Assessment of cytotoxicity exerted by leaf extracts from plants of the genus Rhododendron towards epidermal keratinocytes and intestine epithelial cells

Fresh leaf material of reliably identified Rhododendron species was used in this study (Table 1). The material was collected from January 2012 to December 2013 from plants grown in the Rhododendron-Park Bremen (www.​rhododendronpark​bremen.​de). Each plant species was sampled once without considering seasonal variations. The identities of the plant species used in this study (Table 1) have been verified by reference to the German Gene Bank Rhododendron Database provided by the Bundessortenamt (www.​bundessortenamt.​de/​rhodo) [25]. Material from all plant species used is publicly and freely available from the Rhododendron-Park Bremen upon request.

Leaf material was frozen in liquid nitrogen and powdered using a KSW 3307 mill (Clatronic, Kempen, Germany). Crude extracts were prepared by soaking two grams of Rhododendron leaf powder in 10 mL of 80 % methanol for 24 h at 4 °C with constant shaking. Insoluble material was removed by centrifugation at 3,220 g for 30 min at 4 °C, and supernatants were stored at -20 °C for further use. Methanol was evaporated from the extracts using a Micro Modulyo lyophilizer (Edwards, Crawley, UK). Stock solutions were prepared by dissolving the residues in 100 % dimethyl sulfoxide (DMSO) (Carl Roth, Karlsruhe, Germany). Prior to the in vitro assays, the samples were mixed with the respective cell culture medium such that the final concentration of DMSO did not exceed 0.5 % (v/v), and 5, 50, or 500 μg of lyophilized powder per mL culture medium were applied to confluent IEC6 and HaCaT cell monolayers.

The normal rat small intestine epithelial cell line IEC6 [29, 30] and the human keratinocyte cell line HaCaT [31, 32], purchased from the European Collection of Cell Cultures (Salisbury, UK), were used throughout this study. IEC6 cells were grown in Dulbecco’s modified Eagle’s Medium (DMEM High Glucose) (Lonza Group, Basel, Switzerland) supplemented with 10 % fetal calf serum (FCS) (Perbio Science, Bonn, Germany) and 10 μg/mL insulin (Sigma-Aldrich, Steinheim, Germany). IEC6 cells were incubated at 37 °C in a 5 % CO₂ atmosphere in an incubator (Heraeus, Osterode, Germany). HaCaT cells were cultured in DMEM containing 10 % FCS and incubated at 37 °C in an 8.4 % CO₂ atmosphere. Cell cultures were passaged once per week. All experiments were performed with cultures at approx. 70 % and 95 % confluence for IEC6 and HaCaT cells, respectively.

Effects of Rhododendron leaf extracts on the viability and proliferative activity of cultured IEC6 and HaCaT cells were quantitated using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay (Carl Roth). This test is indicative for mitochondrial NADH-dependent dehydrogenase activity, which is proportional to both cell viability and proliferation rates of treated cultures [33‐35]. A total of 1 × 10⁴ cells/well were seeded in single wells of 96-well plates (Greiner, Essen, Germany) and upon reaching the desired confluence, the cells were incubated with three different concentrations of Rhododendron leaf extracts (5, 50, and 500 μg per mL culture medium, not exceeding 0.5 % DMSO in content) for 24 h in complete medium and at standard culture conditions. Incubation of cells with culture medium containing DMSO at a final concentration of 0.5 % (v/v) was used as a negative control. Culture supernatants containing free-floating dead cells were removed at the end of the incubation period, replaced with fresh culture medium containing MTT at a final concentration of 0.5 % (w/v). The cell layers were then further incubated for another four hours. Subsequently, culture supernatants were removed, the cells adherent to the plate surface were collected in 100 % DMSO and incubated for 15 min at 37 °C to terminate the reaction and to dissolve formazan crystals. The absorbance of formazan formed by Rhododendron leaf extract-treated and non-treated control cell cultures was quantified at 595 nm in a microplate reader against the solvent (Tecan Group, Männedorf, Switzerland). Percentages of cell viability were calculated from triplicates using Eq. (1):

The two cell lines were grown on cover glasses in 24-well Bio-One Cellstar plates (Greiner) to reach the desired degree of confluence. Next, cells were incubated with three different concentrations of Rhododendron leaf extracts (i.e. 5, 50, or 500 μg/mL) for 24 h as described above. Subsequently, cells were washed three times with phosphate-buffered saline (PBS) before being incubated for 45 min in 2 mg/mL propidium iodide (PI) (Carl Roth) and 5 μM Draq5™ (Biostatus, Leicester, UK) in culture medium at 37 °C. After washing three times in PBS, cells were fixed in 4 % paraformaldehyde (PFA) (Carl Roth) in 200 mM HEPES (pH 7.4) at room temperature for 20 min. Cells on cover glasses were washed again in PBS and distilled water before mounting them in Mowiol for subsequent laser scanning microscopy as described previously [36]. PI is not capable of penetrating cells with intact plasma membranes, however, if plasma membrane integrity is lost, PI gains access to the nucleus and forms complexes with the DNA. In contrast, Draq5™ serves as a nuclear counter-stain that transverses the intact plasma membrane and can therefore be used to determine the total cell number. Special care had to be taken when analyzing total cell numbers, because some plant leaf extracts could have exhibited anti-adhesive effects such that total cell numbers were significantly diminished after washing steps. Therefore, total cell numbers were determined and reported herein as a measure for anti-adhesive properties of Rhododendron-derived compounds.

Cells were incubated and treated as described above, before washing twice in phenol red-free HEPES-buffered culture medium for 5 min. Subsequently, the cells were incubated with phenol red-free culture medium containing 20 mM HEPES and 500 nM MitoTracker® Red CMXRos (Molecular Probes, Oregon, USA) for 45 min at 37 °C followed by several washes. The fluorescent dye accumulates in the mitochondrial matrix only when an intact membrane potential, due to active cellular metabolism, is present across the inner mitochondrial membrane. Cells were fixed with 4 % PFA in 200 mM HEPES (pH 7.4) for 20 min at room temperature, rinsed, and mounted on microscope slides as described above for subsequent microscopic inspection.

Stained cells were visualized with an LSM 510 confocal laser scanning microscope (Carl Zeiss, Jena, Germany) at excitation wavelengths of 488 nm, 543 nm and 633 nm for fluorophore excitation to visualize FITC-phalloidin, PI or MitoTracker® Red CMXRos, and Draq5™, respectively. Scans at a resolution of 1024 x1024 pixels were taken in the line averaging mode and at a pinhole setting of one airy unit. Color coding and image analysis was performed by using the LSM 510 software, release 3.2 (Carl Zeiss).

For IEC6 cells, induction of apoptosis upon incubation with R. ferrugineum and R. cinnabarinum leaf extracts at the highest concentration, i.e. 500 μg/mL, was evaluated at different time intervals ranging from 1 to 24 h. The apoptosis assay was performed using the EnzChek Caspase-3 assay kit (Invitrogen, Karlsruhe, Germany) detecting activation of procaspase-3 and other Asp-Glu-Val-Asp (DEVD)-specific proteases. Lysates of treated IEC6 cells and non-treated controls were prepared according to the manufacturer’s protocol. Following clearance by centrifugation, the samples were incubated with 5 mM Z-DEVD-R110 substrate for 30 min at 4 °C. Lysates of IEC6 cells treated for 4 h at 37 °C with apoptosis-inducing staurosporin (10 mM) (Sigma-Aldrich) were used as positive controls, whereas no treatment or incubation with the solvent served as negative controls. Additionally, staurosporin-treated cells incubated with 1 mM of Ac-DEVD-CHO for 10 min served as a negative control since caspase-3 activity is blocked under these conditions. The extent of procaspase-3 activation was determined by fluorescence of liberated rhodamine upon excitation at 496 nm and reading the emission at 520 nm, using a microplate reader (Tecan Group, Männedorf, Switzerland). The values were normalized to equal amounts of DNA in the pellets after lysis, as determined by the Burton assay [37].

The minimum inhibitory concentration (MIC) was defined as the lowest concentration of Rhododendron leaf extract that inhibits visible growth of microorganisms after overnight incubation. The MIC was determined by a two-fold dilution assay in Mueller-Hinton broth (MHB) (Becton Dickinson, Heidelberg, Germany). The Bacillus subtilis strain S168 was tested against 12 Rhododendron crude extracts (Table 1) [25]. All tests were performed in triplicates following the National Center for Clinical Laboratory Standards recommendations [38].

All assays were performed in triplicates and repeated at least three times in independent experiments unless stated otherwise. All data were expressed as means ± standard deviation (SD), as determined by using Origin software (MicroCal Software, Northampton, USA). The profile map shown in Fig. 9 was created using R (RStudio, Boston, USA). Levels of significance were calculated by One-Way ANOVA, and p < 0.05 was considered statistically significant. CellProfiler software [39] was used to determine total cell numbers (Draq5™-positive cells) versus numbers of dead cells (PI-positive cells). This software was also employed to quantify the MitoTracker® Red CMXRos and FITC-phalloidin fluorescence signal intensities as previously described by us [32].

In order to group the 12 selected Rhododendron species [25] according to their antibacterial activities, minimum inhibitory concentration (MIC) tests were conducted against B. subtilis . Accordingly, the plant species were classified into four major groups: six Rhododendron species formed the group with the highest antibacterial activity with an MIC of 50 μg/mL: R. minus, R. racemosum, R. ferrugineum, R. rubiginosum, R. anthopogon ssp. anthopogon, and R. concinnum. Another three species formed the group with moderately active extracts, with an MIC of 100 μg/mL: R. cinnabarinum, R. hirsutum, and R. ambiguum. The remaining Rhododendron species exhibited lower antibacterial activities with R. xanthostephanum and R. polycladum having an MIC of 150 μg/mL and R. hippophaeoides var. hippophaeoides requiring 300 μg/mL to efficiently produce an inhibition zone for B. subtilis.

The effects of leaf extracts prepared from 12 different Rhododendron species on cell viability and proliferation rates were initially estimated with the help of the MTT assay, as this test allows for a rapid screening of many samples. To this end, three different concentrations of Rhododendron leaf extract (5, 50, and 500 μg/mL) were applied to IEC6 and HaCaT cells for 24 h. As demonstrated in Fig. 1, incubation with extracts applied at lower concentrations (5 and 50 μg/mL) revealed no detectable change in MTT conversion rates of HaCaT cells, while the leaf extracts from R. polycladum, R. concinnum, and R. xanthostephanum affected IEC6 cellular metabolism negatively (arrows) in comparison to cells that were not treated at all, or treated with 0.5 % DMSO, suggesting that even 5 μg/mL of these extracts caused cytotoxic effects on intestine epithelial cells. Rhododendron leaf extracts used at the higher concentration of 500 μg/mL were more effective in reducing the MTT conversion ability of the treated cells (Fig. 1). Extracts of R. rubiginosum, R. cinnabarinum, and R. ferrugineum exerted cytotoxic effects as deduced from the significant decrease in the ability of both IEC6 and HaCaT cells to reduce MTT. In addition, samples from R. minus, R. polycladum, R. concinnum, R. ambiguum, and R. hirsutum induced statistically significant reductions in the MTT conversion ability of IEC6 cells, but not HaCaT cells. Interestingly, treatments with leaf extracts of R. hippophaeoides var. hippophaeoides, R. anthopogon ssp. anthopogon, and R. racemosum did not exhibit any significant alterations in metabolic activities or cell viability highlighting their potential non-cytotoxicity (Fig. 1).

In order to verify the initially observed cytotoxicity of the Rhododendron leaf extracts on IEC6 and HaCaT cell lines, changes in the integrity of the plasma membrane of the cells upon incubation with leaf extracts were tested. For this, PI acquisition was assayed, which occurs only in those cells that feature ruptured plasma membranes. Cell staining with Draq5™ allowed an estimation of the total cell number. The results summarized in Table 2, Fig. 2, and Additional file 1: Figure S1 demonstrated that incubation of IEC6 cells with 5 and 50 μg/mL of most Rhododendron leaf extracts did neither significantly reduce the total cell number nor affect plasma membrane integrity. However, incubation of IEC6 cells with 50 μg/mL leaf extracts of R. polycladum, R. concinnum, R. anthopogon ssp. anthopogon, and R. hirsutum resulted in a significant reduction in the total cell number as compared to controls (Fig. 2a). Application of the highest concentration of the majority of leaf extracts resulted in a significant decrease in the total cell number and a dramatic decrease in plasma membrane integrity (Table 2, Additional file 1: Figure S1). This suggested that the observed effects on IEC6 cell cultures were likely due to both, massive cell de-adhesion and cell death via plasma membrane rupturing of the majority of remaining cells, irrespective of the leaf extract used. Interestingly, the leaf extract of R. ambiguum had a remarkably divergent effect on IEC6 cells as opposed to all other extracts at 500 μg/mL: although almost all cells had lost their plasma membrane integrity, they remained adherent to the bottom of the incubation vessels (Additional file 1: Figure S1, Table 2), indicating that the extract of R. ambiguum potentially induces effects different from those of the other species.

HaCaT keratinocytes exposed to Rhododendron leaf extracts at any of the concentrations tested proved more tolerant than IEC6 cells under the same conditions. The total cell number was only significantly diminished upon incubation of HaCaT cells with 500 μg/mL leaf extracts from four Rhododendron species, i.e. R. cinnabarinum, R.concinnum, R. xanthostephanum, and R. racemosum (Fig. 2b, Additional file 2: Figure S2). Three out of those treatments followed the previously observed major trend: A combination of cell de-adhesion and plasma membrane disruption of HaCaT cells was observed when a high concentration of plant extract was applied (Table 2). Interestingly, the extract of R. xanthostephanum led to de-adhesion but not to disruption of plasma membrane integrity. Irrespective of the level of reduction in total cell number caused by 500 μg/mL of extract (Fig. 2b), five out of the 12 Rhododendron leaf extracts did not induce plasma membrane rupture in HaCaT cells (Table 2). This result indicated significant differences in the susceptibility of the two different cell types to the tested Rhododendron leaf extracts.

Changes of the mitochondrial membrane potential of IEC6 and HaCaT cells, induced by Rhododendron leaf extracts were determined by MitoTracker® Red CMXRos. For this, fluorescence intensity of stained mitochondria was quantified by measuring the average intensity over arbitrarily chosen inspection areas (Fig. 3). The measured staining intensity is directly proportional to the extent of metabolically active mitochondria visualized by fluorescence. In addition, Rhododendron leaf extract-treated and MitoTracker® Red CMXRos-stained cells were inspected under a fluorescence microscope in accordance with morphological criteria that allow for the determination of the shape of mitochondria (Fig. 4, Additional file 3: Figure S3). Typically, mitochondria of metabolically active, well-adherent IEC6 cells with an intact cytoskeleton exhibited an elongated appearance (Fig. 4a, control), while the mitochondria of HaCaT keratinocytes appeared oval or doughnut-like in shape (Fig. 4b, control).

IEC6 cells incubated with leaf extracts from all Rhododendron species at the highest concentration were dramatically affected with regard to the mitochondrial membrane potential as deduced from the drastically reduced MitoTracker® Red CMXRos staining although effects were somewhat milder for leaf extracts from R. hippophaeoides var. hippophaeoides, R. xanthostephanum, R. hirsutum, and R. racemosum (Fig. 3a). Alterations in mitochondrial structure of IEC6 cells treated with leaf extracts were frequently observed at all three concentration (Additional file 3: Figure S3). However, IEC6 cells treated with 5 or 50 μg/mL extracts from R. hippophaeoides var. hippophaeoides, R. xanthostephanum, R. hirsutum, and R. racemosum did not show significant differences in the metabolic activity and mitochondrial structure when compared to controls (Figs. 3a and 4a).

Effects of Rhododendron leaf extracts on mitochondrial structure and metabolic activity, i.e. staining intensities, were much less pronounced in HaCaT keratinocytes (Figs. 3b and 4b, Additional file 3: Figure S3b). Exceptions were observed when HaCaT cell cultures were treated with high concentrations of leaf extracts prepared from R. minus, R. cinnabarinum, R. ferrugineum, R. concinnum, R. anthopogon ssp. anthopogon, and R. ambiguum (Fig. 3b) with mitochondria that no longer appeared elongated but were rounded up (Fig. 4b, Additional file 3: Figure S3b).

Next, we inspected the filamentous actin cytoskeleton of Rhododendron leaf extract-treated IEC6 and HaCaT cells as a measure for the preservation of the overall cellular architecture. With regard to the intensity of FITC-phalloidin staining of the F-actin system of both IEC6 and HaCaT cells, the analyses revealed mostly mild effects of the Rhododendron leaf extracts when applied at concentrations of 5 or 50 μg/mL (Fig. 5). Likewise, when morphologically inspecting the cytoskeleton of either cell type, no visible changes to the cortical F-actin system were caused by the lower concentrations of leaf extracts. The corresponding structures remained detectable underneath the plasma membranes of most cells (Fig. 6). Conversely, IEC6 cells exposed to any of the concentrations of R. ambiguum leaf extracts showed a significant decrease in the intensity of phalloidin-staining (Fig. 5a). Five of the Rhododendron leaf extracts applied at the highest concentration, namely R. cinnabarinum, R. ferrugineum, R. concinnum, R. xanthostephanum, and R. anthopogon ssp. anthopogon, resulted in a significantly reduced staining intensity of the filamentous actin system in both cell lines (Fig. 6). In contrast, the extracts of R. minus, R. rubiginosum, and R. polycladum exerted negative effects on the staining of F-actin in IEC6 cells only (Fig. 6a). This suggested a rather heterogeneous spectrum of effects on the actin cytoskeleton by various Rhododendron extracts. No particular morphological phenotype could be observed in association with Rhododendron extract-treated HaCaT keratinocytes (Fig. 6b, Additional file 4: Figure S4).

The findings detailed above suggested that IEC6 and HaCaT cells remained either adherent within or to the monolayers, or that they detached upon incubation with specific Rhododendron leaf extracts. Such observations could be falsely interpreted as both cell types being able to tolerate exposure to cytotoxic agents only to some extent. Because the above assays were technically restricted to adherent cells in monolayers, we next analyzed the fraction of free-floating cells which detached during treatment with Rhododendron leaf extracts using the same staining methods as described above. Therefore, Draq5™ staining additionally served to examine the status of nuclear DNA and to identify morphological alterations of the nuclei, such as those that are typical for cells undergoing programmed cell death.

Treatment of IEC6 cells with 500 μg/mL leaf extracts from R. cinnabarinum and subsequent analysis of the detached cells revealed nuclear condensation, cell shrinkage, rounding-up, loss of contacts with adjacent cells, formation of typical membrane blebs and occurrence of apoptotic bodies (Fig. 7a). These results suggested that leaf extracts of this particular plant species may induce apoptosis.

However, IEC6 cells exposed to 500 μg/mL R. ferrugineum leaf extracts became pycnotic and the actin filaments formed a ring surrounding the nucleus. In addition, IEC6 cells treated with leaf extracts from R. minus, R. rubiginosum, and R. ambiguum showed different stages of chromatin condensation and shrinkage of the nuclei (Fig. 7). Thus, five Rhododendron leaf extracts, namely R. hippophaeoides var. hippophaeoides, R. cinnabarinum, R. ferrugineum, R. xanthostephanum, and R. racemosum induced signs closely related to the classical symptoms of programmed cell death –apoptosis– where the treated cells also exhibited typical phenotypes like formation of plasma membrane blebs.

HaCaT cells too showed cellular changes indicative of cell death upon exposure to 500 μg/mL Rhododendron leaf extracts. However, these were different from the phenotypes observed in treated IEC6 cell cultures. HaCaT cells treated with leaf extracts from either R. cinnabarinum or R. ferrugineum displayed signs of the final stages of cell death, reminiscent of cornification, because they exhibited intense PI staining throughout the nuclei and the cytoplasm, while some cells had lost their nuclei altogether (Fig. 7b, G and H).

Moreover, exposure of HaCaT cells to R. minus, R. concinnum, and R. anthopogon ssp. anthopogon induced changes that featured shrinkage of nuclei and chromatin condensation (Fig. 7b, I).

In order to substantiate the interpretation of some of the observed morphological changes induced by specific Rhododendron leaf extracts, the cells grown in monolayers and treated with the extracts were subjected to an additional apoptosis-proving assay. Besides some of the above noted symptoms, one definitive characteristic of apoptosis is the activation of procaspase-3. Therefore, a caspase-3 activity assay was applied to all treatments of IEC6 cells. Incubation of the cells with 500 μg/mL leaf extracts from any of the 12 Rhododendron species indeed induced apoptosis as evidenced by a significant increase in the levels of caspase-3 activity (data not shown). There was no significant activation of procaspase-3 when IEC6 cells were treated with 5 or 50 μg/mL of all leaf extracts prepared from Rhododendron species. Interestingly, cells treated with extracts from R. cinnabarinum and R. ferrugineum at concentrations of 500 μg/mL clearly exhibited phenotypic changes characteristic to apoptosis (Fig. 7). Thus, we selected the leaf extracts of these two Rhododendron species to analyze procaspase-3 activation in free-floating IEC6 cells in a time-dependent manner (Fig. 8). The results revealed a steady increase in caspase-3 activity levels until 12 h of treatment with extracts of R. cinnabarinum, and to a five-fold lesser extent upon treatment with R. ferrugineum extracts. However, caspase-3 activity decreased during the next 12 h. These results argue that components contained in Rhododendron leaf extracts induce apoptosis in IEC6 cells when applied in high enough concentrations.

The partially complex data acquired herein with different cell toxicity analysis assays are summarized by grouping the 12 Rhododendron species according to their antibacterial effectiveness with respective MICs of 50, 100, 150, or 300 μg/ml, and qualitatively comparing their effects against both cell types (Fig. 9). In general, most Rhododendron leaf extracts exerted more pronounced effects on IEC6 intestine epithelial cells as compared to HaCaT keratinocytes, when applied at high concentrations (500 μg/mL). R. hippophaeoides var. hippophaeoides, which exhibited the lowest antibacterial effect, also proved to be least toxic towards both mammalian cell types. A total of five Rhododendron extracts with high antibacterial potential (MIC of 50 μg/mL) did not reveal cytotoxicity against the mammalian cell lines in any of the tested assays when applied at 50 μg/mL, indicating that these extracts are unlikely to harm mammalian cells while killing bacterial cells. Thus, these five extracts are the candidates to be further assessed for possibly containing bio-active compounds with antimicrobial potencies, while still proving safe to be applied onto epidermal or intestine mucosal cell monolayers. The corresponding plant species were R. minus, R. racemosum, R. ferrugineum, R. rubiginosum, and R. concinnum. Interestingly, only the extracts of R. minus and R. racemosum proved to be non-cytotoxic to both intestine epithelial cells and keratinocytes (Fig. 9), suggesting they are the most promising candidates for future investigations on the search for optimized antibiotics in bio-active plant extract and, therefore, to be used for the identification and purification of specific compounds derived from Rhododendron.