

Periodontal disease is an inflammatory disease caused by growing biofilm, which gradually develops to a bacterial community rich in inflammophilic Gram-negative species [1]. The formation of such biofilm and the subsequent inflammation affects the supporting tissues of the tooth, including the epithelium and the connective tissue. Cytoskeletal intermediate filaments are essential for the structural integrity and function of the epithelial cells. Intermediate filaments constitute the acidic (Type I) and basic (Type II) keratins. The molecular weight of Type I keratin ranges from 40 to 57 kDa and encompasses keratins K9–K20. The molecular weight of Type II keratin ranges from 50 to 70 kDa and encompasses keratins K1–K8.

Keratins play an important functional role in the integrity and mechanical stability of both individual epithelial cells and cell–cell contacts in tissues. Thus, they contribute not only to the stability of the epithelium itself but also to the basement membrane attachment to the connective tissue underlying the epithelium. In the epithelia of internal organs, which are under little mechanical stress, there are only a few loosely-distributed keratin filaments in the cytoplasm. Conversely, keratins are abundant and are densely bound in the lining of outer surfaces. The composition of keratins varies depending on the epithelial cell type and the differentiation status of the epithelial cells. The keratin composition may also be affected by external stimuli, inflammation, or other types of disease development (e.g., cancer). The oral epithelium is a keratinizing form of epithelium and provides an effective physical barrier to microbial invasion. In the oral epithelium, basal dividing cells express simple epithelial cell keratins (K5, K14, and K19), whereas suprabasal cells express keratins typical of differentiated cells (K1, K10, K6, and K16). The keratin profile of the junctional epithelium (JE) differs from the oral epithelium, as only keratins typical for non-differentiated and rapidly dividing cells are expressed both basally (K5, K14, and K19) and suprabasally (K19). In addition, keratins such as K4, K13, and K17 have been found in the sulcular area of the dentogingival junction and may be induced by acute injury, bacterial irritation, smoking, and inflammation [2,3].

In contrast, in patients with severe and rapidly progressive periodontitis (previous term aggressive periodontitis), K17 gene expression was found to be repressed in disease site gingival samples compared to healthy site samples [4]. An open issue of significant interest is whether the expression of those keratins is altered by irritation from biofilm bacteria, such as *Aggregatibacter actinomycetemcomitans*, in periodontal diseases. The Gram-negative bacterium *A. actinomycetemcomitans* is an aggressive pathogen that is frequently associated with subgingival biofilms and has been strongly implicated in the development of rapidly progressive periodontal disease involving the invasion of *A. actinomycetemcomitans* into epithelial layers. Although *A. actinomycetemcomitans* represents only one species in multispecies periodontal biofilm, it most likely is able to suppress the host defense with its virulence factors. To gain insight into this question, we examined keratin K4, K13, K17, and K19 expression and distribution in an organotypic gingival tissue culture model co-cultured with a periodontopathogenic *A. actinomycetemcomitans* biofilm. Keratin K19 was chosen since it is typical and most dominant keratin in the JE [5,6] and we wanted to investigate how well the tissue culture model mimics the JE. We found that the expression of K17 and K19, as well as total keratin expression, decreased in the suprabasal layers of epithelium, which were in close contact with the *A. actinomycetemcomitans* biofilm. The decreased keratin expression may lead to decreased resistance of gingival epithelial cells to bacterial infection.

#### **2. Results**

#### *2.1. Control Cultures Showed Strong Expression of K17 and K19 and Only Weak or No Expression of K4 and K13*

Both control cultures grown without anything on the top of the epithelium or with an empty sterile filter disc showed similar immunohistochemical staining. Pancytokeratin staining was strong and was evenly distributed throughout the epithelium (Figure 1a,b). Similarly, the specific cytokeratins K17 and K19 were found to be expressed from the basal layer throughout the epithelium to the surface (Figure 1a,b). Keratin 4 was not found in the control cultures (Figure 2a,b), and K13 expression was only observed occasionally in single cells (Figure 2a,b).

#### *2.2. A. actinomycetemcomitans Biofilm Decreased the Expression of Keratin 17 and 19, Which Was in Accordance with the Decreased Expression of Total Keratin*

When exposed to pre-grown *A. actinomycetemcomitans* biofilm for 24 h, the human gingival keratinocytes showed no or very weak expression of K17 and K19 throughout the epithelium and especially in areas in close contact to the biofilm (Figure 1c). Some expression of K17 was evident further away from the biofilm-epithelium contact site (Figure 1c) as well as in the basal layer adjacent to the connective tissue. K19 expression was almost totally absent (Figure 1c). No expression of K4 or K13 was observed in the co-cultures with the biofilm (Figure 2c). The decrease in the expression levels of total keratins (pan-cytokeratin staining, Figure 1c) in the co-cultures with the *A. actinomycetemcomitans* biofilm was in accordance with the decrease in specific keratin expression.

**Figure 1.** *A. actinomycetemcomitans*biofilm decreased the total keratin aswell as K17 and K19 expressionin the suprabasal layers of the epithelium. (**a**,**b**) Immunohistochemical staining (peroxidase-3,3 -diaminobenzidine (DAB)) with anti-pan-cytokeratin/anti-cytokeratin 17/anti-cytokeratin 19 shows strong staining in the control cultures. (**c**) Pan-cytokeratin/K17/K19 expression is decreased in the co-cultures with the *A. actinomycetemcomitans* biofilm.

**Figure 2.** The expression of K4 and K13 was low in all of the tested conditions. (**a**,**b**) Immunohistochemical staining (DAB) with anti-cytokeratin 4 in the control cultures or (**c**) in the co-cultures with the *A. actinomycetemcomitans* biofilm shows no expression of K4. (**a**,**b**) Immunohistochemical staining (DAB) with anti-cytokeratin 13 shows only a few stained cells in the control cultures. (**c**) In the co-cultures with *A. actinomycetemcomitans* biofilm, no expression of K13 is observed.

#### *2.3. A. actinomycetemcomitans Biofilm Caused Necrosis of the Epithelial Surface*

Using TEM analysis, the control culture surface showed a thin layer of exfoliating, necrotic cells (Figure 3a,b). The cell structures and cell–cell contacts appeared normal. When exposed to the *A. actinomycetemcomitans* biofilm, the necrotic area of the epithelial surface was thick (Figure 4a,b), and the thickness of the necrotic area increased in areas in close contact to the biofilm and with an increasing amount of biofilm (Figure 4b). Necrosis of the epithelium in contact with the biofilm was consistently seen in EM, as visualized in Figure 4b, where right beneath the increasing mass of *A actinonmycetemcomitans* (on the left side of the picture) the necrotic epithelial layer increases clearly in thickness. This is a descriptive result, but was consistenly seen in EM. In the co-cultures, nuclear breakdown was observed in areas in close contact to the biofilm (Figure 4a,c). When in close contact with the epithelium, the biofilm bacteria released vesicles, and similar structures could be observed intraepithelially (Figure 5a,b). The necrotic area appeared to act as a barrier to the biofilm bacteria; however, a few structures which resembled bacteria were observed inside the epithelium after the 24 h co-culture (Figure 5c).

**Figure 3.** Normal structure of the epithelial cells and the epithelial cell nuclei. (**a**) Transmission electron microscopy (TEM) of the control cultures grown with nothing on the surface and (**b**) with the sterile membrane on the epithelium shows the normal structure of the cells and nuclei. On the top of the epithelium, a thin area of exfoliating and necrotic cells is observed (dark area).

**Figure 4.** *A. actinomycetemcomitans* biofilm caused necrosis of the epithelial cells. (**a**,**b**) TEM of the co-cultures with *A. actinomycetemcomitans* shows the thickening of the necrotic layer in areas in close contact to the biofilm and (**a**,**c**) disruption of the nuclei beneath the biofilm.

**Figure 5.** Vesicle-like structures are released by*A. actinomycetemcomitans*biofilm. (**a**)*A. actinomycetemcomitans* cells in the biofilm release vesicles, and (**b**) similar structures can be observed inside the epithelium. (**c**) (arrows) A few particles resembling single *A. actinomycetemcomitans* cells are observed inside the necrotic epithelium and inside the epithelial cells beneath the necrotic layer.

#### **3. Discussion**

Our major findings were that the *A. actinomycetemcomitans* biofilm caused decreased expression of cytokeratins, which is typical for the dentogingival junction and necrosis of the epithelial surface layers in close contact to the biofilm. Our tissue culture model appeared to mimic JE well, as K19, the typical keratin for JE, was highly expressed in the control cultures [6]. Our finding of decreased K19 expression adjacent to the *A. actinomycetemcomitans* biofilm is in disagreement with earlier reports showing that the inflammation in the periodontal pocket increases K19 expression [3,7]. However, our result may reflect the specific nature of the *A actinomycetemcomitans* biofilm–host tissue interaction. Previous work has shown that *A. actinomycetemcomitans* can invade buccal epithelial cells [8] and gingival tissue [9]. By decreasing the expression of one major structural protein, K19, in JE cells, *A. actinomycetemcomitans* could disturb the epithelial integrity, ease the invasion of the bacteria into deeper tissues, and lead to subsequent periodontal connective tissue destruction. In fact, structures resembling single *A. actinomycetemcomitans* cells were observed in the TEM analysis of epitheliums of co-cultures. Tissue destruction may even be further accelerated by decreased epithelial cell proliferation, which we showed in a previous study of tissue cultures exposed to an *A. actinomycetemcomitans* biofilm [10].

K17 was also strongly expressed in our control cultures. In vivo, K17 has been found in the sulcular epithelium of the dentogingival junction [2]. In addition, JE seems to express the genes encoding K17 [11]. Furthermore, it has been previously suggested that short chain fatty acids produced by some Gram-negative periodontal pathogens increase the expression of K17 [2] and that the inflammatory mediators would play a role in the regulation of the K17 expression [12]. Our finding that an *A. actinomycetemcomitans* biofilm decreased the expression of K17 seems to conflict with the earlier studies showing increased K17 expression after treatment with short chain fatty acids [2]. However, *A. actinomycetemcomitans*, although resistant to the antimicrobial activity of short chain fatty acids [13], produces only long chain fatty acids [14], which may at least partly explain these contradictory findings. Furthermore, K17 gene expression has been found depressed in clinical specimens from patients with severe and rapidly progressive periodontitis, which is in agreement with our results [4].

The model showed no evidence of K4 or K13 expression, which have been shown to be expressed in healthy oral sulcular epithelium [15]. Although K4 is typically absent in JE, its expression can be observed in JE of smoking periodontitis patients [15]. In our model, the *A. actinomycetemcomitans* biofilm did not increase or decrease K4 or K13 expression, of which the latter has been shown increasingly expressed in inflamed tissues [3]. However, the increase in K13 expression in inflammation may require the presence of host immune cells, such as macrophages, which were not included to our tissue co-culture model.

A closer investigation of the epithelial surface structure by TEM revealed that *A. actinomycetemcomitans* biofilm exposure caused necrosis of the epithelial surface in the co-culture models. The thick necrotic layer most likely inhibited the invasion of *A. actinomycetemcomitans* cells into the deeper layers of epithelium, as we could only detect a few structures resembling *A. actinomycetemcomitans* cells in the surface layers of the epithelium. However, the *A. actinomycetemcomitans* biofilm secreted high amounts of vesicles, which could have greater invasive potential than whole bacterial cells. For instance, *A. actinomycetemcomitans* vesicles have been shown to invade human HeLa and gingival fibroblast cells and release cytolethal distending toxin to the host cell nucleus [16,17]. *A. actinomycetemcomitans* vesicle-like structures could be observed in the epithelium. However, these vesicles may also originate from the host epithelial cells, and thus their origin needs to be confirmed in further studies.

In conclusion, *A. actinomycetemcomitans* biofilm decreases the expression of K17 and K19 and the total expression levels of keratins in an organotypic gingival mucosa model. However, the epithelium appears to utilize additional measures to withstand attack by the *A. actinomycetemcomitans* biofilm, which is suggested by the thicker necrotic layer between the epithelial cells and biofilm. Whether host immune cells and macrophages, in particular, change the keratin expression pattern during biofilm attack requires further investigation in a more complex environment.

#### **4. Materials and Methods**

#### *4.1. A. actinomycetemcomitans Biofilm Culture*

The *A. actinomycetemcomitans* biofilm cultures were generated as described previously [10]. Briefly, *A. actinomycetemcomitans* strain D7S was first cultured from trypticase soy agar blood plates. From the plates, an even bacterial suspension was made [18], and 5 <sup>×</sup> <sup>10</sup><sup>7</sup> cells/well were added to a 48-well plate containing porous filter discs. The biofilms were first grown in a rich trypticase soy broth medium for 24 h on a filter disc, and after which they were washed with a 0.85% NaCl solution. The cultivation was continued for an additional 24 h in glutamine supplemented RPMI-1640 medium. The 48-well plate also contained wells with sterile filter discs in the appropriate media for use as controls.

#### *4.2. Gingival Mucosa Co-Culture Models*

The *A. actinomycetemcomitans* biofilm/gingival mucosa co-culture models were constructed as described previously [10]. Briefly, human gingival fibroblasts [19] were grown in a collagen suspension with a cell culture insert (ThinCert™, Greiner Bio-One GmbH, Germany) for one day before <sup>4</sup> <sup>×</sup> 105 spontaneously immortalized human gingival keratinocyte cells [20] were seeded on top of the fibroblast-collagen matrix. The epithelial cells were cultured for one day submerged in the growth medium before the tissue model was lifted to the air–liquid interface. The model was air exposed for five days before a separately cultured *A. actinomycetemcomitans* D7S [21] biofilm (see above) was added on top of the tissue culture. The co-cultures were incubated without antibiotics in the cell culture medium. The gingival mucosa was co-cultured with the biofilm/control disc/no added components for 24 h, and the co-cultures were the fixed with a 10% formalin solution overnight. After the fixation step, the samples were embedded in paraffin and were sectioned.

#### *4.3. Immunohistochemical Staining of Keratins K4, K13, K17, and K19*

Before staining, the specimens were deparaffinized and heat-mediated antigen retrieval in 10 mM citrate buffer (pH 6.0) with microwaving was performed, which was followed by proteinase K treatment [22]. The staining was performed with a Dako TechMate™ 500 Plus Autostainer (Dako, Glostrup; Denmark) using the primary antibodies listed in Table 1 and the Dako REAL™ Detection System, Peroxidase/DAB+, Rabbit/Mouse (Code K5001; Dako) according to manufacturer's instructions.


**Table 1.** Primary anti-keratin antibodies used in the study.

#### *4.4. Transmission Electron Microscopic (TEM) Studies of the Biofilm-Epithelium Contact Site*

The *A. actinomycetemcomitans* biofilm gingival mucosa co-culture models were assembled as described above. The samples were prefixed for TEM with a freshly prepared 5% glutaraldehyde solution (5% glutaraldehyde, 0.16 M s-collidin-HCl buffer, pH 7.4) for at least 3 h at room temperature. After prefixation, the samples were washed with s-collidin-HCl buffer three times for 3 min each. Then, the samples were postfixed (1% OsO4, 1.5% K-ferrocyanide) for2h[23], washed with s-collidin buffer three times for 5 min each, and were dehydrated with ethanol (70% ethanol, 1 min at 4 ◦C; 96% ethanol, 1 min, at 4 ◦C; 100% ethanol, 30 min, at 4 ◦C; 100% ethanol, three times, 30 min each, at 20 ◦C). The dehydrated samples were embedded in epoxy using the following series: propylene oxide, two times, 15 min each; propylene oxide + epoxy resin + DMP (10:10:0.15), 2 h; epoxy resin + DMP (10:0.15), 12 h; epoxy resin + DMP (10:0.15), at 60 ◦C, 36 h. The sectioning of the samples was accomplished with an ultramicrotome to a thickness of approximately 70 nm. After sectioning, the samples were stained with uranyl acetate (1% uranyl acetate in pure water for 30 min) and were rinsed three times with pure water for 30 s. Finally, the samples were stained with lead citrate (0.3% lead citrate in pure water for 3 min) and were rinsed with water as in the previous step. The samples were examined with a JEM-1400 Plus Transmission Electron Microscope (JEOL USA, Inc., Peabody, MA, USA).

**Author Contributions:** Conceptualization, R.I. and M.P.; formal analysis, A.B., R.I., and M.P.; investigation, A.B. and A.T.; methodology, A.T. and M.P.; visualization, R.I. and M.P., project administration, R.I.; resources, R.I. and M.P.; supervision, R.I. and M.P.; writing—original draft, A.B., A.T., R.I., and M.P.; and writing—review and editing, R.I. and M.P., funding acquisition, R.I.

**Funding:** This work was supported by the Academy of Finland grant numbers 126557, 265609, 272960, and 322817 to R.I. A.B. was funded by The Scientific and Technological Research Council of Turkey (TUBITAK).

**Acknowledgments:** We thank Katja Sampalahti, Marja-Riitta Uola, and Essi Hautamäki for their skillful technical assistance in the organotypic tissue cultures and histological staining. The transmission electron microscopic studies were performed in the EM core facility, and the light microscopic imaging was performed at the Cell Imaging Core (Turku Centre for Biotechnology, University of Turku and Åbo Akademi University).

**Conflicts of Interest:** The authors declare no conflict 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*

### **Whole Genome Sequencing of** *Aggregatibacter actinomycetemcomitans* **Cultured from Blood Stream Infections Reveals Three Major Phylogenetic Groups Including a Novel Lineage Expressing Serotype a Membrane O Polysaccharide**

**Signe Nedergaard 1, Carl M. Kobel 1,2, Marie B. Nielsen 1,3, Rikke T. Møller 1, Anne B. Jensen 1,4 and Niels Nørskov-Lauritsen 1,\***


Received: 1 October 2019; Accepted: 19 November 2019; Published: 22 November 2019

**Abstract:** Twenty-nine strains of *Aggregatibacter actinomycetemcomitans* cultured from blood stream infections in Denmark were characterised. Serotyping was unremarkable, with almost equal proportions of the three major types plus a single serotype e strain. Whole genome sequencing positioned the serotype e strain outside the species boundary; moreover, one of the serotype a strains was unrelated to other strains of the major serotypes and to deposited sequences in the public databases. We identified five additional strains of this type in our collections. The particularity of the group was corroborated by phylogenetic analysis of concatenated core genes present in all strains of the species, and by uneven distribution of accessory genes only present in a subset of strains. Currently, the most accurate depiction of *A. actinomycetemcomitans* is a division into three lineages that differ in genomic content and competence for transformation. The clinical relevance of the different lineages is not known, and even strains excluded from the species sensu stricto can cause serious human infections. Serotyping is insufficient for characterisation, and serotypes a and e are not confined to specific lineages.

**Keywords:** taxonomy; core and accessory genes; average nucleotide identity; principal component analysis

#### **1. Introduction**

*Aggregatibacter actinomycetemcomitans* is a fastidious Gram-negative bacterium that inhabits the mucosal surfaces of humans and certain primates [1,2]. The species has attracted attention due to its association with periodontitis [3]. Particularly, a single serotype b clonal lineage is associated with a silent but aggressive orphan disease of adolescents that results in periodontitis and tooth loss [4]. Rather than being the causative agent of aggressive periodontitis, *A. actinomycetemcomitans* may be necessary for the action of a consortium of bacterial partners by suppressing host defences [5].

*A. actinomycetemcomitans* is a member of the HACEK group of fastidious Gram-negative bacteria (*Haemophilus*, *Aggregatibacter*, *Cardiobacterium*, *Eikenella* and *Kingella*), a recognized but unusual cause

of infective endocarditis responsible for 1.4% to 3% of cases [2]. A recent population-based study of the incidence of HACEK bacteraemia in Denmark identified 147 cases corresponding to an annual incidence of 0.44 per 100,000 population [6]. A retrospective study from New Zealand with 87 cases of HACEK bacteraemia confirmed a strong association with infective endocarditis, although the association with endocarditis ranged from 0 of 11 cases (*Eikenella corrodens*) to 18 of 18 cases (*A. actinomycetemcomitans*) [7].

Traditional classification of *A. actinomycetemcomitans* into serotypes is based on the chemical structures of the outer membrane O polysaccharide. Other studies have addressed the species' population structure by subjecting selected strains to multilocus enzyme electrophoresis [8], 16S rRNA gene sequencing [9], or multilocus sequence typing [10]. All methods identified an outgroup consisting of a subset of serotype e strains, but the grouping of serotypes was not consistent between methods. Finally, whole genome sequences (WGSs) of *A. actinomycetemcomitans* have become available. The first comparison of 14 strains found two major groups composed of serotypes a, d, e, plus f, and b plus c, respectively, while a serotype e strain outgroup showed a conspicuous lack of the cytolethal distending toxin gene cluster [11]. Jorth and Whiteley added three additional WGSs and calculated average nucleotide identity (ANI); strains within the two major groups were ~99% identical, while comparisons of strains belonging to separate groups disclosed significant differences (ANI of ~97%), and the outgroup strain was positioned outside the recommended species boundary [12]. The most recent comparison of WGSs included non-*actinomycetemcomitans Aggregatibacter* strains and was restricted to 397 concatenated core genes. The analysis suggested a division of the species into five clades: clade b (serotype b), clade c (serotype c), clade e/f (serotypes e and f), clade a/d (serotypes a and d), and clade e (outgroup serotype e strains) [13].

The few studies that have characterised *A. actinomycetemcomitans* from cases of bacteraemia were limited to serotyping [14], or serotyping supplemented with arbitrarily primed PCR [15]. Here, we present WGSs of 29 Danish bacteraemia isolates. We include five additional oral strains to characterise a novel group within the species designated lineage III.

#### **2. Results**

Twenty-nine blood stream isolates of *A. actinomycetemcomitans* were identified as part of an investigation of Danish HACEK bacteraemia cases [6]. Serotyping by PCR identified seven serotype a, 11 serotype b, 10 serotype c, and one serotype e strain—this distribution is similar to the observed prevalence among oral strains in Scandinavia ([16] and references therein). Comparison of WGSs did, however, show that the single serotype e strain (PN\_561) was unrelated to common isolates of this serotype, but clustered with the clade e' outgroup [13]. Moreover, one serotype a strain (PN\_696) was unrelated to the other six isolates of this serotype and did not cluster with any *A. actinomycetemcomitans* WGSs present in the public databases. We tried to identify further isolates of this peculiar genotype. An early characterisation by multilocus enzyme electrophoresis of 97 strains isolated over a period of 45 years identified four minor groups that deviated from the two major divisions [8]. By WGSs, the three strains of division V (HK\_907, HK\_973, and HK\_974) clustered with the aberrant serotype a blood stream isolate. An investigation of stored serotype a strains revealed two additional members of this lineage (K51, HK\_1710), and these five oral isolates are included in the comparison.

The neighbour-joining comparison of core gene sequences of 35 study strains (including the type strain NCTC 9710 of serotype c) plus seven selected reference sequences downloaded from the public databases is shown in Figure 1. One blood stream isolate (PN\_561) plus a reference sequence comprise the clade e' outgroup that is used to root the tree. Although serotype b and serotype c strains are distributed in two separate branches, the overall population structure of the species consists of three separate lineages.

**Figure 1.** Neighbour-joining dendrogram of *Aggregatibacter actinomycetemcomitans* based on 1146 concatenated core genes (1,104,001 nucleotides) of 42 whole genome-sequenced strains. The type strain is designated with a superscript T. Twenty-nine strains were from cases of bacteraemia (designated PN), five oral strains were included to describe lineage III, and seven WGSs were downloaded from Genbank; see Supplementary Table S1 for further description and origin of strains. The outgroup (strains PN\_561 and SC1083) reduces the number of core genes and should probably be excluded from the species. Serotypes and phylogenetic lineages are shown; nt, non-typeable by immunodiffusion with antisera [8]. The bar represents 5000 residue substitutions.

#### *2.1. Delineation of the Species*

Serotype e strain PN\_561 was cultured from a case of bacteraemia in 2005. The patient underwent aortic valve replacement for infective endocarditis, and the bacterium was also cultured from the removed valve. The strain was identified as *Actinobacillus (Aggregatibacter) actinomycetemcomitans* based on selected phenotypic tests; re-examination using matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry confirmed the identification with a log-score value above 2. The clinical case demonstrates the aggregative potential of the clade e' outgroup, and the WGSs reveal the presence of a *tad* cluster and leukotoxin operon, but a lack of the cytolethal distending toxin gene cluster (GenBank accession number VSEC00000000). We performed in silico DNA–DNA hybridisation (DDH) by use of the Genome-to-Genome Distance Calculator 2.1, which estimates the DDH values that would have resulted from classic hybridisation experiments [17]. The in silico DDH value for strain PN\_561 versus the type strain NCTC 9710 was 54.7%, which is below the phylogenetic species boundary of 70% suggested by classic hybridisation [18]. Similar in silico DDH values were obtained for PN\_561 versus selected reference strains from other serotypes (range: 54.2% (D18P-1, serotype f) to 55.4% (HK\_1651, serotype b)).

Average nucleotide identity (ANI) is a powerful method to estimate overall genome relatedness and is widely used as a substitute of the classic DDH methods. ANI is calculated for two genome sequences by breaking the genome sequence of the query strain into 1020-bp-long fragments. Then, nucleotide identity values for individual fragments of the query strain and the genome of the subject strain are calculated using the NCBI BLASTn program. Using OrthoANI [19], where both genomes are

fragmented and only reciprocal BLASTn hits are included, strain PN\_561 was 93.89% identical with the type strain of *A. actinomycetemcomitans*; restricting ANI calculation to 1146 concatenated core gene sequences gave an ANI value of 94.72%. An ANI threshold value of 95% is considered the species boundary [20,21], and the two clade e' outgroup strains are excluded from further analysis.

#### *2.2. Natural Competence and Genomic Characteristics of Lineages*

Twenty-eight invasive strains plus five oral strains of *A. actinomycetemcomitans* were tested for natural competence by plating on kanamycin-containing agar in the presence of donor DNA. All strains belonged to the three dominant serotypes a–c. In accordance with previous findings [22], only a subset of serotype a strains were competent for transformation, while serotype b and c were invariably noncompetent. Specifically, competence was associated with lineage II (strains PN\_437, PN\_559, PN\_563, PN\_567, and PN\_688), while all strains of lineage III were noncompetent. Competence is a primary mechanism of horizontal gene transfer and DNA acquisition in bacteria, and it has previously been shown that competent strains of *A. actinomycetemcomitans* are, on average, 200,000 bp larger than noncompetent strains [12]. In accordance with this, the mean genome size of strains of lineage II was 2.26 Mb, while the mean genome size of invariably noncompetent strains of lineage I was 2.07 Mb. With an average genome size of 2.22 Mb, lineage III was closely related to the mean size of competent lineage II, but the range (2113–2316 Kb) did overlap the size of the largest genome in lineage I (strain PN\_738; 2134 Kb).

Analysis of competence genes in the six strains of lineage III revealed major disruptions within the essential repository (Figure 2). All strains carried large (11–28 Kb) mobile elements impeding *comM*, but no genome contained all 16 genes, and 31 of 84 identified competence genes were inactivated. It is possible that future analysis will reveal some competent strains of lineage III, but the investigated genomes indicate an ancestral noncompetent lineage, which questions the relationship between competence and genome size.

**Figure 2.** Intact and inactivated competence genes from lineage III strains using strain HK\_1651 as reference. Red genes have premature stop codons caused by single base substitutions, single base deletions, or double nucleotide insertions. Yellow genes are disrupted by large insertions (11–28 Kb). Green genes have deletions (150–722 bp) of either the first (*sxy*, *pilC*, *comE*) or the last part (*comC*) of a gene. The first 73 amino acids of gene *sxy* in strains HK\_907 and PN\_696 are replaced with an unrelated coding sequence (blue genes). Green and blue arrows mark the transition from competence gene to unrelated sequence, or vice versa.

Excluding the two clade e' outgroup strains from Figure 1 increased the number of core gene sequences from 1146 to 1357 but did not distort the dendrogram (not shown). The total number of genes identified by Prokka was 4631, and the distribution of accessory genes is of interest. We used principal component analysis (PCA) of the dichotomous presence/absence of orthologs, analysing 3274 genes present in 1–39 of 40 strains. Figure 3A shows a scatterplot of this accessory genome of 40 *A. actinomycetemcomitans* strains, represented by the two principal components that account for a major part of the variance in the gene presence/absence matrix (Figure 3B). Analysis of the accessory genome supports the division of the species into three distinct lineages. The first principal component (PC) primarily serves to separate lineage I from lineages II and III, while PC2 dissociates all three lineages (Figure 3A). Indeed, nine of the 10 annotated genes (excluding hypothetical genes) with the highest loadings in PC1 were only present in 23 strains of lineage I, while nine of the 10 annotated genes with the highest negative loadings were predominantly associated with 17 strains of lineage II plus III (range 15–18). For PC2, the highest loadings were associated either with 34 strains of lineage I plus II, or with six strains of lineage III, while the highest negative loadings were more unevenly distributed among lineages (not shown).

**Figure 3.** (**A**), principal component analysis of presence/absence of accessory homologs in 40 strains of *A. actinomycetemcomitans*. (**B**), line plot of the eigenvalues of factors or principal components in the analysis. The two principal components depicted in 3A comprise 30% of the sum of variances of all individual principal components (3B).

By scatterplot of PC1 vs. PC2, strains of lineage III appear more diverse than those of lineage II, which are more diverse than those of lineage I (Figure 3A); this may, in part, be caused by the decreasing number of strains included in the lineages. The accessory gene content of serotype g strain NUM4039 was the most divergent in lineage II, followed by serotype a strain D7S-1 (Figure 3A); this segregation is not apparent from single-nucleotide polymorphism (SNP) analysis of core gene sequences (Figure 1). Three strains of lineage III clustered closely by PCA, while HK\_907, HK\_973 and PN\_696 were more individually positioned (Figure 3A). Again, this pattern of accessory gene content is not reflected in the SNP analysis of core gene sequences (Figure 1).

Lineage-specific gene homologs were abstracted from the Roary output, and the annotated genes (excluding hypotheticals) are listed in Supplementary Table S2. Thirty-seven gene homologs (14 annotated, 23 hypothetical) were detected in 23 lineage I strains and not in strains of other lineages; the corresponding numbers are 16 annotated genes only present in 11 lineage II strains, and 20 annotated genes only present in six lineage III strains. Several unexpected associations are observed, such as homologs of the multidrug exporter MdtA restricted to lineage II, and several CRISPR-associated nucleases confined to lineage II or III. Clearly, phenotypic and pathobiological significance must be addressed in biologic experiments, and the true relationship between marker genes and lineages awaits analysis of a larger number of strains. Nevertheless, the existence of lineage-specific marker genes encourages the development of lineage-specific PCRs that will be simpler and more informative than the currently employed serotype-specific PCRs.

#### **3. Discussion**

*A. actinomycetemcomitans* was linked to aggressive periodontitis in 1976, and this association was supported by elevated serum antibodies in patients [5]. Three distinct bacterial surface antigens were identified by 1983, and this typing has remained the cornerstone of the initial characterisation of cultured strains. More advanced methods for dissection of the population structure [8–10] have not gained acceptance, although multilocus sequence typing (MLST)holds promise as a general, versatile typing method. WGSs have unequivocally shown that serotyping is inadequate for assessment of the phylogenetic positioning of a clinical strain.

We performed whole genome sequencing of *A. actinomycetemcomitans* cultured from blood stream infections. By serotype, the population resembled oral strains from our region with almost equal proportions of the three dominant serotypes. However, comparison of WGSs revealed some interesting observations. First, the serotype e strain PN\_561 was not related to common serotype e strains within the species but belonged to an outgroup that has been designated clade e' [13]. Aberrant strains of serotype e also deviate by 16S rRNA [9] and MLST [10,23], but distinctive phenotypic markers have not yet been described. Assessment of overall genome relatedness by ANI and in silico DDH positioned PN\_561 and the reference clade e' outgroup strain outside the species boundary. These strains are negative for the cytolethal distending toxin genes but encode the *tad* cluster that is decisive for autoaggregation and adherence to a wide range of solid surfaces. Strain HK\_921 of the clade e' outgroup was included in the investigation of *Aggragatibacter* strains that resulted in the description of the new species *Aggregatibacter kilianii*, and the difference between HK\_921 and the other strains of *A. actinomycetemcomitans* was similar to or exceeded the difference between *Aggregatibacter kilianii* and *Aggregatibacter aphrophilus* [24]. Valid publication of bacterial names generally requires key phenotypic tests for discriminatory purposes, while identification by matrix-assisted laser desorption/ionisation time-of-flight (MALDI-TOF) mass spectrometry is only dependent on the composition of mass spectre in the database. The frequent detection of these outgroup strains in clinical samples may give impetus to taxonomic rearrangements in genus *Aggregatibacter*; the clinical significance of the clade e' outgroup is emphasised by strain PN\_561 being the cause of infective endocarditis.

Second, strain PN\_696 of serotype a was neither related to other strains of this serotype nor to any previously deposited WGSs of *A. actinomycetemcomitans*. We were able to identify five additional strains cultured from the oral cavity with high genotypic resemblance to PN\_696. In contrast to the clade e' outgroup, aberrant serotype a strains were positioned inside the *A. actinomycetemcomitans* species boundary. Accepted and used designations are helpful, and clade a' would be in line with a recent description of the population structure of *A. actinomycetemcomitans* by WGSs [13]. We suggest a different term. The population structure of *A. actinomycetemcomitans* is more accurately described by a division of the species into three phylogenetic lineages I–III. Although clearly separate entities, this designation would bundle serotype b and c strains into a common lineage. Strains of five separate serotypes are present in lineage II, while a subset of serotype a strains constitute lineage III.

#### **4. Materials and Methods**

#### *4.1. Bacterial Strains and DNA Accession Numbers*

Twenty-nine strains of *A actinomycetemcomitans* were collected from blood stream infections from seven Danish departments of clinical microbiology. Initial analysis of WGSs revealed a peculiar sequence from a serotype a strain, and we were able to identify five additional oral strains with high resemblance to the invasive strain; these five strains were included in the study, as was the type strain NCTC 9710 of serotype c. Representative WGSs of serotypes other than c were downloaded from the public databases, including an additional serotype e strain allocated to the outgroup tentatively designated clade e' [13]. Supplementary Table S1 lists the origin, host characteristics, and accession numbers of all investigated strains and sequences.

#### *4.2. Identification and Phenotyping*

All clinical strains were subjected to renewed identification by matrix-assisted, laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF) as described [24]. Serotyping by PCR was performed as previously described [25]. Natural competence was investigated by transformation assays [22] using donor DNA from the D7S *hns* mutant that carries the kanamycin resistance gene cluster from pUC4K in the *H-NS* gene [26]. In brief, 20 μl of a dense bacterial suspension (OD600nm = 0.3) was spread in a small area (diameter of ~10 mm) on a brain-heart infusion (BHI) agar plate. After incubation for 2 h, 10 μl of donor DNA (~1 μg) was added, gently mixed by using a loop, and further incubated for an additional 6 h. The bacteria were collected with a cotton swap, suspended in 400 μl of BHI broth, and plated on selective media containing kanamycin 50 (μg/mL); in parallel, diluted samples were plated on chocolate agar plates. Colonies were counted after two and seven days, and the transformation frequency was the ratio of the number of transformants to the number of cells plated.

#### *4.3. DNA Sequencing, Genome Assembly, and Analysis*

DNA libraries were prepared from 200 ng of genomic DNA with a Sciclone NGS robot (PerkinElmer), using the QIAseq FX DNA Library Kit (QIAGEN), according to the manufacturer's protocol. Quality control of the libraries was conducted by on-chip electrophoresis (TapeStation, Agilent) and by Qubit (Thermofisher) concentration measurements. Dual-indexed paired-end sequencing (2 by 150 bp) was performed with an Illumina NextSeq 500 system (Illumina) aiming at 200 x coverage. Paired demultiplexed FASTQ files were generated using CASAVA software (Illumina), and initial quality control was performed using FastQC. Reads were assembled using Unicycler (version 0.4.7), an optimiser for SPAdes (version 3.13.9). Contigs with a length below 500 nt were disregarded. Draft assemblies of study strains plus FASTA files from reference strains downloaded from GenBank were annotated with Prokka [27]. Roary [28], a rapid, large-scale, prokaryote pan-genome analysis tool, was used with default settings for identification of core genes to create clusters of genes that share amino acid sequence similarity and coverage above a given threshold and orders strains by the presence or absence of orthologs. Core genes (present in all strains) were aligned with ClustalW and concatenated, before SNPs were called and evolutionary analyses conducted in MEGA X [29].

In silico DNA hybridisation between selected strains was performed with the Genome-to-Genome Distance Calculator (version 2.1), using standard settings and the recommended identity/high-scoring segment pair length calculation [17]. Average nucleotide identity (ANI) values of draft genomes were calculated using online tools (http://www.ezbiocloud.net/sw/oat) [19]; additionally, ANI values of concatenated *A. actinomycetemcomitans* core genes were calculated with Panito (version 0.0.2b1) (https:// github.com/sanger-pathogens/panito). Principal component analysis (PCA) of binary absence/presence gene matrix from Prokka was computed in R with built-in packages.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2076-0817/8/4/256/s1, Table S1: List of strains and sequence accession number; Table S2: Lineage-specific, annotated genes.

**Author Contributions:** All authors made a substantial, direct, and intellectual contribution to the work. N.N.-L. conceived and planned the study and made the first draft of the manuscript. A.B.J., M.B.N., R.T.M., and S.N. performed molecular and phenotypic experiments. C.M.K. completed the bioinformatics. All authors approved the manuscript for publication.

**Funding:** This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

**Acknowledgments:** Members of the Danish HACEK Study Group Jens J. Christensen, Slagelse Hospital; Gitte Hartmeyer, Odense University Hospital; Kristian Schønning, Hvidovre Hospital; Lisbeth Lützen, Aarhus University Hospital; Claus Moser, Rigshospitalet; Bente Olesen, Herlev and Gentofte Hospital; and Henrik C. Schønheyder, Aalborg University Hospital provided strains and clinical information on patient cases.

**Ethical Statement:** The study is an extension the incidence of HACEK bacteraemia in Denmark [6] and was conducted in accordance with the regional guidelines for the use of clinical and laboratory data and was approved by the Danish Data Protection Agency (record number 2012-58-0004) and the National Board of Health (record number 3-3013-1170/1).

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

#### **Abbreviations**

ANI, average nucleotide identity; BHI, brain-heart infusion; DDH, DNA–DNA hybridization; HACEK, *Haemophilus*, *Aggregatibacter*, *Cardiobacterium*, *Eikenella* and *Kingella*; MALDI-TOF, matrix-assisted laser desorption/ionisation time-of-flight; MLST, multilocus sequence typing; PCA, principal component analysis; SNP, single-nucleotide polymorphism; WGSs, whole genome sequences.

#### **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*

### **Di**ff**erential Cell Lysis Among Periodontal Strains of JP2 and Non-JP2 Genotype of** *Aggregatibacter actinomycetemcomitans* **Serotype B Is Not Reflected in Dissimilar Expression and Production of Leukotoxin**

**Anne Birkeholm Jensen 1,2, Marianne Lund 2, Niels Nørskov-Lauritsen 2, Anders Johansson 3, Rolf Claesson 4, Jesper Reinholdt <sup>5</sup> and Dorte Haubek 1,\***


Received: 11 October 2019; Accepted: 26 October 2019; Published: 30 October 2019

**Abstract:** Leukotoxic potential of *Aggregatibacter actinomycetemcomitans* strains has been studied by the use of several methods, and results differ depending on the methods used. The aim of the present study was to perform a comprehensive examination of the leukotoxic potential of a collection of *A. actinomycetemcomitans* strains by use of three quantitative methods, Western blotting, ELISA, and mRNA expression assay and compare these results with previous data obtained by a cell lysis assay. A higher leukotoxic potential among JP2 genotype strains compared to non-JP2 genotype strains of *A. actinomycetemcomitans* was found by Western blotting, ELISA and mRNA expression assay. Leukotoxicity as determined by cell lysis assay showed a variation among strains examined, not only depending on being part of JP2 genotype *vs*. non-JP2 genotype group of *A. actinomycetemcomitans*. The leukotoxicity of *A. actinomycetemcomitans* strains as determined by cell lysis assay did not correspond to the leukotoxic potential of *A. actinomycetemcomitans* strains as determined by three quantitative methods. A comparison of the results obtained by ELISA and mRNA expression assay showed a reasonable correlation between these two methods. It seems important to use more than one method to assess the LtxA-related virulence capacity of *A. actinomycetemcomitans* in order to obtain comprehensive understanding of the leukotoxic potential of *A. actinomycetemcomitans* strains.

**Keywords:** mRNA assay; quantitative ELISA; cell lysis assay; leukotoxin; JP2 genotype

#### **1. Introduction**

*Aggregatibacter actinomycetemcomitans* is a Gram-negative member of the human oral microbiota [1] and is involved in human infections and diseases [2–5]. One of these diseases is periodontitis, characterized by destruction of tooth-supporting periodontal tissues due to an inflammatory response [6]. *A. actinomycetemcomitans* possesses several virulence factors [3], of which the production of the RTX (repeats-in-toxin) leukotoxin (LtxA) has received much attention [7–12]. LtxA induces cell lysis, degranulation and an inflammatory response in human leukocytes by interaction with the β2-integrins in the cell membrane of human immune cells [8,13,14]. A 530-bp deletion in the promoter region of the

LtxA gene operon was identified in the original JP2 strain, cultured from a young individual diagnosed with juvenile periodontitis [15]. This discovery marked the beginning of an era, where the detection of the 530-bp deletion categorized highly leukotoxic *A. actinomycetemcomitans* strains as a member of the JP2 clone of *A. actinomycetemcomitans*, possessing enhanced virulence. Based on in vitro studies, it was reported that the JP2 genotype of *A. actinomycetemcomitans* has a 10–20 fold higher lytic activity than the non-JP2 genotypes of *A. actinomycetemcomitans* [15]. Furthermore, longitudinal clinical studies reported a clear correlation between being carrier of the JP2 genotype of *A. actinomycetemcomitans* and having an increased risk of developing periodontitis at a young age [16,17].

Often the leukotoxic potential of *A. actinomycetemcomitans* is characterized by the leukotoxicity in in vitro studies [11,15,18–21]. This leukotoxicity is determined by use of cell lysis assays, where human immune cells are exposed either to the bacteria or to purified LtxA, and most often the leukotoxicity is correlated with the JP2 or non-JP2 genotype of the *A. actinomycetemcomitans* strain. Therefore, it has been the common believe that the leukotoxicity of an *A. actinomycetemcomitans* strain could be explained by the 530-bp deletion, an explanation supported by recent results presented by Sampathkumar and co-workers [21]. However, *A. actinomycetemcomitans* strains without the 530-bp deletion with high leukotoxicity according to results obtained by cell lysis assay have been reported on [20]. Furthermore, the leukotoxic potential of *A. actinomycetemcomitans* strains has been explained by different mechanisms [15,18,19]. A higher expression of the mRNA, encoding the *ltxA,* has been reported to correspond with a higher leukotoxic potential [15,18], but also a variation in the activity of LtxA in some *A. actinomycetemcomitans* has been reported [19]. Only a few studies have compared the leukotoxicity of *A. actinomycetemcomitans* with the LtxA expression and production defined as a quantification of the LtxA, e.g., by enzyme-linked immunosorbent assay (ELISA), or mRNA coding for LtxA [15,18,21], but it seems reasonable to assume that the two factors are related.

In the present study, we aimed to investigate if the leukotoxicity of a collection of Ghanaian *A. actinomycetemcomitans*, serotype b, is related to the LtxA production of the strains determined by Western blotting and ELISA, and LtxA expression as determined in a mRNA expression assay. Furthermore, we wanted to compare the results obtained by ELISA and mRNA expression assay to elucidate the relationship between the two quantification methods used to determine leukotoxic potential as the LtxA production and expression of *A. actinomycetemcomitans*, respectively.

#### **2. Results**

#### *2.1. LtxA Expression and Production of the 20 Ghanaian Strains by Western Blotting, ELISA and mRNA Expression Assay*

Western blotting clearly showed that the JP2 genotype strains of *A. actinomycetemcomitans* had a higher LtxA production than the non-JP2 genotype strains (Figure S1).

The non-JP2 genotype strains 575G, 605G, 638G, 443G, and 486G were previously characterized as having high leukotoxicity by cell lysis assay (Table 1) [20].

Strains 443G and 486G might have a tendency towards a higher LtxA production than the other non-JP2 genotype strains, but not at comparable levels as to the JP2 genotype strains. Furthermore, the Western blotting demonstrated that the strains with high LtxA production had a higher amount of LtxA in the growth supernatant than in the cell pellet extract (Figure S1).

By visual inspection of Figure 1, the ELISA showed the same division of JP2 and non-JP2 genotype strains of *A. actinomycetemcomitans* as demonstrated by the Western blotting (Figure S1). In Figures 1 and 2, the cell lysis assay (LDH) characterizes the leukotoxicity of the strains, and the results are from the publication by Höglund Åberg and coworkers (2014) [20]. The results are given as a percentage of total lysis of THP-1 cells by Triton X. The ELISA determines the leukotoxic production of the *A. actinomycetemcomitans* strains as a percentage of the reference *A. actinomycetemcomitans* strain HK921, and results are the mean of two separate runs. The mRNA expression assay determines the leukotoxin expression of *A. actinomycetemcomitans* strains as a ratio to *adk* and *pgi*, and the results are the mean of three separate runs.


**Table 1.** Characterization of the collection of *A. actinomycetemcomitans* strains, serotype b.

\* As defined by Höglund Åberg and coworkers (2014): 0–30% LDH release = low leukotoxicity, 31–60% = average leukotoxicity and 60% <sup>≤</sup> <sup>=</sup> high leukotoxicity. <sup>1</sup> Originating from Portugal, <sup>2</sup> Originating from Cape Verde Islands, <sup>3</sup> Originating from Tel Aviv, <sup>4</sup> Originating from Algeria, <sup>5</sup> Originating from Morocco, <sup>6</sup> Originating from Ghana.

**Figure 1.** Leukotoxicity, leukotoxin production and leukotoxin expression of 20 Ghanaian *A. actinomycetemcomitans*, serotype b, JP2 and non-JP2 genotype strains determined by a cell lysis assay (LDH), an ELISA, and a mRNA expression assay.

**Figure 2.** Box plot of the distribution of the 20 Ghanaian *A. actinomycetemcomitans* strains according to JP2 genotype and non-JP2 genotype strains.

Two non-JP2 genotype strains of *A. actinomycetemcomitans* (443G and 486G) showed almost the same LtxA production as one of the JP2 genotype strains of *A. actinomycetemcomitans*. However, in general the JP2 genotype strains of *A. actinomycetemcomitans* showed a higher LtxA production than the non-JP2 genotype strains of *A. actinomycetemcomitans*. Furthermore, this finding was

supported by comparing the group of JP2 genotype strains of *A. actinomycetemcomitans* to the group of non-JP2 genotype strains of *A. actinomycetemcomitans* (Figure 2). The group of JP2 genotype of *A. actinomycetemcomitans* had a statistically significant higher LtxA production than the group of non-JP2 strains of *A. actinomycetemcomitans* (*p* < 0.05) (Figure 2).

Expression of mRNA encoding LtxA did not correlate with the leukotoxicity either (Figure 1). By visual inspection, the mRNA expression assay showed a greater variation in the LtxA expression of the *A. actinomycetemcomitans* strains and found one non-JP2 genotype strain (638G) with a LtxA expression at comparable levels to the JP2 genotype strains. The average mRNA expression of JP2 genotype strains of *A. actinomycetemcomitans* was higher than observed for non-JP2 genotype strains, and the difference did attain statistical significance (Figure 2) (*p* < 0.05).

Conclusively, the high leukotoxicity of some of the non-JP2 strains of *A. actinomycetemcomitans* found by Höglund Åberg and coworkers (2014) could not be reproduced according to the LtxA expression and production found by neither the Western blotting, the ELISA, nor the mRNA expression assay.

#### *2.2. Comparison of Results Obtained by ELISA and mRNA Expression Assay*

For further comparison of the two quantitative methods, ELISA and mRNA expression assay, 45 strains of *A. actinomycetemcomitans*, serotype b, were analyzed (Figure 3). The ELISA determines the leukotoxin production of the *A. actinomycetemcomitans* strains as a percentage of the reference strain HK921, and the results are the mean of two separate runs. The mRNA expression assay determines the leukotoxin expression of the *A. actinomycetemcomitans* strains as a ratio to *adk* and *pgi*, and the results are the mean of three separate runs.

**Figure 3.** The distribution of the expanded bacterial collection of 45 *A. actinomycetemcomitans* strains for comparison of the results obtained by ELISA and in mRNA expression assay.

By visual inspection of Figure 3, both methods divided the *A. actinomycetemcomitans* strains, serotype b, according to their genotype, although the division was most clearly demonstrated by ELISA. A comparison of the group of JP2 genotype strains of *A. actinomycetemcomitans* and the group of non-JP2 genotype strains of *A. actinomycetemcomitans* is illustrated in Figure 4.

**Figure 4.** A comparison of the group of JP2 genotype strains and the group of non-JP2 genotype strains of *A. actinomycetemcomitans* based on the expanded bacterial collection of 45 strains for comparison of results obtained by ELISA and in mRNA expression assays by use of log-transformed data.

As for the analysis of the 20 Ghanaian *A. actinomycetemcomitans* strains, the group of JP2 genotype strains had a higher LtxA production and expression than the group of non-JP2 genotype strains of *A. actinomycetemcomitans* (*p* < 0.05) by both ELISA and mRNA expression assay.

By visual inspection of the scatter plot (Figure 5), where the points correspond to a particular strain, a reasonable fit between the results obtained by the ELISA and in the mRNA expression assay was seen.

**Figure 5.** A comparison of ELISA and mRNA expression assay by an ordinary least square regression model based on the expanded bacterial collection consisting of 23 non-JP2 genotype strains of *A. actinomycetemcomitans* (black dots) and 22 JP2 genotype strains of *A. actinomycetemcomitans* (white dots) by use of log-transformed data.

However, the points tended to fall above the identity line indicating bias in the data. An ordinary least squares regression model was used to compare the ELISA and the mRNA expression assay based on the results of the collection of 45 *A. actinomycetemcomitans*, serotype b, strains. The regression fit of the two methods is given by the R2 at 0.37 indicating a reasonable relationship between the methods. The intercept differed from 0.0 (0.85) indicating that the regression model possesses consistent bias. A slope > 0.00 (1.00) (*p* < 0.05) illustrated a proportional (log-log linear) relationship between the two methods. Furthermore, the scatter plot (Figure 5) illustrated a greater variation in the data among the non-JP2 genotype of *A. actinomycetemcomitans*, whereas the data for the JP2 genotype strains of *A. actinomycetemcomitans* was more equal.

#### **3. Discussion**

In the present study, we aimed to investigate if the leukotoxic potential determined by a leukotoxicity assay of 20 Ghanaian *A. actinomycetemcomitans*, serotype b, is related to the LtxA production and expression of the strains determined by three different quantitative methods. Quantification of the LtxA production and expression using Western blotting, ELISA and mRNA expression assay could not reproduce the high leukotoxicity found among some of the Ghanaian non-JP2 genotype strains by use of cell lysis assay previously reported on by Höglund Åberg and coworkers (2014) [20]. All JP2 genotype strains of *A. actinomycetemcomitans* showed high LtxA expression and production. Furthermore, the non-JP2 genotype strains of *A. actinomycetemcomitans*, serotype b, previously characterized as having a high leukotoxicity, were all below the JP2 genotype strains when using LtxA expression and production assays. Furthermore, supplementary analysis of the expanded bacterial collection of *A. actinomycetemcomitans* strains, serotype b, revealed a reasonable relationship between the results obtained by ELISA and mRNA expression assay.

To our knowledge, this is the first study relating the leukotoxicity of a large collection of *A. actinomycetemcomitans* strains, serotype b, to the LtxA expression and production determined by three different quantitative methods. Studies have previously related the leukotoxicity of *A. actinomycetemcomitans* strains to the LtxA expression and production, often determined by quantification of mRNA or by Western blotting [15,18,21]. In addition, many studies have related a high leukotoxicity and a high LtxA expression to the presence of the 530-bp deletion in the strains [14,15,18,21,22]. The results from the present study do agree with such previous findings (Figure S1, Figures 1–5). Both ELISA and Western blotting characterize the JP2 genotype of *A. actinomycetemcomitans* as having a higher LtxA production than the non-JP2 genotype strains (Figure S1 and Figure 1). Furthermore, the comparison of the group of JP2 genotype strains and the group of non-JP2 genotype strains did classify the JP2 genotype strains with a higher LtxA expression and production than the non-JP2 genotype strains regardless of using mRNA expression assay or ELISA (Figures 2 and 4). The difference between JP2 and non-JP2 genotype strains of *A. actinomycetemcomitans* applied to both the collection of the 20 Ghanaian *A. actinomycetemcomitans* and to the expanded collection consisting of 45 *A. actinomycetemcomitans*, serotype b. Surprisingly, a poor relation between the leukotoxicity and the LtxA expression and production of the Ghanaian *A. actinomycetemcomitans* strains, serotype b, was found (Figure S1 and Figure 1).

Since other researchers have reported on correlating data concerning the leukotoxicity and the LtxA expression, e.g., the mRNA expression, of *A. actinomycetemcomitans,* one has to speculate if methodological complexities may explain the lack of correlation between the results obtained by cell lysis assay compared to the three quantitative assays in the present study. It has been proposed that other factors than the *ltxA* expression, LtxA production, and LtxA secretion may influence the leukotoxic potential of a strain; Diaz and co-workers did describe a higher activity of the LtxA in a fresh clinical strain compared to the original JP2 strain [19]. Still, this higher activity of the protein did correlate to a higher amount of LtxA detected on the cell pellet of the clinical strain analyzed by Western blotting [19]. Diaz and co-workers also reported that the LtxA from the fresh clinical *A. actinomycetemcomitans* strain of JP2 genotype was located on the cell pellet in larger amounts than

in the supernatant [19]. However, the clinical strain was described with features comparable to a rough strain phenotype. Kachlany and co-workers studied the difference between rough and smooth strains of *A. actinomycetemcomitans*, and they found that the smooth strains secreted more LtxA into the growth supernatant than the rough strains [23]. In order to accommodate the risk of differences between the strains due to rough-smooth phenotype in the present study, all strains were converted to smooth phenotypes, and the procedures of each method were adjusted according to this. Also in the study by Åberg and coworkers, analyzing the leukotoxicity of the Ghanaian *A. actinomycetemcomitans*, serotype b, smooth strain variants were used [20]. Still, the LtxA was purified from the cell pellet and not from the growth supernatant. However, when using the Peptone-Yeast-Glucose (PYG) medium (as done in the study by Åberg and co-workers) LtxA stays attached to the cell membrane and is not secreted into the growth medium [10]. The LtxA was extracted by NaCl in the study by Åberg et al. 2014; a protocol that has been criticized by others [19]. In addition, the use of different culture broth could also interfere with the LtxA production. However, whether or not these methodological aspects may explain the lack of correlation between the leukotoxicity and the LtxA expression and production among the Ghanaian *A. actinomycetemcomitans*, serotype b, found in this study, has to be a matter of further investigation in the future.

Environmental factors are of great importance and have been reported to influence on both the production and the secretion of the LtxA [18,22–25]. Previous studies have reported that the secretion of LtxA occurs into the environment both as a water-soluble protein and as an attached protein to membrane vesicles [26–28]. The LtxA attached to the vesicle is expected to be found in the supernatant used in the ELISA and the Western blot; however, this is not the case for the cell lysis assay used by Åberg and coworkers [20]. Although, the amount of cell membrane-attached LtxA and the vesicle membrane-attached LtxA should be relative [29], Kato et al. showed a 5-fold greater leukotoxic activity in the LtxA purified from the vesicles compared to the cell membrane-attached LtxA [29]. Therefore, the aspect of membrane vesicles may contribute to the surprising results in the present study as well.

The results from the present study show that the leukotoxicity of the Ghanaian *A. actinomycetemcomitans* strains is not explained by a greater leukotoxin expression and production. Johansson and coworkers have described different genetic similarities among the highly leukotoxic strains in the Ghanaian collection of *A. actinomycetemcomitans* [30]. The genetic similarity among the strains characterized by the *cagE* may be an indicator of a higher toxicity of the strains, but perhaps not an indicator of a higher leukotoxicity. The protocol used for purification of the LtxA in the study by Åberg and coworkers [20] may lead to presence of other protein and membrane components in the supernatant added to the THP-1 cells used in the cell lysis assay to measure the leukotoxicity of the *A. actinomycetemcomitans* strains [31]. Therefore, it is possible that the activity of the Ghanaian *A. actinomycetemcomitans* strains towards THP-1 cells is influenced by other components, either by interference directly with the THP-1 cells or by interference with the activity of the LtxA. It would be interesting to reproduce the leukotoxicity of the Ghanaian *A. actinomycetemcomitans* strains with other LtxA-purification protocols, where the LtxA is more strictly purified [11,19,32].

The leukotoxicity reported by Åberg and co-workers is given as a percentage of total lysis of the THP-1 cells by Triton X [20]. The results of the three LtxA-quantitative methods (Western blotting, ELISA and mRNA expression assay) used in the present study do not have a maximum value in the same way, and therefore it is possible for a strain to be characterized with a value above, e.g., 100%. If a certain amount of LtxA is needed for the total lysis of THP-1 cells [18], it is possible for strains with different amounts of produced and secreted LtxA to be classified with the same level of high leukotoxicity. However, they may still be very different according to LtxA expression and production. Therefore, some of the Ghanaian non-JP2 genotype strains of *A. actinomycetemcomitans,* serotype b, might produce enough LtxA to lead to a high percentage of cell lysis, but still not produce comparable amounts of LtxA to the JP2 genotype strains of *A. actinomycetemcomitans*. Of course, such presumptions depend on the fact that there is no difference in the activity of the LtxA. This is a notion that is unanswered for the bacterial collections used in the present study.

The results obtained by use of the ELISA and the mRNA expression assay did not correlate completely. Still, there is a reasonable relationship between the methods by visual inspection of Figures 3 and 5. Furthermore, both methods did determine the group of JP2 genotype of *A. actinomycetemcomitans*, serotype b, with a statistical significant higher leukotoxin expression and production than the group of non-JP2 genotype strains (Figure 4). The ELISA and the mRNA expression assay are both quantitative assays, and comparison of the results by a regression model seems reasonable [33,34]. Therefore, we performed a Bland-Altman plot for detection of difference between methods (not shown) [34], but this analysis showed no bias in our data; a result refuted by our scatter plot analysis. However, Ludbrook (2002) does discuss that the Bland-Altman analysis is not always safe to use when comparing methods that do not measure the exact same quantity and have a proportional relationship [33]. The results in the present study did show a reasonable relationship (Figure 5) when studying the scatter plot and the regression model, but with a higher agreement between the two methods among the JP2 genotype strains of *A. actinomycetemcomitans* than for the non-JP2 genotype strains. Generally, the data based on analysis of JP2 genotype strains seems more equal than for the non-JP2 genotype strains. This is probably reasonable considering that the group of JP2 genotype strains of *A. actinomycetemcomitans* is a very genetically homogeneous group of isolates, whereas the non-JP2 genotype strains, although all being of serotype b, is a more genetically diverse group of isolates. Still, based on the visual inspection of the Figures 3 and 5, it seems reasonable to assume that one can predict the LtxA production based on an analysis by use of a mRNA expression assay.

Some limitations of the present study should be addressed. Our mRNA analysis consists of quite high standard deviations, but our results from the mRNA expression assay are still consistent with previous findings by others [18,24,25]. In addition, the method is very similar to the one used by Longo et al. [25], but they reported no standard deviations, and, therefore, comparison is impossible. Whether or not the high standard deviation is caused by a biological phenomenon or by technical difficulties is unknown and needs to be investigated further. We are currently preparing for analysis of the *ltxA* gene expression by Nanostrings technologies; Nanostrings technologies are based on hybridization probes and require no amplification of RNA, thus, the inherent variabilities in the PCR technique will be avoided.

In the present study, we characterized 20 Ghanaian *A. actinomycetemcomitans*, serotype b, according to the LtxA expression and production by three different quantitative methods. The LtxA expression and production of the Ghanaian *A. actinomycetemcomitans*, serotype b, was not fully related with previously published results on leukotoxicity of the strains [20]. Therefore, these findings indicate that the non-JP2 genotype strains of *A. actinomycetemcomitans*, serotype b, originating from Ghana and previously reported on as highly leukotoxic, may have enhanced leukotoxicity, but probably not at comparable levels to JP2 genotype strains. Furthermore, the results from the present study showed a reasonable relationship between the expression of *ltxA* and the amount of produced and secreted LtxA. Conclusively, it is possible that only using one method to describe the LtxA-related virulence potential of *A. actinomycetemcomitans* could result in incomplete or misleading conclusions. Therefore, it is recommendable to use more than one method, combining analysis of both the leukotoxicity and the LtxA expression and/or production, when characterizing the leukotoxic potential of *A. actinomycetemcomitans*.

#### **4. Materials and Methods**

#### *4.1. Bacterial Strains and Cell Lysis Assay*

Table 1 shows the characteristics of 45 *A. actinomycetemcomitans* strains of serotype b, originally collected as subgingival plaque samples. The highly leukotoxic JP2 genotype of *A. actinomycetemcomitans*, characterized by a 530-bp deletion in the LtxA gene operon, belongs to serotype b, and is in particular linked to individuals of African descent [20,35,36]. Out of the 45 strains studied, 20 isolates from Ghanaian adolescents were previously analysed in a cell lysis assay based on the determination of lactate dehydrogenase (LDH) release from the monocytic leukemia cell line, THP-1 [20] as an

expression of the leukotoxicity. Of these 20 *A. actinomycetemcomitans* strains of serotype b, 15 were non-JP2 genotype strains, and, among these strains, five were characterized as low leukotoxic, five as intermediate leukotoxic, and five as highly leukotoxic (0–30% LDH release = low leukotoxicity, 31–60% = average leukotoxicity, and 60% ≤ = high leukotoxicity, respectively) according to the study by Åberg and coworkers [20] (Table 1). The remaining five Ghanaian strains from that study belonged to the JP2 genotype of *A. actinomycetemcomitans* and were all highly leukotoxic towards THP-1 cells (Table 1).

The results obtained by the three quantitative assays being Western blotting, ELISA, and mRNA expression assay in the present study were compared with cell lysis assay results obtained in a previous study reported by Åberg and coworkers [20]. In addition to the 20 strains from Ghana, the bacterial collection in the present study was expanded with another 25 strains collected mainly from individuals originating from Africa or with African origin for comparison reasons. These 25 *A. actinomycetemcomitans* strains were 17 JP2 and five non-JP2 genotypes of *A. actinomycetemcomitans*, serotype b, previously characterized by Haubek et al. [36], and three non-JP2 genotype of *A. actinomycetemcomitans*, serotype b, from a collection of Moroccan isolates not previously reported on. The complete collection of 45 *A. actinomycetemcomitans* strains, serotype b, was analysed by ELISA and in an mRNA expression assay for comparison of the two different methods determining leukotoxic production and expression of *A. actinomycetemcomitans*, respectively (Table 1).

All strains were transformed into smooth variants by repeated subculture before being analysed by three different biochemical methods.

#### *4.2. LtxA Isolation from the Cell Pellet and the Growth Supernatant*

For isolation of LtxA, 50 ml volumes of pre-warmed TY×2 (Tryptone-yeast) medium were inoculated with a few colonies from a chocolate agar plate and incubated for 24 h in an atmosphere of 5% CO2 in air to an OD600, 1 cm at approximately 0.3. Tubes were centrifuged at 3000× *g* for 20 min, and 1 mL of the supernatant was transferred to Eppendorf tubes, supplied with sodium azide to 3 mM, and stored at −20 ◦C for later quantification of LtxA by ELISA and Western blotting. The cell pellet was isolated and five ml of 10 mM phosphate-buffered saline (PBS, 0.3 M NaCl) pH 7.2, containing 3 mM sodium azide, was added to the bacterial pellet with the purpose of releasing the cell membrane-attached LtxA as described by Johansson and co-workers [10]. The tube was vortexed and rotated end-over-end for 30 min at room temperature. One ml of the suspension of the cell pellet in buffer was transferred to an Eppendorf tube and prepared for quantification of LtxA by Western blotting.

#### *4.3. Western Blotting*

Western blotting was performed on the 20 Ghanaian *A. actinomycetemcomitans* (Table 1) for semi-quantitative comparison of the level of LtxA between the different strains and for semi-quantitative comparison of the level of LtxA in the cell pellet extract and the growth supernatant. Cell pellet extracts for SDS-page were prepared from suspensions of the bacteria in five ml 0.3 M NaCl (as described above) and subsequently diluted 10-fold with SDS-containing sample buffer in order to reflect bacteria-bound LtxA in the 50 mL culture volume. Supernatant samples were prepared as described above. The Western blotting was performed as described by Reinholdt and co-workers [14] using a polyclonal rabbit anti-LtxA antibody (Ab-LtxA) (unlabelled for coating, labelled for detection) raised against the C-terminal half (recombinant) of the LtxA molecule in collaboration with the DAKO laboratories (Glostrup, Denmark). Briefly, samples of supernatant and cell pellet extract were applied to a 7% Tris-acetat buffered gel (Nu-PAGE™, Invitrogen) in identical volumes in neighbouring lanes, providing visual semi-quantitative comparison of LtxA.

#### *4.4. ELISA for Quantification of LtxA in Growth Supernatants*

Quantification of the production of LtxA was performed by an ELISA as described by Reinholdt and co-workers [14]. Briefly, the LtxA isolated in the growth supernatant treated with sodium azide to 3 mM was used to quantify the LtxA production of the different strains. To detect LtxA by ELISA, polystyrene microplates (Nunc, Roskilde, Denmark) were coated overnight with Ab-LtxA in 10 mM phosphate-buffered saline, pH 7.4 (PBS). After washing and blocking the plate with washing solution (PBS containing 0.25 M NaCl and 0.15%Tween 20), test samples appropriately diluted in washing solution were incubated in wells for 2 h. Bound LtxA was detected by sequential incubations with biotinylated Ab-LtxA and alkaline phosphatase-conjugated streptavidine (DAKO). The assay was developed with a chromogenic substrate of p-nitrophenylenephosphate in diethanolamine buffer, pH 9.0, and plates were read at 405 nm by a Multiscan RC reader (Labsystems). The analysis was performed in duplicates, and the original strain HK921 (JP2) served as a reference. The results are the mean of the two runs and are given as a percentage of the results when testing *A. actinomycetemcomitans* strain HK921 (JP2).

#### *4.5. mRNA Analysis*

The purification of RNA, the synthesis of cDNA, and the performing of the real-time PCR for the determination of the expression of mRNA, coding for the production of LtxA in each strain, were performed as described by Søndergaard and co-workers with a few modifications [37]. Briefly, bacterial suspensions were made with smooth colonies in 1.5 ml Brain Heart Infusion (BHI) broth to an OD600 at approximately 0.4. The suspensions were centrifuged, and the cell pellets were isolated. The cell pellets were re-suspended in one ml of RNAprotect (Qiagen, by GmbH, Hilden, Germany), and the RNA was purified with Magna Pure Compact instrument using a MagnaPure Compact Nucleic Acid Isolation kit (large volume) (Roche Diagnostics GmbH, Mannheim, Germany). Residual DNA was degraded from the RNA sample using a Turbo DNA free kit (Ambion by ThermoFischer Scientific, Waltham, Massachusetts, USA) and 50% more DNase than recommended by the manufacturer in a Veriti 96-well Thermal Cycler (Applied Biosystems by ThermoFischer Scientific, Waltham, Massachusetts, USA) for 30 min at 37 ◦C, and 5 min at 95 ◦C. cDNA was prepared with TaqMan Reverse Transcription reagents (Life Technologies) in a Veriti 96-well Thermo Cycler for 10 min at 25 ◦C, 30 min at 48 ◦C and 5 min at 95 ◦C. The complete digestion of the genomic DNA in each RNA sample was confirmed by running a cDNA reaction without reverse transcriptase in the cDNA reaction mix, followed by PCR. The cDNA was mixed with primers, probes, and TaqMan Fast Advanced Master Mix (Life Technologies by ThermoFischer Scientific, Waltham, Massachusetts, USA) and quantitative PCR was run in triplicates in a LightCycler 480 (Roche, Germany). Relative gene expression of *ltxA* was quantified with LightCycler Relative Quantification software (Roche Applied Science) and normalized to the single-copy housekeeping genes *adk* and *pgi*. The assay was performed in triplicates, and the results are given as the mean of the three runs. Primers and probes are listed in Table 2.



#### *4.6. Statistical Analysis*

All the statistics were analyzed by the use of SciPy [38] that is an open source scientific tool for Python® (Beaverton, USA). The statistical analysis were performed on log-transformed data in order to attain normal distributed data.

The difference between the group of JP2 and the group of non-JP2 genotype strains of *A. actinomycetemcomitans*, serotype b, determined for each method separately was analyzed by an unpaired sample t-test for parametric data.

A scatter plot for comparison of ELISA and the mRNA expression assay was performed as described by Ludbrook [33] with a few modifications. The relationship between the two methods was tested by use of a regression analysis. Because the ELISA quantifies the protein and the mRNA expression assay measures the mRNA coding for the protein, the two methods do not measure the exactly same quantity of the *A. actinomycetemcomitans* strains. Therefore, an ordinary least squares regression model was used, since the relationship between the two methods can be viewed as a calibration problem [33].

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2076-0817/8/4/211/s1, Figure S1. A semi-quantitative determination of the leukotoxin expression by Western blotting of 20 Ghanaian *A. actinomycetemcomitans*, serotype b, both JP2 and non-JP2 genotypes. The cell membrane-attached LtxA isolated from the cell pellet (C), and the released LtxA into the growth supernatant (S). HK921 (JP2) is illustrated as a reference. The LtxA is given as the band with a size of 116 kDa.

**Author Contributions:** Conceptualization, D.H., N.N.-L., A.J., and A.B.J.; methodology, J.R., M.L., N.N.-L., A.J., R.C. and A.B.J.; software, N.K.C. and A.B.J.; validation, D.H., N.N.-L., A.J. and A.B.J.; formal analysis, A.B.J. and N.K.C.; investigation, J.R., M.L., N.N.-L., D.H., A.J., R.C. and A.B.J.; resources, D.H., N.N.-L. and A.J.; data curation, D.H., A.J., N.N.-L., M.L. and A.B.J.; writing—original draft preparation, A.B.J.; writing—review and editing, A.B.J., D.H., N.N.-L., M.L., A.J., and R.C.; visualization, A.B.J.; supervision, D.H., A.J. and N.N.-L.; project administration, A.B.J., N.N.-L. and D.H.; funding acquisition, D.H. and A.J.

**Funding:** This research was funded by the University Strategic Fundings (USM-fundings) of Aarhus University, Danish Dental Association, the Ingeborg and Leo Danin Foundation, and by the Västerbotten County Foundation (TUA), Sweden.

**Acknowledgments:** We thank laboratory technician, Mette Nikolajsen for the preparation of bacterial isolates and growth media for the use in the ELISA and Western Blot analysis. We thank laboratory technician Mette Jensen for laboratory work carried out while performing the mRNA analysis. We thank Nikolaj Kruse Christensen (N.K.C) for a great contribution in the statistical analysis performance.

**Conflicts of Interest:** The authors declare to have no conflict 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* **Genetic Profiling of** *Aggregatibacter actinomycetemcomitans* **Serotype B Isolated from Periodontitis Patients Living in Sweden**

**Anders Johansson 1, Rolf Claesson 2, Carola Höglund Åberg 1, Dorte Haubek 3, Mark Lindholm 2, Sarah Jasim <sup>2</sup> and Jan Oscarsson 2,\***


Received: 28 August 2019; Accepted: 15 September 2019; Published: 17 September 2019

**Abstract:** The bacterium *Aggregatibacter actinomycetemcomitans* is associated with aggressive forms of periodontitis and with systemic diseases, such as endocarditis. By assessing a Ghanaian longitudinal adolescent cohort, we earlier recognized the *cagE* gene as a possible diagnostic marker for a subgroup of JP2 and non-JP2 genotype serotype b *A. actinomycetemcomitans* strains, associated with high leukotoxicity as determined in a semi-quantitative cell assay. This group of *A. actinomycetemcomitans* is associated with the progression of attachment loss. In the present work, we used conventional polymerase chain reaction (PCR) and quantitative PCR to perform the *cagE* genotyping of our collection of 116 selected serotype b *A. actinomycetemcomitans* strains, collected over a period of 15 years from periodontitis patients living in Sweden. The *A. actinomycetemcomitans* strains carrying *cagE* (referred to as *cagE*+; n = 49) were compared to the *cagE*-negative strains (n = 67), present at larger proportions in the subgingival plaque samples, and were also much more prevalent in the young (≤35 years) compared to in the old (>35 years) group of patients. Our present results underline the potential use of *cagE* genotyping in the risk assessment of the development of periodontal attachment loss in Swedish adolescents.

**Keywords:** *Aggregatibacter actinomycetemcomitans*; *cagE*; *virB1*; *virB4*; genotype; virulence

#### **1. Introduction**

*Aggregatibacter actinomycetemcomitans* is a Gram-negative opportunistic pathogen associated with rapidly progressing periodontitis and with extra-oral diseases, such as endocarditis [1–3]. Several longitudinal studies have demonstrated that adolescents colonized with *A. actinomycetemcomitans*, as compared to those that are not, have a significantly increased risk of the development of periodontal attachment loss (AL) [4–7]. *A. actinomycetemcomitans* produces an array of virulence factors that allow this bacterium to evade and suppress the host immune response, including two exotoxins, i.e., leukotoxin and cytolethal distending toxin (CDT) [8–10]. A large genetic diversity within the *A. actinomycetemcomitans* species has been found, and seven different serotypes (a–g) exist, representing genetically divergent lineages [11–13]. *A. actinomycetemcomitans* genotypes can have extensively different pathogenic potentials [5,14,15]. For example, carriers of the JP2 serotype b-specific genotype of *A. actinomycetemcomitans* are at higher risk of development of AL compared to carriers of a non-JP2 genotype of *A. actinomycetemcomitans*. Typical for JP2 genotype strains is the deletion of 530 base pairs (bp) in the promoter region of the *ltxCABD* gene operon, which encodes leukotoxin (LtxA), and an enhanced leukotoxicity [16,17]. LtxA is a virulence factor of *A. actinomycetemcomitans* with the

capacity to cause imbalance in the host inflammatory response [9]. The *ltx* promoter deletion has been frequently used as genetic marker to identify *A. actinomycetemcomitans* carriers with an increased risk for periodontal disease onset and progression [17], and this genotype is easily detected using a DNA-based assay (PCR) [18]. In addition to the JP2 genotype, a subgroup of non-JP2 genotype serotype b strains exhibits a similar disease association and high leukotoxicity, as has been shown in a semi-quantitative cell assay [14]. Genetic characterization has revealed that this particular subgroup of non-JP2 genotype of *A. actinomycetemcomitans* strains of serotype b are genetically closely related to the JP2 genotype by sharing the same arbitrarily-primed (AP) PCR gel electrophoresis banding pattern, referred to as AP-PCR genotype 1 in the present work [14]. Another property shared between the JP2 genotype and highly leukotoxic non-JP2 genotype serotype b strains was recently recognized, i.e., the carriage of the *cagE* gene sequence [19]. The *cagE* gene in *A. actinomycetemcomitans* was initially characterized by Teng and Hu [20], presenting evidence that the encoded CagE protein could induce apoptosis on primary human epithelial cells. However, consistent with leukotoxicity being a major virulence property of *cagE*-positive *A. actinomycetemcomitans* serotype b strains, one JP2 genotype bacterial cell was enough to lyse the majority of macrophage cells in vitro [20], whereas, as in comparison, a ratio of 50,000 JP2 genotype bacterial cells per epithelial cell was used to detect the CagE-induced apoptotic effects in vitro [21]. This suggests that CagE may have limited overall contribution to the virulence at biologically-relevant bacterial levels. In the present study, we utilized the *cagE* gene sequence as a diagnostic risk marker for the PCR detection of highly leukotoxic JP2 and non-JP2 genotypes of *A. actinomycetemcomitans* serotype b [19].

Furthermore, a type IV secretion system (T4SS) is a large macromolecular complex in Gram-negative bacteria which mediates conjugation, DNA transport and the secretion of virulence factors. Experimental work on model organisms, such as *Agrobacterium tumefaciens* and *Helicobacter pylori*, has revealed an archetypal T4SS system composed of 12 proteins, referred to as VirB1–VirB11, and VirD4 [22,23]. *A. actinomycetemcomitans* T4SS gene clusters are found in approximately 50% of strains and can be encoded both on the chromosome and on plasmids [24–26]. Interestingly, CagE exhibits homology to two T4SS proteins. The CagE N-terminus is homologous to VirB1 (lytic transglycosylase; also known as MagB01), and the CagE C-terminus is homologous to VirB4 (ATP:ase; MagB03) [19]. As judged by in silico analysis of the serotype b genomes available in the National Center for Biotechnology (NCBI) database, the *cagE* gene locus is not present in any of the strains encoding VirB1 and VirB4 on the chromosome. Whether the *cagE* and *virB1*/*virB4* genes are consistently inversely carried in serotype b strains has not earlier been thoroughly assessed but would support the notion that CagE may represent the result of a recombination event in which parts of the *virB1* and *virB4* genes were fused together to encode a chimeric VirB1–VirB4 protein in *A. actinomycemcomitans* [19].

In our previous work, delineating the role of *cagE* as a potential diagnostic marker, we studied a collection of *A. actinomycetemcomitans* strains collected from a prospective cohort of Ghanaian adolescents [19]. To further evaluate the role of *cagE* and *virB1*/*virB4* as diagnostic tools, we assessed our collection of *A. actinomycetemcomitans* strains that were collected during 15 years from periodontitis patients living in Sweden [27]. Data from microbiological analyses of this collection revealed that the young individuals (≤35 years) had a higher prevalence of *A. actinomycetemcomitans* and larger proportions of it in the samples compared to the older patients (>35 years). Moreover, serotype b was highly prevalent in the samples collected from young patients [27]. The aim of the present work was to determine the prevalence of the *cagE* genotype among the serotype b strains from this aforementioned collection (n = 116) and also to evaluate the potential use of *cagE* as a diagnostic marker for the carriage of highly leukotoxic serotype b strains among periodontitis patients living in Sweden.

#### **2. Results**

*2.1. Validation of PCR Assays to Detect VirB1 and VirB4 Sequences in A. actinomycetemcomitans Serotype B Reference Strains*

All A. *actinomycetemcomitans* serotype b strains assessed in the present study were grouped according to their AP-PCR genotype—1, 2 or "other" (i.e., AP-PCR types 3–11 as defined earlier [27]) (Figure 1A).

**Figure 1.** PCR genotyping of *Aggregatibacter actinomycetemcomitans* serotype b strains. (**A**) Distinct arbitrarily-primed (AP)-PCR banding patterns distinguish *cagE*-positive and *cagE*-negative serotype b strains of *A. actinomycetemcomitans*. The approximately 3000-bp DNA-band (arrowed) detected in AP-PCR type 1 is unique for this genotype and was earlier demonstrated to contain the *cagE* gene sequence [19]. Typically, this DNA band reflects the difference between AP-PCR types 1 and 2. The presence/absence of the *cagE* gene and *ltxA* JP2 promoter type in AP-PCR types 1 and 2 is indicated. (**B**) PCR detection of *cagE*, *virB1*, and *virB4*, respectively. An amplicon specific for *cagE* was revealed in both JP2 and non-JP2 AP-PCR genotype 1 strains. In AP-PCR genotype 2 strains, amplicons specific for *virB1* and *virB4* were detected, whereas *cagE* was not. Sizes (bp) of selected bands in the DNA molecular weight marker (M) are indicated. Figures illustrate representative experiments.

To test the hypothesis that the presence of chromosomal *virB1* and *virB4* genes can serve as genetic markers that are suitable for the detection of *cagE*-negative serotype b strains, PCR was employed as described in the Materials and Methods section. To evaluate the PCR approach, we initially assessed 25 *A. actinomycetemcomitans* strains of serotype b which have previously been subject to whole genome sequencing (Table 1) (Figure 1B). As expected, this revealed presence of both *virB1* and *virB4* in the *cagE*-negative strains only (n = 7; 4 type 2 AP-PCR and 3 "other" AP-PCR type), whereas neither *virB1* nor *virB4* were detected by PCR in the *cagE*-positive strains (n = 18; all AP-PCR type 1). This finding prompted us to further investigate this apparent inverse relationship between the carriage of *cagE* and *virB1*/*virB4* in the assessment of our local collection of serotype b *A. actinomycetemcomitans* strains. As *virB1* and *virB4* were carried simultaneously in the strains studied, we continued our analyses, mainly screening for the presence of *virB4*.


**Table 1.** Genotyping of *A. actinomycetemcomitans* serotype b strains (n = 25) that were earlier subjected to whole genome sequencing.

<sup>a</sup> Geographic location of laboratories from where strains were obtained/origin of donor (where known)**;** <sup>b</sup> determined by PCR as described in Materials and Methods; <sup>c</sup> previously determined by PCR [14,28,29]; <sup>d</sup> previously determined by PCR [19]; <sup>e</sup> previously determined by AP-PCR [19], or deduced in the present work; <sup>f</sup> Sweden residents unless specified otherwise.

*2.2. Screening of CagE and VirB4 in Serotype B A. actinomycetemcomitans Strains Collected from Patients with Periodontitis Living in Sweden*

We screened the 116 serotype b *A. actinomycetemcomitans* strains, collected from periodontitis patients living in Sweden, using qPCR to determine the prevalence of the *cagE* and *virB4* genes (Table 2) (Table S1) (Figure 2).

**Table 2.** Inverse relationship in the carriage of *cagE* and *virB4*. Presence of chromosomal *cagE* and *virB4* genes in serotype b strains of *A. actinomycetemcomitans* (n = 116) in different AP-PCR genotypes. The number of strains and percent (%) of all strains are indicated. The *cagE*-positive strains all (100%) belong to AP-PCR type 1 and lack the *virB4* gene.


**Figure 2.** Genotype patterns of *A. actinomycetemcomitans* serotype b strains. Schematic overview of AP-PCR type, as well as the JP2- and *virB4*-genotype patterns of the *cagE*-positive and *cagE*-negative strains, respectively. The collection of 116 serotype b strains was earlier sampled from periodontitis patients living in Sweden [27].

According to our results, *cagE* was present in 49 (42.2%) strains, including all 16 JP2 genotype strains, and hence absent in 67 (57.8%) strains. Of the *cagE*-positive strains, all (100%) belonged to AP-PCR genotype 1. Interestingly, three *cagE*-positive strains (all non-JP2 genotypes) were found to carry the *virB4* gene. However, PCR analysis, using the primers *magB01*-F and *ssb*-R, supported that all three strains most likely carried *virB4* on a plasmid rather than on the chromosome (data not shown). Thus, we concluded that a property common among the *cagE*-positive strains is an apparent lack of a chromosomal *virB4* gene. Of the *cagE*-negative *A. actinomycetemcomitans* strains, 25 (37.3%) belonged to AP-PCR genotype 2, and 42 (62.7%) belonged to AP-PCR genotypes 3–11. The prevalence of *virB4* was somewhat higher among the AP-PCR genotype 2 *A. actinomycetemcomitans* strains (44%) compared to the strains belonging to AP-PCR genotypes 3–11 (35.7%), suggesting that *virB4* might be usable as a genetic marker for a subgroup of *cagE*-negative strains. Thus, taken together, as none of the 116 strains studied encoded both *cagE* and *virB4* on the chromosome, we concluded that there is an apparent inverse relationship in the carriage of these genes in the *A. actinomycetemcomitans* strains of serotype b.

#### *2.3. Higher Proportions of CagE-Positive A. actinomycetemcomitans Serotype B in Subgingival Plaque Samples*

Furthermore, we assessed whether the *cagE* genotype may correlate with the proportion of *A. actinomycetemcomitans* found in the subgingival plaque samples. For this, the serotype b *A. actinomycetemcomitans* strains (n = 116) were divided into two groups, i.e., *cagE*-positive (n = 49) and *cagE*-negative (n = 67), and then they were matched with the determined total viable counts (%) of *A. actinomycetemcomitans* in the respective samples [27]. This clearly revealed that *cagE*-positive strains were carried in patients at significantly higher (*p* < 0.001) proportions than *cagE*-negative *A. actinomycetemcomitans* (Figure 3A). However, among the *cagE*-positive, the proportion of *A. actinomycetemcomitans* in samples with a JP2 genotype strain (n = 16) was not significantly different from that with a non-JP2 genotype strain (n = 33) (Figure 3B).

**Figure 3.** Higher proportions of *A. actinomycetemcomitans* in subgingival plaque samples containing a *cagE*-positive serotype b. The proportion of *A. actinomycetemcomitans* (total viable count—TVC; %) in the subgingival plaque samples was determined earlier for each of the serotype b strains (n = 116) [27]. (**A**) The *cagE*-positive strains (n = 49) were present in significantly higher (*p* < 0.001) proportions than the *cagE*-negative strains (n = 67). (**B**) The JP2 (n = 16) and non-JP2 (n = 33) strains were present at similar proportions. Median and quartiles from the samples are shown in each panel.

#### *2.4. Higher Prevalence of CagE-Positive A. actinomycetemcomitans Serotype B in Young Patients*

To further evaluate the virulence of the *cagE*-positive serotype b *A. actinomycetemcomitans* strains (n = 49) among periodontitis patients living in Sweden, we also assessed the age-associated prevalence of these strains. For this purpose, the patients (n = 116) were grouped into young (≤35 years; n = 62) and old (>35 years; n = 54) groups (Table 3). This revealed that among the young patients, *cagE*<sup>+</sup> *A. actinomycetemcomitans* strains (n = 40; 64.5%) were much more common than among the older patients (n = 9; 16.7%), i.e., these strains had a significantly higher (*p* < 0.001, odds ratio (OR) = 9.1, 95% CI: 3.8–22.0) prevalence among the young patients.

**Table 3.** Age-associated distribution of the *cagE* genotype of serotype b. The prevalence of *cagE*-positive and *cagE*-negative strains among the *A. actinomycetemcomitans* serotype b strains (n = 116) sampled from young (≤35 yr; n = 62) and from old patients (>35 yr; n = 54). The numbers and percentages (%) of young, old, and all patients are indicated. The prevalence of *cagE*-positive strains was significantly higher (*p* < 0.001; odds ratio (OR) = 10.5, 95% CI: 4.2–26.1) in the young compared to old patients.


#### **3. Discussion**

In the present work, we used conventional PCR and qPCR to genotypically analyze our collection of 116 *A. actinomycetemcomitans* serotype b strains, collected during 15 years from periodontitis patients living in Sweden, and our present results underline the potential use of *cagE* genotyping in the risk assessment of the development of periodontal attachment loss in adolescents living in Sweden.

Each of the 116 serotype b strains were matched both with its load (% of total viable count) in the respective subgingival plaque sample and with its age-associated prevalence category [27]. As *cagE*-positive, in contrast to *cagE*-negative serotype b *A. actinomycetemcomitans* strains, were found at larger proportions in the plaque samples and exhibited a much higher prevalence in the young compared to in the old patients of this population, our present results are consistent with our findings assessing the longitudinal Ghanaian adolescent cohort [19]. Whereas the proportions of *A. actinomycetemcomitans* genotypes in the total viable counts of subgingival plaque samples had not earlier been assessed in patient cohorts, the ratio of *cagE*-positive among serotype b strains carried by young patients in the Swedish population (64.5%) was similar to that of the adolescents in the Ghanaian cohort, exhibiting an association between the progression of attachment loss and exposure to this particular *cagE*-positive genotype [14,19]. As the *cagE*-positive serotype b strains sampled in both Ghana and Sweden were found at larger proportions in the plaque samples and exhibited a much higher prevalence in the young group of patients [19,27], the results from our present study are consistent with the notion that *cagE*-positive strains (including both the JP2 and non-JP2 genotypes) represent a subgroup of highly virulent *A. actinomycetemcomitans* serotype b.

The genetic similarity of *cagE*<sup>+</sup> serotype b strains is supported by the observation that they share the same AP-PCR genotype, as well as the fact that they have all a complete *cdtABC* gene operon [30]. We speculated earlier that they may belong to a clonal lineage that is closely related to the JP2 genotype ancestor [19]. As *cagE*-positive strains include both the JP2 and non-JP2 genotypes but no identified JP2-genotype strain has thus far been found to be *cagE*-negative, it is hypothesized that the JP2 genotype-associated deletion in the *ltxCABD* promoter once originated in a *cagE*-positive serotype b strain (Figure 4). It is tempting to speculate that high leukotoxicity may have been a characteristic of this ancestral *A. actinomycetemcomitans* strain, as that is a property common among *cagE*-positive strains, regardless of whether they are of the JP2 genotype or not.

**Figure 4.** Hypothetical origin of the *cagE* and JP2 genotypes in serotype b *A. actinomycetemcomitans*. The genetic similarity between *cagE*-positive strains and the apparent absence of the JP2 genotype among *cagE*-negative strains suggests the possibility that the JP2-associated 530-bp deletion in the *ltx* promoter might have originated in a *cagE*<sup>+</sup> strain. The JP2 genotype of *A. actinomycetemcomitans* initially appeared as a distinct genotype in the Mediterranean part of Africa approximately 2400 years ago [29]. As *cagE*<sup>+</sup> strains consistently lack chromosomal copies of *virB1* and *virB4*, an earlier recombination event causing fusion of a *virB1*- and a *virB4*-like gene sequence resulting in the *cagE* determinant might have taken place in a common serotype b *A. actinomycetemcomitans* ancestral strain.

Consistent with our earlier in silico analysis of the genome-sequenced serotype b strains in the NCBI database [19], another property shared between the *cagE*-positive strains assessed in the present work was a lack of chromosomal genes encoding the T4SS-associated proteins VirB1 and VirB4. Based on the homology between VirB1 and VirB4 with the CagE N-, and C-terminus, respectively, we suggested earlier that CagE may represent a fusion product of a VirB1- and a VirB4-like amino acid sequence [19]. A scenario where the origin of the *cagE*<sup>+</sup> serotype b *A. actinomycetemcomitans* strains is a recombination event on the chromosome, generating a fusion of parts of the genes encoding *virB1* and *virB4* (as illustrated in Figure 4), is plausible considering that chimeric proteins do exist in a number of bacterial T4SS gene clusters. For example, it was reported that *H. pylori* VirB3 and VirB4 is a fusion product, i.e., the first 150 amino acids of VirB4 have weak similarity with VirB3 although the motifs are conserved [31]. Similarly, a Western blot assay indicated a CagE-like protein pattern when prototypical *virB3* and *virB4* genes of *A. tumefaciens* were fused together and expressed [32]. Chimeric proteins are also encoded in a number of T4SS gene clusters of other species, including VirB3–VirB4 in *Campylobacter jejuni* [33,34], VirB1–VirB8 in *Bordetella pertussis* [35], and VirB11–VirD4 (MagB11–MagB12) in at least one strain of *A. actinomycetemcomitans* [25]. Moreover, observations with *H. pylori* are consistent with the notion that T4SS gene clusters can include regions that are prone to genetic rearrangements, resulting in the disruption or activation of the secretion system [36]. Results from our present work show that CagE and VirB1/VirB4 can be encoded in the same *A. actinomycetemcomitans* strain, albeit, as supported by PCR, with the T4SS genes most likely encoded on plasmids. The carriage of plasmids encoding T4SS genes, such as *virB1* and *virB4,* has been demonstrated in some *A. actinomycetemcomitans* strains [25,26]. In contrast, *cagE* appears not to be encoded on plasmids. According to the sequences available in the NCBI database, no hitherto sequenced *A. actinomycetemcomitans* plasmid carries a *cagE* gene locus. We were unable to detect by PCR the presence of a T4SS-encoding plasmid in the *cagE*-positive serotype b strain HK1651, which was earlier reported [25]. The reason for this discrepancy is not known but may reflect the possibility that this plasmid was lost in the strain preserved in our stocks upon repeated in vitro cultivation. The loss of plasmids of *A. actinomycetemcomitans* strains during in vitro cultivation is a phenomenon that has been reported earlier, albeit then related to growth in an antibiotic free medium [37].

Taken together, our present results further support the usefulness of the *cagE* gene as a potential diagnostic marker in the risk assessment of the development of attachment loss among young individuals. We conclude that *cagE* positive *A. actinomycetemcomitans* strains of serotype b among periodontitis patients living in Sweden consist of the JP2 and non-JP2 genotypes with phenotypic characteristics similar to the ones seen for the JP2 genotype strains but with a leukotoxin promoter region lacking the 530-bp deletion. Their origin, evolution, and extent of genetic similarity will be further explored by whole genome sequencing.

#### **4. Materials and Methods**

#### *4.1. Collection of A. actinomycetemcomitans Strains and Clinical Data Used in the Present Study*

For the present work, we used data from our microbiological analyses of 3459 subgingival plaque samples, collected from 1445 patients during 15 years (2000–2014) that included 337 'younger' patients (≤35 years of age) and 1108 'older' patients (>35 years of age) [27]. At the specialist clinics, it is recommended that microbial analysis is performed to study the microbial biofilm profiles of individuals ≤35 years affected by periodontal attachment loss, and of patients >35 years with rapidly progressive periodontitis, not responding to conventional periodontal therapy. The samples were sent from the Specialist Clinic of Periodontology at the Dental School in Umeå, Sweden, and from external specialist dental clinics throughout Sweden to be analyzed at the laboratory for microbiological diagnostics, Dental School, Umeå. The samples were collected from individuals between 9 and 92 years of age that were all diagnosed with periodontitis and referred to specialist clinics for periodontal treatment. However, due to the many clinics involved and the retrospective nature of the present study,

clinical and other parameters were not systematically reported in the patient information attached to the referral to the laboratory for microbiological diagnostics. Therefore, the classification of the patients was dichotomized only and was based on the old definition of early onset periodontitis, which distinguished patients ≤35 years versus those >35 years of age [38]. An *A. actinomycetemcomitans* strain was collected and isolated from 347 patients [27]. PCR characterization revealed that 118 (34.0%) of the *A. actinomycetemcomitans* strains were serotype b, and 17 (14.4% of the serotype b strains) were characterized by 530-bp deletion in the promoter region of the leukotoxin gene operon (JP2 genotype). Among these 118 serotype b strains, we were able to cultivate and characterize 116 for use in the present study: 100 non-JP2 genotype and 16 JP2 genotype. For the present work, each of these 116 unique *A. actinomycetemcomitans* strains was combined with recorded clinical data, i.e., the age group of the patient (>35 or ≤35 years), and proportion of *A. actinomycetemcomitans* of the total cultivable microflora (TVC) in the sample.

#### *4.2. Bacterial Strains and Growth Conditions*

In the present work, we used a collection of 116 unique *A. actinomycetemcomitans* serotype b strains that were collected from periodontitis patients living in Sweden [27]. A list of these strains is presented in Table S1. The sampling of this collection and the subsequent characterization of the serotype, the AP-PCR genotype, and the leukotoxin promoter type (JP2/non-JP2 genotype) has been described earlier [27]. In the present study, 25 serotype b *A. actinomycetemcomitans* strains were used as reference, as they were subject to prior whole genome sequencing [15,19,39–41] (Table 1). Among these, five belong to a collection of oral *A. actinomycetemcomitans* strains previously reported on by Prof. Sirkka Asikainen: ANH9381, I23C, S23A, SCC1398 and SCC4092. Nine strains belong to the collection of serotype b *A. actinomycetemcomitans* strains, sampled from periodontitis patients living in Sweden: 133A1-08U, 196A1-10U, 115A-11U, 245-12U, 338A1-13U, 304A1-14U, 299A1-15U, 456A1-13U, and 520A-01U [28]. *A. actinomycetemcomitans* strains 443G, 486G, 575G, 605G, and 638G were sampled from a Ghanaian cohort of adolescents [6,42]. Finally, six type strains were included in the study: HK908 [29], HK909 [43], HK912 [29], HK921 [43], HK1651 [39], and Y4 [44,45]. All strains were cultured on blood agar plates (5% defibrinated horse blood, 5 mg of hemin/l, 10 mg of vitamin K/l, Columbia agar base) and incubated in air supplemented with 5% CO2 at 37 ◦C.

#### *4.3. DNA Isolation and Polymerase Chain Reaction Analysis*

DNA templates for PCR and qPCR analysis were obtained by boiling a loopful of fresh *A. actinomycetemcomitans* colonies in 100 μl of water. *A. actinomycetemcomitans* genomic DNA to be used in AP-PCR was isolated using the GenElute™ Bacterial Genomic DNA kit (Sigma-Aldrich, St. Louis, MO, USA), following the manufacturer's instructions. For the isolation of plasmids from *A. actinomycetemcomitans* strains, a QIAprep®Spin miniprep kit was used (QiaGen, Venlo, The Netherlands). Reaction mixtures for PCR were prepared using illustra™ PuReTaq™ Ready-To-Go™ PCR beads (GF Healthcare, Buckinghamshire, UK), whereas we used a KAPA SYBR®FAST qPCR Kit (KAPA Biosystems, Wilmington, MA, USA) for qPCR. The AP-PCR type was analyzed as earlier described [19,27], using the random sequence oligonucleotide OPB-3 (5 -AGTCAGCCAC-3 ) (Invitrogen, Carlsbad, CA, USA) at 0.4 μmol/l and cycling conditions according to Dogan and coworkers [46]. The *cagE* gene was amplified by PCR as a 1020-bp DNA fragment, using a *cagE* forward primer (5'-GGATCCGTCCCTGAAATTTTATTAGCTTG-3') and a *cagE* reverse primer (5-CTGCAGTTAAACGACCTTTAAACATTTTTTTA-3') [20]. In qPCR analysis, *cagE* was detected as earlier described [19] using the *cagE*\_F2 (5'-TGGATTGGGACAAGTGAACA-3') and *cagE*\_R2 (5'-CAATAATGGCTCGTGCAATATC-3') primers to amplify a 623-bp internal fragment of the *cagE* gene. A ≈630-bp fragment of the lytic transglycosylase, *virB1* gene was amplified using a *cagE* forward primer and a *virB1* reverse primer (5'-GTTTTTAATCAATCTTCCTGATTG-3'). The amplification of the ATP:ase-encoding *virb4* gene, as a ≈900-bp DNA fragment, was carried out by PCR or qPCR using the *virB4* forward primer (5'-GTGCAGAAGCCTGTATTCGTGC-3'), and the *virB4* reverse primer

(5'-CCAGTCATTAGTGGCTTCGCC-3'). The *magB01*-F (5'-GCCATCTACTACGCCTATCGC-3') and *ssb*-R (5'-TTATCGCCGTCAAGCGGAAG-3') primers [25] were used in PCR to assess the presence of plasmids encoding T4SS genes. PCR cycling conditions were 94 ◦C for 1 min, followed by 35 cycles of 94 ◦C for 30 sec, 54 ◦C for 30 sec, and 72 ◦C for 1 min, and then finally 72 ◦C for 7 min. The cycling conditions for qPCR were 95 ◦C for 10 min, followed by 45 cycles of 95 ◦C for 10 sec, 54 ◦C for 5 sec, and 72 ◦C for 22 sec. The complete genome sequences of serotype b strains SCC1398 (VirB1; GenBank accession KND83482), and I23C (VirB4; KOE53154) [40] were used as reference in oligonucleotide synthesis.

#### *4.4. Statistical Analysis and Image Processing*

The rank test was used to calculate the strength of the association between the *A. actinomycetemcomitans cagE* and JP2 genotypes and proportion of TVC in subgingival plaque samples (IBM SPSS Statistics for Windows, Version 25.0, Armonk, New York). An odds ratio (OR) was used to quantify the strength of the association between the *A. actinomycetemcomitans cagE* genotype and age group (MedCalc for Windows, MedCalc Software, Ostend, Belgium). No normalization of the data or test unit was used in the present work.

#### *4.5. Ethical Considerations*

All procedures were conducted according to the guidelines of the local ethics committee at the Medical Faculty of Umeå University, which are in compliance with the Declaration of Helsinki (64th WMA General Assembly, Fortaleza, October 2013). The characterization of the *A. actinomycetemcomitans* strains was made utilizing clinical samples from patients visiting the Specialist Clinic of Periodontology at the Dental School in Umeå. Data from specific strains were grouped in relation to age (>35 or ≤ 35 years) and could not be traced to a specific individual.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2076-0817/8/3/153/s1. Table S1. The 116 *A. actinomycetemcomitans* serotype b strains, collected from periodontitis patients living in Sweden, and which were used in the present work.

**Author Contributions:** Conceptualization, A.J., R.C., and J.O.; methodology, R.C., and J.O.; validation, R.C., M.L., S.J., and J.O; formal Analysis, A.J., R.C., S.J, and J.O.; investigation, R.C., M.L., S.J., and J.O.; resources A.J., R.C., C.H.Å., D.H., and J.O; data curation, A.J., R.C., and J.O.; writing—original draft preparation, A.J., R.C., and J.O.; writing—review and editing, A.J., R.C., C.H.Å., D.H., M.L., S.J., and J.O.; visualization, A.J., R.C., and J.O.; supervision, A.J., R.C., C.H.Å., and J.O.; project administration, A.J. and J.O; funding acquisition, A.J., M.L. and J.O.

**Acknowledgments:** We are grateful to Elisabeth Granström for valuable technical assistance. This work was supported by TUA grants from the County Council of Västerbotten, Sweden (to J.O. and A.J.), by funds from Insamlingsstiftelsen, Medical Faculty, Umeå University (to J.O. and A.J.), and from Svenska Tandläkare-sällskapet, Kempe Foundation, and Thuréus Foundation (to M.L.).

**Conflicts of Interest:** The authors declare no competing interests.

#### **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/).

### *Obituary* **In Memoriam: Edward "Ned" Lally**

#### **Nataliya Balashova**

Department of Basic and Translational Sciences, School of Dental Medicine, University of Pennsylvania, Philadelphia, PA 19104, USA; natbal@upenn.edu

Received: 23 February 2020; Accepted: 27 February 2020; Published: 1 March 2020

On February 11, 2019, welost a colleague and friend Dr. Edward "Ned" Lally. Hewas aninternationally recognized leader in research on the periodontal pathogen *Aggregatibacter actinomycetemcomitans*.

Ned succeeded in his professional career as a clinical pathologist, scientist and educator. He received his B.S. degree in 1966 and his D.M.D. degree in 1968 from the University of Pittsburgh. He was a resident in both the United States Naval Hospital and the Department of Pathology at the Hospital of the University of Pennsylvania. Ned received his Certificate in Oral Pathology in 1973 and earned a Ph.D. in Immunology in 1978 from the University of Pennsylvania. He then joined the Department of Pathology at the University of Pennsylvania, School of Dental Medicine, where he had grown to the rank of Professor with tenure. In addition to his academic career, Ned distinguished himself in service to the United States Navy and Naval Reserve.

In early years, Ned's research was focused on understanding mucosal immunity. In his practice he met young patients with an aggressive form of periodontal disease and investigated familial aggregation of the disease. Ned had a longstanding interest in the field of microbial pathogenesis and the role of *A. actinomycetemcomitans* in the development of aggressive periodontitis. He was especially intrigued to unravel the mechanism of action of an RTX toxin, LtxA, an immune cell killer. In the 1980s, Ned made his seminal observation that LtxA interacted with β2 integrin receptor LFA-1 on the surface of immune cells. He believed this was the most significant of his contributions related to the field. Following his study, β2 integrin receptors were reported for RTX toxins from other bacteria. In later years, Ned successfully employed various immunological, biophysical, biochemical and molecular techniques for the characterization of the toxin interaction with the host cell membrane. Ned found that the interaction of LtxA with immune cells is both complex and multifaceted, involving both β2 integrin LFA-1 and cell membrane cholesterol. His results provide new insight into the mechanism by which the RTX group of bacterial toxins kill cells. Ned published over 90 peer-reviewed manuscripts and successfully maintained continuous NIH funding.

Ned will be remembered not just for his scientific contributions, but also as a great charismatic teacher for young researchers. Outside his professional career, Ned was a devoted husband, father and grandfather. Personally, I will always remember him as one of the best friends I have ever met in my professional life.

© 2020 by the author. 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/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Pathogens* Editorial Office E-mail: pathogens@mdpi.com www.mdpi.com/journal/pathogens

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18

www.mdpi.com