Sideritis scardica extracts inhibit aggregation and toxicity of amyloid-β in Caenorhabditis elegans used as a model for Alzheimer’s disease

Introduction

Alzheimer’s disease (AD) is the most common type of dementia and also the most common neurodegenerative disorder in general. As nowadays more people reach a high age than in the past and a cure is still missing, AD is a rising concern for modern civilizations. According to the World Alzheimer Report 2015, about 47 million people suffered from dementia in 2015, and numbers may double every 20 years (Prince et al., 2015). A variety of possible causes for AD are being discussed, of which amyloid-β peptides (Aβ) still play a key role in Alzheimer research and which could be targeted by drugs therapeutically or preventively (Hardy & Selkoe, 2002). Aβ is derived from the amyloid precursor protein (APP) by cleavage through β- and γ-secretases (Selkoe, 1997). The monomers aggregate to oligomers, to polymers and finally to senile plaques, which are abundant in the brain of patients suffering from AD (Lansbury, 1999). The traditional formulation of the amyloid hypothesis blamed those mature aggregates for neurodegeneration, but the smaller oligomers were discovered to be the most neurotoxic Aβ species (Lambert et al., 1998; Walsh & Selkoe, 2004).

Sideritis scardica Griseb. (Lamiaceae) is a perennial shrub endemic to the Balkan peninsula, with Bulgaria as its main habitat. Depending on the area, it is commonly known as Greek mountain tea, Shepherd’s tea, Ironwort, Mursalski tea, Pirinski tea, or Caj Mali. A broad range of traditional uses of S. scardica are known, including the treatment of bronchitis, asthma, sore throat, the prevention of anemia, and the use as tonic or poultice (Todorova & Trendafilova, 2014). Concerning the traditional use against cough associated with common cold and gastrointestinal discomfort, a HMPC (2016) monograph is available. The plant is rich in polyphenols, such as flavonoids, hydroxycinnamic acid derivatives, and phenylethanoid glycosides (Evstatieva, 2002; Petreska et al., 2011). Pharmacological activities like antimicrobial, gastroprotective and anti-inflammatory activity are mostly accredited to this class of secondary metabolites (Tadic et al., 2007; Tadic et al., 2012a; Tadic et al., 2012b). Recently the effects of S. scardica extracts on the central nervous system were addressed in a number of studies. Hydroalcoholic extracts were able to inhibit the reuptake of the monoamine neurotransmitters noradrenaline, dopamine and serotonin in vitro (Feistel & Appel, 2013; Knörle, 2012). Furthermore, they showed antidepressant and psychostimulating effects, as well as a modulation of AMPA-dependent neurotransmission in rats (Dimpfel, 2013; Dimpfel, Schombert & Feistel, 2016a). In mice, cognitive enhancement and Aβ-counteracting effects were observed (Hofrichter et al., 2016). Also, clinical studies have already been performed. S. scardica extracts were able to improve the mental performance of healthy subjects under stress conditions and of subjects suffering from mild cognitive impairment (MCI), which is a precursor of AD (Behrendt et al., 2016; Dimpfel, Schombert & Biller, 2016b). A double-blind, randomized, and placebo-controlled clinical trial currently demonstrates a significant effect of a combination of S. scardica and Bacopa monnieri extract (memoLoges®) on the mental performance of subjects suffering from MCI (Dimpfel et al., 2016c).

To further investigate the influence of hydroalcoholic S. scardica extracts on neurodegenerative diseases and especially on Aβ toxicity and aggregation, we have chosen Caenorhabditis elegans as a model organism (Link, 2006). In transgenic strains expressing human Aβ (1–42), in vivo effects can be observed, that, unlike in vitro studies, also consider bioavailability and other biological influences on a multicellular organism. In the present study, we also tried to figure out the influence of extraction solvents on the content of polyphenolic compounds and pharmacological activity, if a dose-dependency exists, and which extract fractions are potent in order to explore the active principle.

Materials and Methods

Phytochemical analysis

The extracts and fractions were analysed for total polyphenols with a Folin-Ciocalteu UV method following chapter 2.8.14. of the European Pharmacopoeia (2017a) and for specific polyphenolic compounds (flavonoids, acteoside, caffeoylquinic acids) with a HPLC method. For this purpose, a Luna® C18/2 column (Phenomenex, Torrance, CA, USA; 250 mm length, 4.6 mm inner diameter, 5 µm particle size) was used at a temperature of 40 °C in a Shimadzu LC10 HPLC system. 10 µL of about 5 mg/mL sample were injected. The mobile phase was composed of water + 0.1% H₃PO₄ (H₂O) and acetonitrile + 0.1% H₃PO₄ (ACN) with the following gradient: From 95% H₂O/5% ACN (0 min) to 50% H₂O/50% ACN in 41 min; 100% ACN from 45 to 50 min to 95% H₂O/5% ACN until 52 min; 65 min in total. The compounds were detected by DAD at 330 nm and calculated through scutellarin, chlorogenic acid, and acteoside (Phytolab, Vestenbergsgreuth, Germany) as external standards (Fig. 1). Additionally, thin layer chromatography (TLC) was conducted to highlight differences of the fractions. As the stationary phase, silica gel 60 F₂₅₄ was used. The plate was cleaned and activated with ethyl acetate/methanol 50:50 (V/V) and dried at 105 °C for 30 min. 10 µL of preparations from 1 g S. scardica extract and 10 mL ethanol 50% (10 min at 65 °C, filtered) were applied and separated within 15 cm using dichloroethane/acetic acid/methanol/water 50:25:15:10 (V/V/V/V) as mobile phase. After drying, anisaldehyde solution R (European Pharmacopoeia, 2017b) was sprayed on the plate, which was dried again for 3 min at 120 °C.

Results

The phytochemical analysis revealed that basically the content of all tested plant compounds increased with decreasing polarity of the extraction solvent (Table 1). Nevertheless, compared to extraction solvents of stronger lipophilicity, the amount of total phenols in the 40% ethanolic extract was surprisingly high which may be based on the lower drug-extract ratio. From water to 70% ethanol the content of acteoside was enriched almost 13-fold in concentration. The more lipophilic fractions of the 40% ethanolic primary extract (Liq-Liq BuOH, Reprecip. supernat, Resin EtOH) showed higher concentrations of total polyphenols, acteoside and caffeoylquinic acids than the polar ones (Liq-Liq H₂O, Reprecip. precip., Resin H₂O) and also compared to the primary extract (Fig. 2). The butanolic fraction of the liquid–liquid extraction contained especially high yields of the analyzed polyphenolic compounds. It contained concentrations of acteoside and flavonoids that were approximately three times higher than the primary extract. Taking the distribution of mass into consideration, acteoside especially seemed to selectively accumulate in lipophilic solvents used for fractionation. The amount of flavanoids in the fractions obtained from reprecipitation in 70% ethanol constituted the only case of a lower concentration in the unpolar fraction (supernatant) compared to the polar one (precipitate) and to the primary extract. However, the TLC consistently displays the generally higher content of polyphenolic constituents in the lipophilic fractions (flavonoids Rf 0.3–0.5 and tannins/hydroxycinnamic acids Rf 0.7–0.8) in comparison to those with stronger polarity and to the original extract (Fig. 3).

The transgenic C. elegans strain CL2006 constitutively expresses human Aβ (1–42) (Link, 1995). These peptides form aggregates, which were stained with thioflavin S for quantification; Fig. 4 shows the visualized plaques in the head region of the worms (Fig. 4A). In a concentration of 600 µg/mL all S. scardica extracts significantly reduced the number of Aβ aggregates (Fig. 4B). The extract made of 20% ethanol clearly showed a concentration-dependent activity (Fig. 4C), whereas the treatment with 50% ethanolic extract seemed to lose effectiveness when the concentration was raised from 400 to 600 µg/mL. Worms that were treated with 400 µg/mL of lipophilic fractions prepared from the 40% ethanolic extract showed similar plaque numbers to the actual primary extract in the same concentration (Fig. 4D). In contrast, the hydrophilic fractions showed weak or no significant activity (water phase of resin separation: 10.9 ± 0.9% reduction). Taken together, the lowest numbers of Aβ plaques were seen in worms treated with 1,000 µg/mL of the 20%, 400 and 600 µg/mL of the 40%, and 400 µg/mL of the 50% ethanolic extract (20.3 ± 1.4–22.4 ± 0.4% reduction), which was slightly better than the positive control EGCG (19.5 ± 2.1%) studied previously in our laboratory (Abbas & Wink, 2010).

The temperature-inducible expression of human Aβ (1–42) makes worms of strain CL4176 paralyse over time, which is an outcome of Aβ toxicity (Dostal & Link, 2010). The control strain CL802 does not express Aβ. The progression of this paralysis was traced for at least 12 h (Fig. 5). The PT₅₀, a median value describing the point in time when exactly 50% of the worms were paralysed, was calculated to test for statistically significant differences. Extracts and fractions were tested in two sets of experiments, which showed slightly, but not significantly differing values of the negative control (0.5% ethanol) (Table 2). Nevertheless, all treatments were compared to the respective negative control of the test series. All worms treated with 600 µg/mL of the different S. scardica extracts showed a delay of the Aβ-induced paralysis similar to or better than the positive control 100 µg/mL EGCG. The most active extract was the one prepared from 50% ethanol (more than 10% delay), which was also acting in a concentration-dependent manner (Fig. 5B). 600 µg/mL of the 40% ethanolic extract showed about 5% delay, but was tested in another series of experiments, which makes a direct comparison inappropriate, especially as the percentage delay of EGCG also differs from 6 to 3% in the two sets. Treatment with 400 µg/mL of the more lipophilic extract fractions attenuated the progression of the Aβ-induced paralysis (about 6% delay each), whereas the polar fractions failed to increase the PT₅₀ significantly (Table 2; Fig. 5C). Worms of the control strain (not expressing Aβ) that were also treated with the extracts or fractions in the highest used concentration, did not exhibit any paralysis.

Discussion

All S. scardica extracts tested significantly reduced the number of Aβ aggregates and alleviated Aβ toxicity in transgenic C. elegans strains in a concentration of 600 µg/mL or less. Taken together, the 40 and 50% ethanolic extracts were the most active, although EtOH40 did not show a similar percentage delay of paralysis. But a direct comparison of the values is difficult, as they were obtained from two different test series that were performed about 1.5 years apart. In an in vivo system like C. elegans a lot of factors, including behavior, can change to some extent. The worms of the second set basically started to paralyse some hours later and showed slightly different paralysis progression (see Fig. 5). But obviously data are consistent within the sets and none of the active substances tested in the same set showed a significantly higher activity than EtOH40. Thus, compared to extracts prepared from solvents of higher or lower polarity, the two mid-polar extracts showed the strongest activity, although they did not contain the highest content of flavonoids, acteoside, and caffeoylquinic acids as polyphenolic lead compounds. Reasons for this may be based on pharmacodynamic synergisms of certain extract constituents, as plant extracts always embody multicomponent mixtures. To elucidate this question, further studies must be performed. Also, bioavailability of active compounds or even bioenhancing effects may play a role. Polyphenolics contained in S. scardica were reported to be bioavailable, as in a clinical study 5% of polyphenols ingested with a cup of tea were found as metabolites in urine samples via a HPLC-MS measurement (Petreska & Stefova, 2013). Drug absorption is not very well investigated in C. elegans itself, but doubtless compounds have to show similar properties as in vertebrates to be well absorbed from the intestine. According to Zheng et al. (2013) the amount of drugs being absorbed is similar in C. elegans and mice. Furthermore, the specifity of extraction solvents regarding the ratio of active to inactive constituents could lead to higher activity of the mid-polar extracts.

The 20 and 50% ethanolic extracts were tested in different concentrations, showing a dose-dependence in both assays. Only worms of strain CL2006 treated with 600 µg/mL EtOH50 did not show a lower number of Aβ plaques compared to 400 µg/mL. Beginning toxic effects at this concentration are highly improbable, as all chosen treatments were tested for their toxicity on C. elegans (data not shown). More likely, the extract is exhibiting a U-shaped dose–response curve, that is more realistic in biological systems than linear responses (Calabrese & Baldwin, 2001).

All the more lipophilic fractions of the primary 40% ethanolic extract showed a significant reduction in number of Aβ aggregates as well as a delayed Aβ-induced paralysis, with the level of activity being similar to the primary extract or just slightly higher, which points out that the lipophilicity of the extract constituents is perhaps not important alone, otherwise the extract prepared from 70% ethanol would have also shown better results than the mid-polar extracts. However, in most cases the polar fractions did not reveal significant effects, but show trends. So, synergistic effects, maybe of polar and unpolar constituents, are still worth being discussed and investigated. Considering the phytochemical profile of the extract fractions, it is not completely clear if their Aβ-counteracting activities can be attributed to their content of total polyphenols, or to a more specific class of compounds. But as the precipitated fraction of the reprecipitation of EtOH40 in 70% ethanol that contained more flavonoids than its lipophilic counterpart always showed lower activity, this group of compounds may not play a central role. Contemplating the enrichment of acteoside in the unpolar fractions, this phenylethanoid glycoside remains the most promising compound for a potential causal correlation of content and activity.

The Aβ-counteracting activity of hydroalcoholic S. scardica extracts has already been shown in mice (Hofrichter et al., 2016). Here the number of Aβ depositions, as well as the level of soluble Aβ (1–42) was decreased, which is coherent with the results of the present study. Hofrichter et al. (2016) also provided some evidence about the mode of action. They found an intensified Aβ clearance via enhancement of phagocytosis in microglia and induction of ADAM10 expression, a crucial α-secretase, which cleaves Aβ (Esch et al., 1990). An induction of ABC transporter could not be found. An influence of S. scardica on secretases cannot be discussed using the results of the present study as the worms were expressing Aβ through a minigene, not by processing APP. Other possible mechanisms of action against Aβ toxicity involve anti-inflammatory and antioxidant activities (Gilgun-Sherki, Melamed & Offen, 2001; Heneka et al., 2015; Kadowaki et al., 2005; Shelat et al., 2008). As S. scardica has already shown anti-inflammatory properties (Tadic et al., 2007), the inhibition of neuroinflammation is a possible mechanism in vertebrates. But as the nematodes are lacking important structures and mediators which promote inflammation, this is not applicable in the chosen model. Also, several antioxidant activities of Greek mountain tea have been described in vitro (Todorova & Trendafilova, 2014), but no antioxidant effects of the extracts or fractions, including the level of intracellular ROS (reactive oxygen species) and defense against the pro-oxidant compound juglone, was observed in C. elegans (data not shown). S. scardica is rich in polyphenols, which makes a direct interaction with Aβ peptides highly probable, as this is described for many polyphenolic compounds (Porat, Abramowitz & Gazit, 2006; Stefani & Rigacci, 2013). Assembly of peptides can be inhibited by hydrogen or ionic bonds (hydroxyl groups of polyphenols and amino groups of peptides), or by hydrophobic interactions. This direct inhibition of Aβ aggregation and oligomerisation is also described for EGCG, which was used as the positive control (Abbas & Wink, 2010; Del Amo et al., 2012; Wang et al., 2010). A reduced oligomerisation could likewise explain the alleviated Aβ toxicity.

Conclusions

In conclusion, it can be stated that hydroethanolic S. scardica extracts inhibit Aβ aggregation and toxicity in C. elegans with the mid-polar extracts being the most active. This augments existing evidence and makes S. scardica highly interesting for the treatment or prevention of neurodegenerative diseases like Alzheimer’s. Acteoside, a phenylethanoid glycoside, represents a promising, potentially active substance in the extracts and fractions. Nonetheless, further steps have to be taken to investigate the active principle of the extracts and potential synergistic actions of its constituents. In addition, a detailed mechanism of action cannot be stated at the moment; the hypothesized direct inhibition of Aβ aggregation needs further elucidation.

Supplemental Information