Background
The human oral microbiome is composed of a wide variety of microorganisms that play important roles in health and disease. Bacteria, for example, maintain oral and systemic health [
1], but can also cause disease. Bacteriophages shape microbial diversity [
2]. Protozoans and fungi consume food debris and other microbes [
3,
4] and archaea utilize fermentation byproducts to produce methane [
5]. Yet, in terms of microbial species, our current understanding of the human oral microbiome is limited to a few well-studied microorganisms. Based on what is known about these microbes, researchers have developed conceptual models with the explicit purpose of determining the etiology of oral diseases (e.g., [
6,
7]). These models suggest putative interactions among the microorganisms as well as between microorganisms and the host. Although these studies have significantly advanced the field, recent studies show that the oral cavity is a complex and dynamic habitat consisting of hundreds of different interacting species [
8,
9].
The conceptual model for periodontitis is that
Porphyromonas gingivalis, Tannerella forsythia, Treponema denticola and
Aggregatibacter actinomycetemcomitans are responsible for the disease [
6,
10,
11]. Evidence from animal models indicated that these bacteria may disrupt tissue homeostasis by manipulating signaling pathways of the host and once the innate immunity is compromised, cause a shift in the relative abundance of microbes, resulting in inflammation and bone loss [
12,
13]. Yet, it is well established that these bacteria i.e.,
P. gingivalis, T. forsythia, T. denticola, A. actinomycetemcomitans, are also found in healthy oral cavities (e.g., [
14‐
19]). Moreover,
P. gingivalis is not found in up to half of the patients with chronic or aggressive periodontitis [
20]. The paradox of these microbes being present in both health and disease questions the validity of the conceptual underpinnings of the etiology of periodontitis, which invites for alternative ways of thinking about periodontitis as well as other oral diseases.
Conventional periodontal therapies aim at controlling supra- and subgingival biofilms and managing prognostic factors such as poor glycemic control in subjects with diabetes mellitus and active smoking [
21,
22]. Although periodontal therapy has been shown to improve periodontal health and reduce the rate of further clinical attachment loss [
23,
24] and tooth loss due to periodontitis [
25‐
27], it falls short of completely arresting the disease. Almost all subjects compliant with periodontal maintenance care continue to show signs of progressive clinical attachment loss, and one- to two- thirds of them, lose one or more teeth during an extended period of periodontal maintenance care [
28,
29]. We believe new thinking is necessary for finding successful ways of treating periodontitis.
The knowledge from a different field indicates that an appropriate microbial community is a key for maintaining resistance against infection. A classic example is the gut microbiome, which protects the host from
Clostridium difficile (CD) infection. Specifically, some patients develop CD colonization of their gut upon antibiotic treatment, which eradicates out their innate microbiome. A successful strategy for combating CD by transplanting gut microbiota from a healthy donor was first medically recorded more than 50 years ago [
30]. (The origins of the procedure date back to the 4
th century in China [
31]). Specifically, a suspension of stool taken from a systemically healthy intimate partner, relative or friend is introduced into the gastrointestinal tract of the recipient via colonoscopy, enema or a nasogastric tube. Recent systematic review reports approx 90 % efficiency of eradicating the CD infection by a fecal transplant [
32]. Besides the CD infection, other potentially dysbiotic diseases were found to be responsive to the microbial transplantation, such as inflammatory bowel disease, obesity. For a review on current developments in the gut microbial transplant see Kelly et al. [
33].
Inspired by the success of the fecal transplant, here we put forward a concept of a microbial transplant as a potential therapy for periodontitis. We envision the therapy of consisting of three steps: (i) harvesting sub- and supra-gingival microbiota from a healthy donor, e.g., spouse or a partner; (ii), performing deep cleaning, root planning and applying a broad-spectrum antimicrobial agent to the periodontitis patient; and (iii) neutralizing the antimicrobial agent immediately following by a rinsing with a microbial suspension harvested from the healthy donor in the periodontitis patient.
The objectives of the present study were two-fold. The first objective was to assess if the entire oral microbial community of periodontitis was different from established caries, edentulism, and oral health. The second objective was to test an approach for applying a broad-spectrum safe antimicrobial agent (NaOCl) to significantly reduce the load of existing microbiome followed by neutralization of this agent for subsequent microbial transplantation.
Methods
Study subjects
Adult subjects were recruited at the Department of Periodontics, University of Washington, Seattle, WA, U.S.A. and Section of Periodontics, University of Düsseldorf, Germany. A written informed consent for participation in the study was obtained from the subjects. The study was approved by the University of Washington Institutional Review Board, ref. number 3570. Subjects were enrolled if they had one of the following clinical conditions: severe periodontitis, caries, edentulism, or oral health. A periodontitis case was defined as having at least 2 interproximal sites at different teeth with clinical attachment loss (CAL) of 6 mm or greater and at least 1 interproximal site with probing depth (PD) of 5 mm or greater [
34] and a minimum of 20 permanent teeth, not including 3
rd molars. Subjects were excluded from the periodontitis group if they had multiple established caries lesion or wore a removable partial denture. A caries case was defined as having the following number of teeth with established caries lesions: 6 or more teeth in subjects 20 to 34 years of age; 4 or more teeth in subjects 35 to 49 years of age; and 3 or more teeth in subjects 50 years of age and older. Established caries was defined as a class 4 lesion according to the International Caries Detection and Assessment System. The number of teeth with caries lesion in caries cases was greater than one standard deviation above the mean of caries extent in respective age group the U.S.A. [
35]. Exclusion criteria for a caries case were interproximal sites with CAL of 4 mm or greater or PD of 5 mm or greater [
34]. An edentulous case had to be completely edentulous in both jaws and their teeth had to be extracted more than one year before the enrollment in the study. A healthy case was defined as having 28 teeth, not counting 3
rd molars, or 24 or more teeth, not counting 3
rd molars if premolars had been extracted for orthodontic reasons or were congenitally missing with no signs of oral disease. Exclusion criteria for a healthy case included: smoking, loss of permanent teeth due to caries or periodontitis, any interproximal sites with CAL of 4 or greater or PD of 5 mm or greater, or any established caries lesions. Exclusioncriteria for all groups included: oral mucosal lesions, systemic diseases, and use of antibiotics or local antiseptics within 3 months prior to the study.
Sample collection
For all but the edentulous patients, supra- and subgingival plaque was collected from six sites with the deepest probing depth in each sextant. One sterile paper point per site was inserted into the deepest aspect of the periodontal pocket or gingival sulcus. Biofilm from oral mucosae was collected by swiping a sterile cotton swab over the epithelial surfaces of the lip, left and right buccal mucosae, palate, and dorsum of the tongue [
36]. Samples were stored at −80 °C.
Molecular methods
Microbial DNA was isolated from cells by physical and chemical disruption using zirconia/silica beads and phenol-chloroform extraction in a FastPrep-24 bead beater [
37]. Prokaryotic 16S rRNA genes were amplified using universal primers (27 F and 1392R) using the GemTaq kit from MGQuest (Cat# EP012). The PCR program involved a pre-amplification step of 10 cycles with annealing temperature of 56 °C followed by 20 amplification cycles with annealing temperature 58 °C. In each cycle, elongation time was 1 min 10s, at 72 °C. PCR was finalized by extended elongation for 5 min. PCR products were purified with DNA Clean & Concentrator columns (Zymo Research, USA) and quantified using the NanoDrop (Agilent, USA).
Due to technical reasons, for some subjects, subgingival and mucosal microbiotas were pooled together into one vial, while for other subjects the subgingival and mucosal microbiotas were stored separately. Since our goal was to investigate the entire microbial community, for the separately stored supra and subgingival microbiota samples, equal quantities of PCR product derived from swab and paper point samples were pooled together for each patient. For edentulous patients, there were no paper point samples. Each purified PCR product, 500 ng, was labeled with a Multiplex Identifier (MID) during the Roche Rapid Library preparation step. Four MID-tagged sequences, representing each of the conditions, were combined in equimolar concentrations and subjected to emPCR and DNA sequencing protocols as specified by the manufacturer’s recommendations for the Roche 454 Jr. instrument.
Data analysis
The obtained sequences were separated out by their respective Multiplex Identifier (MID) and uploaded to the MG-RAST web server [
38]. The MG-RAST pipeline assessed the quality of sequences, removed short sequences (multiplication of standard deviation of length cutoff of 2.0) and removed sequences with ambiguous bp (non-ACGT; maximum allowed number of ambiguous base pair was set to 5). The pipeline annotated the sequences and allowed the integration of the data with previous metagenomic and genomic samples. The RDP database was used as annotation source, with minimum sequence identity of 97 %, maximum e-value cutoff at 10
−5, and minimum sequence length of 100 bases. Alpha diversity analysis was conducted in MG-RAST.
Orthogonal transformation of the annotated rRNA genes to their principal components (PC) was conducted using normalized abundances [
39,
40]. Normalization of the abundance was performed identically to the procedure used by MG-RAST. Specifically, abundances were increased by one, log2 transformed, and centered to produce relative values. Relative values were standardized by dividing them by the standard deviation of the log2 values [
38]. The data were graphed on a 2 dimensional ordination plot. To determine the relative contribution of the microbial species to the plot, we let X denote the 16 by 578 (patients by species) matrix of the normalized abundance values. The matrix X was used to produce a 16 × 16 matrix D of distances between all pairs of subjects. Principal coordinate analysis (PCoA, i.e., multidimensional scaling) was achieved by performing principal component analysis (PCA) on the matrix of distances, D. The Euclidean distance metric was used to create this matrix, but this metric produced results similar to those when the alternative Bray-Curtis distance was used. To investigate and visualize differences between the four patient groups, the first two principal components, PC1 and PC2, of the distance matrix D were retained. To establish which species were most prominently responsible for the groupings, the projection of each species onto the (PC1, PC2) plane was calculated; those species with the largest projections are displayed in the right panel of Fig.
3. That is, this figure shows a biplot of the species most highly correlated PC1 and PC2.
Mann–Whitney test (alpha = 0.05) was used to investigate significant difference in normalized relative abundances. The abundance data were normalized in two different ways: MG-RAST normalization as described above and raw abundances normalized to the total number of reads in a sample. Differences in alpha diversities by condition were determined using ANOVA. Two-tailed T-tests, assuming unequal variance, were used investigate if there was significantly different in the means (alpha = 0.05). These analyses were performed using SAS JMP.
Sodium hypochlorite experiments
According to previously suggested concentration [
41], sodium hypochlorite experiments were conducted with a 1:50 dilution of the 6 % household bleach (Chlorox, The Clorox Company, USA), the “hypochlorite working solution”. The dilution corresponds to 16 mM NaOCl. The neutralizing agent was a sodium ascorbate buffer produced by adjusting pH of a 23 mM ascorbic acid solution with concentrated NaOH until pH 5.3.
Three samples of dental plaque from a volunteer was subjected to three challenges: (i) resuspension in saline, (ii) resuspension in the hypochlorite working solution and (iii) resuspension in the hypochlorite working solution that was previously neutralized with equal volume of the ascorbate buffer. Resuspended plaque samples were plated on blood agar plates.
Active chlorine concentration was crudely assessed by iodine colorimetry. A colorimetric solution contained 1 volume of 0.1 % stabilized starch solution (Fisher Scientific) mixed with 0.1 volume of 1 M potassium iodide. Two volumes of the colorimetric solution were mixed with 1 volume of the solution containing NaOCl. Intense blue color indicated detectable active chlorine. Iodine colorimetry was calibrated with a dilution series of the hypochlorite working solution with the following dilutions 1:200, 1:400, 1:800, 1:1600, 0. The dilution1:800 still showed a hint of blue, while 1:1600 was completely colorless.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
AEP, TB, TFF, PAN designed the experiments; AP, PAN conducted the experiments; BGL, TWR conducted statistical analysis; AP, PAN wrote the paper. All authors read and approved the final manuscript.