Candidate probiotics comprised Lactobacillus rhamnosus Goldin and Gorbach (GG) (ATCC 53103), Lactobacillus reuteri ATCC 55730, and Streptococcus salivarius K-12. The periodontal pathogens were Fusobacterium nucleatum ATCC 10953, Porphyromonas gingivalis ATCC 33277, and Aggregatibacter actinomycetemcomitans ATCC 33384. Aggregatibacter actinomycetemcomitans was grown in Tryptic Soya Agar and broth supplemented with 0.6% yeast extract (TSA and TSB) and incubated in a 5% CO₂ atmosphere. All other bacteria were grown using Wilkins-Chalgren broth or agar (Oxoid, Basingstoke, United Kingdom) at 37°C and incubated in an anaerobic cabinet (atmosphere, 10:10:80, H₂, CO₂, N₂). All bacteria were incubated at 37°C. For both in vitro and in vivo experiments, 10 ml volumes of each bacterium were prepared by growing cells overnight either anaerobically or in a 5% CO₂ environment. A 100-fold dilution of overnight culture was made in sterile broth, which was incubated until the desired growth phase was reached according to constructed growth curves. Cells were harvested by centrifugation at 3220 × g for 15 min and the pelleted cells were washed and re-suspended in sterile phosphate buffer saline (PBS) (0.01 M PBS; NaCl 8 g/L, KCL 0.2 g/L, Na₂HPO₄ 1.42 g/L, and KH₂PO₄ 0.24 g/L, pH 7.4). These steps were repeated twice, and the cell suspension adjusted to a final optical density at 600 nm of 0.1. Each larva received aliquots of 5 μl of this bacterial suspension injected directly to the hemocoel. For experiments using lysates and cell-free extracts, 10 ml of 10⁸ Colony Forming Units (CFU)/ml of the appropriate strain was centrifuged. The supernatant was reserved to use as a cell-free extract. The pellet was washed, concentrated in 1 ml of Wilkins Chalgren broth, and lysed using a bead beater (FastPrep FP120; Thermo Electron Corporation). Samples were filter sterilized to remove any whole bacteria remaining.

G. mellonella were obtained from Live Foods Direct, Sheffield, United Kingdom. The larvae were in the last instar stage (shedding of the exoskeleton), and were selected based on their weight (275–300 mg), the presence of a fresh cream color, and no gray markings. All larvae were used within 3 days from shipment.

The test organisms were grown to the stationary phase (10⁸ CFU/ml). Cultures of pathogenic bacteria were diluted 1:100 in Wilkins Chalgren agar. After careful mixing, 20 ml agar plates were poured and left to set. Once set, cup cuts were aseptically made within the agar (8 mm wells) and filled with 100 μl of the probiotic cell culture (∼1.5 × 10⁸ CFU/ml) of each tested probiotic organism. The plates were incubated at 37°C anaerobically for 48 h and the diameter of the zone of inhibition produced measured using calipers.

Human oral keratinocytes (HOKs, Sciencell Research Laboratories, United States) were used to assess the effect of probiotics on their viability. HOKs were maintained in oral keratinocyte medium (OKM, Sciencell Research Laboratories, United States) supplemented with oral keratinocyte growth supplement (OKGS) and 100 U/ml of both penicillin and streptomycin (OKM, Sciencell Research Laboratories, United States). The medium was substituted twice weekly and cells were incubated in a humid atmosphere with 5% CO₂ at 37°C. Cells were cultured in T-25 or T-75 vented culture flasks and 24 well plates (Corning, Sigma, United States). The cells were plated, at a density of 10 × 10⁴ cells per cm², in 1 ml of the appropriate medium either in 12 or 24 well plates according to the experiment and used after 24 h incubation at 37°C at ∼90–100% confluence. Cells were exposed to 10⁸ CFU/ml of each probiotic cell suspension for 24 h. Viability was determined using the trypan blue exclusion assay (Prince et al., 2012). Uninfected cells were included as a control. Probiotic lysates and cell-free extracts (100 μl) were added simultaneously with the periodontopathogen to the human oral keratinocytes.

A modified version of the assay described by Ramarao et al. (2012) was performed. Larvae of G. mellonella were incubated for 30 min at room temperature before injection. Overnight cultures of each microorganism were centrifuged (3220 × g, 15 min) and suspended in PBS. This was repeated twice. Cultures were adjusted to an OD_{600 nm} of 0.1. For intrahemocoelic injection, bacterial suspensions were prepared with final concentrations in the range of 10⁴ CFU/ml to 10⁸ CFU/ml. Volumes of 5 μl of each strain, cell-free extract or lysate were delivered directly to the hemocoel through an injection in the rear left pro-leg using a 26-gauge needle Hamilton microsyringe (Sigma, United Kingdom). Sterile PBS (5 μl) was injected into the “trauma” control group and additionally, a “no treatment” control group was added. The right pro-leg was used as the injection site. Different sites were used for pathogenic and probiotic strains to reduce the risk of injection site infection. Infected larvae were incubated in a petri dish in groups of 10 at 37°C in the dark for the duration of the experiment (5–7 days).

Larval mortality was determined daily over a week. Larvae that had turned black and that were not moving in response to a gentle shaking of the dish or touching with a pipette tip were considered dead. Dead larvae were removed from the petri dish and the death was recorded. The experimental endpoint was designated by either the death of all the larvae in the tested groups or the conversion of larvae into pupae. Pupae were identified via a color change to white (Jorjao et al., 2018). Five Petri dishes containing 10 larvae each were assigned to each experiment and control groups (50 larvae total for each sample). Dead G. mellonella were placed into sterile Universal bottles and homogenized in 10 ml of sterile PBS. This suspension was then serially diluted, and spot plated onto Wilkins Chalgren agar to calculate bacterial load per individual larva. The experiments were terminated once two of the control individuals had died or pupated.

Galleria mellonella data were plotted as survival curves using the Kaplan–Meier estimator in Microsoft Excel 2010. The survival values were considered significantly different if the p-value was < 0.05. For cell culture work paired student’s T-tests done using Microsoft Excel 2010. Results were considered significant if p ≤ 0.05.

In vitro testing of antibacterial activity using an agar well-diffusion assay indicated that all three tested pathogens were inhibited by the three investigated probiotics to varying degrees. A. actinomycetemcomitans showed the greatest sensitivity to all probiotics with significantly larger zones of inhibition produced than for other pathogens. Streptococcus salivarius K-12 was the most effective against A. actinomycetemcomitans (Table 1). For both F. nucleatum and P. gingivalis the effect of all probiotics was comparable with P. gingivalis exhibiting slightly more sensitivity (Table 1).

Data in Figures 1–3 show when control HOKs were incubated for 24 h, ∼97% ± 0.2 of the cells remained viable; whereas the percentage of cells that remained viable following 24 h inoculation with the periodontopathogens was significantly (p > 0.01) lower. P. gingivalis decreased the viability of treated cell monolayers to 40% (Figure 1), A. actinomycetemcomitans decreased the viability of cell monolayers to 51% (Figure 2), and F. nucleatum decreased the viability of treated cell monolayers to 34% (Figure 3).

In the G. mellonella model (Figure 4), no mortality was observed in either control (non-treated control or PBS control). All three pathogens caused the death of at least 50% of larvae by the experimental endpoint. However, P. gingivalis and F. nucleatum caused significantly higher mortality (p < 0.05) than A. actinomycetemcomitans in the order F. nucleatum > P. gingivalis > A. actinomycetemcomitans (Figure 4). Bacterial lysates and broth culture filtrates of pathogens reduced the viability of larvae, but the mortality rate was less than with live bacteria. Lysates were more lethal than cell-free extracts. These data show that HOKs and G. mellonella are susceptible to infection with selected periodontal pathogens. By contrast, none of the candidate probiotics (L. rhamnosus GG, L. reuteri, and S. salivarius K-12) induced significant mortality in HOKs or G. mellonella (Figures 1–4).

When S. salivarius K-12 was injected with either F. nucleatum or A. actinomycetemcomitans there was a higher larval viability 24 h post-injection (p ≤ 0.01) than when G. mellonella was injected with the pathogens alone (Figure 5). However, S. salivarius K-12 did not protect larvae from the effects of P. gingivalis (Figure 5). L. reuteri afforded some protection against P. gingivalis but had a limited effect against F. nucleatum and A. actinomycetemcomitans. L. rhamnosus GG had some protective effect on G. mellonella viability when injected in a mixture with F. nucleatum and P. gingivalis but increased the mortality of larvae when injected in a mixture with A. actinomycetemcomitans (Figure 5). Probiotic protection in G. mellonella according to bacterial species was in the following order: for F. nucleatum SS > LR > LHR, for P. gingivalis LR > LHR > SS, and for A. actinomycetemcomitans, SS > LHR = LR. Administering probiotic strains, their supernatants and lysates simultaneously with F. nucleatum (Figure 3) or P. gingivalis (Figure 1) to human oral keratinocytes significantly (p < 0.05) increased the viability compared to when infected with the pathogen alone. Protection against A. actinomycetemcomitans inoculation in HOKs (Figure 2) was variable and was only statistically significant (p < 0.05) when lysates or supernatants were administered (p < 0.05).

The effects of injecting pathogens and probiotics into separate prolegs were evaluated. This was to exclude any effects due to an inhibitory interaction between microorganisms. The viability of larvae injected with F. nucleatum was increased when injected simultaneously with all probiotics. The most effective was S. salivarius K-12 (Figure 6). For P. gingivalis, L. rhamnosus GG reduced the lethal effect of G. mellonella (Figure 6), while with the other two probiotics the viability was lower than the larvae injected with P. gingivalis alone. L. reuteri decreased conferred protection against A. actinomycetemcomitans, as did L. rhamnosus GG and S. salivarius K-12 to a lesser extent (Figure 6). Probiotic protection according to bacterial species occurred in the following order: SS > LR > LHR for F. nucleatum, LR > LHR > SS for P. gingivalis, and LR > LHR = SS for A. actinomycetemcomitans.

None of the larvae were killed when F. nucleatum was injected into larvae that were pre-injected with L. rhamnosus GG or S. salivarius K-12 (Figure 7). Furthermore, there was a highly significant decrease (p < 0.001) in the mortality of larvae pre-treated with L. reuteri (Figure 7). The effects of injecting P. gingivalis on larvae that were pre-treated for 24 h with probiotics was total protection for larvae pre-treated with L. reuteri whereas the least protection was conferred by L. rhamnosus GG (Figure 7). For A. actinomycetemcomitans the best protection was conferred by S. salivarius K-12 followed by L. reuteri and L. rhamnosus GG was the least effective (Figure 7) in terms of the increase of viability of larvae compared to larvae treated by pathogen only (p < 0.01). Probiotic protection according to bacterial species was in the following order: LHR > SS > LR for F. nucleatum, LR > SS > LHR for P. gingivalis, and SS > LR > LHR for A. actinomycetemcomitans.