**3. Results**

#### *3.1. Antibacterial Activity of Proanthocyanidins from Pelargonium sidoides Root Extract*

For antibacterial efficiency evaluation, bacterial strains *S. aureus*, *S. epidermidis*, *A. actinomycetemcomitans* and *E. coli* were subjected to PSRE and PACN treatments were used at the concentration ranging from 10 to 100 μg/mL. Antibacterial activity results are summarized in Figure 1.

Both PSRE and PACN significantly reduced bacterial metabolic activity in comparison to the untreated control, starting from concentrations of 50–70 μg/mL. However, some differences were noticed between the tested strains. In the case of *S. aureus*, PSRE was more effective than PACN (Figure 1a). Fifty μg/mL of PSRE was enough to significantly decreased metabolic activity of this strain, whereas PACN had a similar result at 80 μg/mL (*p* < 0.05 vs. control, indicated by the \*). For *A. actinomycetemcomitans*, the results were opposite (Figure 1d); PACN significantly decreased metabolic activity at 50 μg/mL, but it required 80 μg/mL PSRE to achieve this level of activity. The metabolism of the clinical isolate *S. epidermidis* (Figure 1b) was significantly decreased by both PSRE and PACN applied at a 70 μg/mL concentration.

Interestingly, the non-pathogen *E. coli* demonstrated the highest resistance to the treatment (Figure 1c) when the 0–80 μg/mL window was considered. After the 80 μg/mL PSRE treatment, metabolic activity of *E. coli* decreased by 26% compared with untreated control, whereas metabolic activity of *S. aureus, S. epidermidis* and *A. actinomycetemcomitans* was reduced by 75%, 97% and 57%, accordingly. In the presence of 80 μg/mL PACN, the loss of metabolic activity of these four strains listed in the same order was 47%, 74%, 96% and 99%. Moreover, even after addition of 100 μg/mL PACN to the growth medium, *E. coli* still preserved nearly half of the control activity level, but in the case of *A. actinomycetemcomitans*, the twice less amount of this preparation almost completely blocked bacterial growth. Summarizing, the results of the antibacterial evaluation indicate stronger toxicity of PSRE and PACN to *S. aureus*, *S. epidermidis* and *A. actinomycetemcomitans* than to non-pathogenic *E. coli* strain when the extracts were used within the 80 μg/mL range. Conversely, at higher 90 and 100 μg/mL concentrations a similar broad range effect was observed for all the tested strains as all results of both PSRE and PACN were significantly different in comparison with untreated controls (*p* < 0.05, indicated by the \*).

**Figure 1.** Antibacterial activity of *Pelargonium sidoides* root extract (PSRE) and proanthocyanidins from PSRE (PACN) towards *Staphylococcus aureus* (**a**), *S. epidermidis* (**b**), *Escherichia coli* (**c**) and *Aggregatibacter actinomycetemcomitans* (**d**). RFU—relative fluorescence units. Bars represent means and standard deviations of six experimental replicates. \* = *p* < 0.05 vs. untreated control.

*3.2. The E*ff*ect of Proanthocyanidins from Pelargonium sidoides Root Extract on Gingival Fibroblast Viability under Conditions of Bacterial Infection*

3.2.1. *Pelargonium sidoides* Root Extract and Proanthocyanidins Preserve Gingival Fibroblasts in the Presence of Bacteria

Next in the study, we investigated whether antibacterial properties of PSRE and PACN are efficient in protecting human gingival fibroblasts during bacterial infection. For this purpose, two microenvironments were simulated: oral, by growing together gingival fibroblast cells and bacteria *A. actinomycetemcomitans*, and mucosal, by co-culturing gingival fibroblasts and *S. aureus*. Results obtained by the co-culture "race for the surface" simulation are reported in Figure 2.

Both the presence of PSRE and PACN at 100 μg/mL were effective in protecting cells from bacterial infection; the number of viable cells counted after 24 h of direct contact with bacteria was comparable with the control consisting of cells cultivated in fresh medium without bacterial presence. Conversely, when bacteria were cultivated in the same wells of cells but in the absence of PSRE or PACN (no treatment), they were able to fully colonize the well. In this scenario, no viable cells were detected for both applied models. The results indicate that the presence of either of the substances was effective in preserving gingival fibroblast cell viability by lowering bacteria proliferation.

**Figure 2.** The effect of *Pelargonium sidoides* root extract (PSRE) and proanthocyanidins from PSRE (PACN) on human gingival fibroblast viability in co-culture "race for the surface" assay. Human gingival fibroblasts were infected with *A. actinomycetemcomitans* (**a**) or *S. aureus* (**b**); bars represent means and standard deviations of six experimental repeats. "Control" represents not infected cells, and "n.d." means there were no detectable cells in the samples. In (**c**), there are representative images of human gingival fibroblasts with *A. actinomycetemcomitans* after 24 h of infection.

3.2.2. *Pelargonium sidoides* Root Extract and Proanthocyanidins Protect Gingival Fibroblasts from Necrotic Cell Death Induced by Bacterial Lipopolysaccharide

The protective effect of PSRE and PACN under conditions of bacterial infection might be mediated not only by suppression of bacterial growth, but also by increasing the resistance of the cells to bacterial metabolites. Therefore, we examined how PSRE and PACN affect the viability of primary murine gingival fibroblasts in the presence of bacterial LPS. To select the concentrations of PSRE and PACN that has no harmful effect for the cells, the concentration-dependent toxicity was defined by double nuclear staining with propidium iodide and Hoechst33342.

The level of necrotic cells with propidium iodide-positive nuclei significantly increased after 24 h treatment with 200 μg/mL and higher concentration of both PSRE and PACN (Figure S1). PSRE demonstrated significantly higher toxicity compared to PACN in the concentration range between 200 and 350 μg/mL, and this was reflected by established LD50 for the preparations: 209 μg/mL for PSRE and 288 μg/mL for PACN. In the concentration range from 400 to 550 μg/mL, there were no further difference between the effect of both preparations and nearly all cells in the culture were found necrotic. Based on these data, there were two concentrations of PSRE and PACN selected for antiinflammatory effect study: 50 and 100 μg/mL. Both solutions were not toxic to gingival fibroblasts when applied for 24 h at these concentrations.

After treatment with 2 μg/mL LPS, the amount of necrotic cells with propidium iodide-positive nuclei in fibroblast cell culture increased from 2.7% ± 2.2% in control to 31.7% ± 11.9%, reflecting a statistically significant loss in cell viability (Figure 3). However, the treatments with 50 and 100 μg/mL PSRE and PACN prevented cell viability loss induced by LPS. The data indicate that both PSRE and PACN could protect gingival fibroblasts from bacterial LPS-induced necrosis.

**Figure 3.** The effect of *Pelargonium sidoides*root extract (PSRE) and proanthocyanidins from PSRE (PACN) on the viability of gingival fibroblasts affected by bacterial lipopolysaccharide (LPS). (**a**) Representative fluorescent images after cell treatment with LPS and 100 μg/mL PSRE or PACN and nuclear viability staining; propidium iodide-positive necrotic nuclei are red, all nuclei are stained blue with Hoechst33342. (**b**) Quantitative viability results. Fifty and 100 represent the concentrations (μg/mL). The data are presented as means plus standard deviation of five experiments. \*—significant difference compared to the untreated control, and #—significant difference compared to the LPS only treatment, *p* < 0.05.

3.2.3. *Pelargonium sidoides* Root Extract and Proanthocyanidin Fraction Prevent Caspase Activity Induced by Bacterial Lipopolysaccharide

Next to necrosis, bacterial toxicity might initiate apoptosis that also contributes to the loss of gingival tissue. Cysteine aspartic proteases (caspases) are key mediator enzymes in apoptosis [23]. There are two main groups of caspases according to their place in the apoptotic event cascade: initiator caspases acting as apoptotic triggers and effector caspases that are amplified by triggers and execute proteolysis leading to the characteristic biochemical and morphological changes in the apoptotic cell. To determine apoptosis-preventing capacity of PSRE and PACN, we assessed the level of active initiator caspase-8 and effector caspase-3 in LPS-treated fibroblast cells.

A 24 h LPS treatment induced elevation in caspase-8 substrate cleavage activity; it increased from 0.007 ± 0.001 nmol/min/mg in control samples to 0.047 ± 0.013 nmol/min/mg (Figure 4a). PSRE at a 50 μg/mL concentration did not significantly affect LPS-induced caspase-8 activity. However, treatments with 100 μg/mL PSRE, 50 μg/mL PACN and 100 μg/mL PACN significantly prevented caspase-8 activation by LPS by 2.2, 2.0 and 5.9 times, respectively. Treatments with 100 μg/mL PSRE and PACN without LPS had no significant effect on the activity of the caspase.

After treatment with LPS, effector caspase-3 substrate cleavage rate increased from 0.012 ± 0.01 to 0.189 ± 0.06 nmol/min/mg (Figure 4b). Both PSRE and PACN at all tested concentrations significantly reduced the effect of LPS on caspase-3 activity. When LPS was applied together with 50 μg/mL PSRE, caspase-3 activity was nearly two times lower than in LPS only-treated samples. 100 μg/mL PSRE decreased the effect of LPS on caspase-3 activity 4.7 times. Fifty and 100 μg/mL PACN lowered LPS-induced caspase-3 activity 3 and 8.8 times, respectively. There was no significant effect on caspase-3 substrate cleavage rate observed after treatment with 100 μg/mL PSRE or PACN alone, without LPS.

Overall, caspase activity evaluation indicates that both PSRE and PACN could suppress apoptotic protease activity evoked by bacterial LPS. Both preparations have a stronger effect in decreasing the executing caspase-3 activity compared with the effect on trigger caspase-8. However, PACN was more efficient in suppressing caspase-8 activity than PSRE, thus, PACN is a more powerful suppressor of apoptosis at the early stages. None of the substances was toxic to gingival fibroblasts when applied for 24 h at these concentrations.

**Figure 4.** The effect of *Pelargonium sidoides* root extract (PSRE) and proanthocyanidins from PSRE (PACN) on LPS-induced caspase-8 (**a**) and caspase-3 (**b**) activation in gingival fibroblasts. Fifty and 100 represent the concentrations expressed in μg/mL. The data are presented as means with standard deviation of seven experimental repeats. \*—significant difference compared to untreated control, #—significant difference compared to LPS only treatment and &—significant difference compared to LPS plus 50 μg/mL PSRE treatment, *p* < 0.05.

*3.3. The E*ff*ect of Proanthocyanidins from Pelargonium sidoides Root Extract on Inflammatory Responses to Bacterial Lipopolysaccharide*

3.3.1. The effect of *Pelargonium sidoides* Root Extract and Proanthocyanidin Fraction on Lipopolysaccharide-Induced Secretion of Inflammatory Mediators

Infection-induced inflammation is the main responsible for gingival and dental tissue loss in the pathogenesis of periodontal disease. Thus, for successful treatment of the disease it is important to defeat both infection and inflammation. Next in the study, we examined antiinflammatory properties of PSRE and PACN on LPS-induced release of inflammatory mediators from gingival fibroblasts and blood leukocytes.

Increased IL-8 production by gingival fibroblasts is responsible for attraction of neutrophils to inflamed regions and rapid tissue loss during periodontitis [24,25]. Evaluation of IL-8 amounts secreted in the medium by cultured gingival fibroblasts revealed that after 24 h with LPS the amount of the cytokine increased from nearly zero to 1022 ± 75 ng/mL (Figure 5a). In the presence of 50 μg/mL PSRE, the level of IL-8 after same LPS stimulation was only 344 ± 49 ng/mL, i.e., three times lower than without the extract. Increasing PSRE concentration to 100 μg/mL further suppressed IL-8 release to 201 ± 33 ng/mL (five times less than after LPS-only treatment). A similar suppression level was achieved by 50 μg/mL PACN, and in the presence of 100 μg/mL PACN, the level of IL-8 after LPS stimulation had further dropped to 32 ± 12 ng/mL, making the level 16 times lower than after treatment with LPS alone.

The level of PGE2, a mediator of cyclooxygenase-2 inflammatory pathway, was found to be dramatically increased in the cell culture medium after LPS treatment (Figure 5b). In control samples, the average amount of PGE2 was 5 ± 2 ng/mL, but after 24 h with LPS, it had climbed up to 2372 ± 194 ng/mL. In the presence of both 50 and 100 μg/mL PSRE, the levels of PGE2 after LPS treatment were 637 ± 58 ng/mL and 613 ± 133 ng/mL, respectively, i.e., nearly four times lower compared with the level without the extract. Treatments with 50 and 100 μg/mL PACN were even more efficient, further decreasing PGE2 level in the medium to 174 ± 41 ng/mL and 72 ± 16 ng/mL, or 14 and 33 times compared with LPS only treatment, respectively. Thus, the effects of PACN on LPS-stimulated PGE2 release were significantly stronger than those of PSRE.

Bacterial invasion also cause an infiltration of leukocytes that mediate inflammation and disturb osteoblast-osteoclast balance via release of IL-6 [23]. Thus, we investigated how PSRE and PACN affect the release of IL-6 from PBMCs. One hundred μg/mL of PSRE and PACN significantly decreased LPS-induced secretion of IL-6 from PBMCs to 67% and 18% of the level caused by LPS stimulation, respectively (Figure 5c). Note that neither PSRE nor PACN were toxic to the cells at the concentrations applied as revealed by metabolic viability analysis (Figure S2).

**Figure 5.** The effect of *Pelargonium sidoides* root extract (PSRE) and proanthocyanidins from PSRE (PACN) on (**a**) interleukin-8 (IL-8) and (**b**) prostaglandin E2 (PGE2) secretion from gingival fibroblasts, and (**c**) interleukin-6 (IL-6) secretion from peripheral blood mononuclear cells after LPS treatment. Fifty and 100 represent the concentrations (μg/mL). The data are presented as means and standard deviations of seven experiments. \*—significant difference compared to untreated control, #—significant difference compared to LPS only treatment and &—significant difference compared to LPS plus a 50 μg/mL PSRE treatment, *p* < 0.05.

The data about inflammatory mediator secretion indicate that both PSRE and PACN efficiently suppress LPS-induced IL-8 and PGE2 release from gingival fibroblasts and IL-6 release from mononuclear leukocytes. PACN had slightly stronger IL-8 and IL-6 release suppressing activity, and significantly stronger PGE2 release suppressing activity than PSRE.

#### 3.3.2. The Effect of *Pelargonium sidoides* Root Extract and Proanthocyanidin Fraction on Lipopolysaccharide-Induced Expression of Inflammation-Related Genes

The release of inflammatory factors is the first step of innate immune response to pathogens. The next step leading to prolonged and enhanced inflammatory reaction is induction of inflammatory genes to produce new mediators. If uncontrolled at this stage, acute inflammation may become chronic and contribute to tissue loss in periodontitis. Activated macrophages and leukocytes are the main source of interleukin-1β (IL-1β) and TNF-<sup>α</sup>, the acute phase pyrogenic cytokines involved in most of the processes that maintain inflammation [26]. Bacterial infection also triggers inducible NO synthases (iNOS) to produce reactive nitrogen species stress on pathogens [27]. Therefore, we have examined the capacity of PSRE and PACN to modulate expression of inflammatory genes IL-1β, TNF-α and iNOS in primary murine bone marrow-derived macrophages and human mononuclear leukocytes under treatment of LPS. Stimulation of macrophages in the presence of interferon-γ (IFN-γ) acting as enhancer of LPS-induced gene expression [28] induced an increase in transcription of all the three genes investigated (Figure 6a–c).

Both preparations at a dose of 100 μg/mL significantly suppressed the mRNA transcription of IL-1β and iNOS (Figure 6a,b). The level of the IL-1β mRNA decreased by 78% of the initial level with LPS after treatment with PSRE, and by 89%—after treatment with PACN. For iNOS, the decrease in mRNA level after PSRE and PACN treatment was 53% and 64%, respectively. However, the incubation with both substances did not affect LPS plus IFN-γ-induced TNF-α gene expression (Figure 6c). 6 h treatment with LPS caused significant increase in cyclooxigenase-2 (COX-2), TNF-α and IL-1β gene transcription in human PBMCs (Figure 6d–f). PSRE and PACN at a concentration of 100 μg/mL significantly suppressed mRNA transcription of COX-2 and IL-1β. When LPS was together with PSRE, COX-2 and IL-1β mRNA levels dropped by 50%, 73% and 56%, respectively. For PACN, mRNA synthesis for these cytokines was suppressed by 63%, 89% and 76%. Similarly to BMDMs case, neither PSRE, nor PACN significantly affected TNF-α gene expression. Both PSRE and PACN at a concentration of 100 μg/mL were not toxic for PBMCs and BMDMs as revealed by metabolic activity and apoptosis evaluation (Supplementary Figures S2 and S3).

**Figure 6.** The effect of *Pelargonium sidoides*root extract (PSRE) and proanthocyanidins from PSRE (PACN) on proinflammatory gene expression in bone marrow-derived macrophages (**<sup>a</sup>**–**<sup>c</sup>**) and peripheral blood mononuclear cells (**d**–**f**) after LPS or LPS and IFN- stimulation. (**<sup>a</sup>**,**d**) IL-1β (**<sup>a</sup>**,**b**) iNOS, (**<sup>c</sup>**,**f**) TNF- and (**e**) COX-2. The data are expressed as a fold change of glucose-6-phosphate isomerase gene transcription and presented as mean ± SD of three independent measurements. \*—significantly different from the LPS-treated samples (ANOVA followed by a Tukey's multiple comparison test, *p* < 0.05).

The gene expression analysis indicate that both preparations acted as inflammatory signal suppressors preventing expression of pro-inflammatory cytokine IL-1β and prostaglandin producing enzyme COX-2 genes, as well as decreasing synthesis of iNOS and protecting tissues from the damage of reactive nitrogen species. However, neither PSRE nor PACN significantly influenced TNF-α gene expression.

#### 3.3.3. The Effect of *Pelargonium sidoides* Root Extract and Proanthocyanidin Fraction on Lipopolysaccharide-Induced Macrophage Conversion to M1 Phenotype

Activated macrophages can be polarized into a proinflammatory M1 phenotype and alternative anti-inflammatory M2 phenotype [29]. Increase in the M1/M2 macrophage ratio can lead from an antibacterial defense to the development of periodontitis and positively correlates with the severity of the disease [30]. Next in the study, we tested how PSRE and PACN affect LPS and IFN-γ-stimulated macrophage polarization to proinflammatory M1 phenotype characterized by the presentation of surface markers CD80 and CD86 [31,32].

Flow cytometry analysis revealed that in response to LPS and IFN-γ, the amount of M1-polarised macrophages increased 9.3 times compared to the untreated control (Figure 7). Both PSRE and PACN at a concentration of 100 μg/mL were effective in reducing the level of CD80 and CD86-positive cells. The population of cells with the exposed markers after treatment with PSRE was by 58% lower, and after treatment with PACN by 71% lower than after LPS and IFN-γ stimulation without the

treatments. The results indicate that both substances were potent in preventing macrophage conversion to proinflammatory M1 phenotype under exposure to LPS and IFN-γ treatment.

**Figure 7.** Expression of proinflammatory cell surface markers CD80 and CD86 analyzed by flow cytometry 24 h after treating LPS + IFNγ-activated BMDMs with PSRE and PACN. (**a**) Characteristic mouse macrophage marker F4/80-positive cells were gated for double CD80 and CD86 analysis as a measure of M1 macrophage phenotype (top right small quadrant). Representative plots of a total of three independent experiments in three replicates are presented in the bottom. (**b**) Mean ± SD of three independent measurements in three parallels. Differences between the measurements were tested using a one-way ANOVA followed by a Tukey's multiple comparison test. \*—significantly different from the LPS and IFN-γ treatment (*p* < 0.05).
