Profiling of acetylcholinesterase inhibitory alkaloids from some Crinum, Habranthus and Zephyranthes species by GC–MS combined with multivariate analyses and in silico studies

Abstract Acetylcholinesterase (AChE) inhibitors remain the class of drugs used for the treatment of Alzheimer disease (AD). For the aim of discovering new sources of potent AChE inhibitors, a combined AChE-inhibitory activity together with alkaloid profiles by GC-MS, combined with multivariate statistical analysis for biomarkers determination and in silico studies were attempted. Strategy was applied on leaves, roots and bulbs of six aquatic and terrestrial Amaryllidaceae species. Thirty alkaloids were identified and the AChE inhibitory activities of the extracts were tested by in-vitro Ellman method. Principal bioactive markers were discovered by correlating AChE inhibitory activity with chemical fingerprints via PLS and OPLS modeling which revealed that galanthamine, lycoramine, caranine, tazettine and N-demethylgalanthamine were the most bio-significant markers. Furthermore, the molecular docking was performed to illustrate binding orientations of the top scoring alkaloids in the active site of human acetylcholinesterase. Suggested strategy revealed that, beside galanthamine, caranine, N-demethylgalanthamine, and lycoramine are promising AChE inhibitors. Graphical Abstract


Introduction
The World Alzheimer Report 2018 revealed that more than 46.8 million individuals overall suffered from dementia in 2018 (Patterson 2018). Many Amaryllidaceae alkaloids are known to exhibit a wide range of interesting biological actions (Cabezas et al. 2003). Galanthamine is the best known Amaryllidaceae alkaloid, as an acetylcholinesterase (AChE) inhibitor (Hopkins et al. 2012), which has been widely used in the management of mild to moderate Alzheimer's Disease (AD) marketed with the names Reminyl V R and Razadyne V R (Howland 2010).
Crinum and Zephyranthes species have been the subject of several investigations regarding their acetylcholinesterase inhibitory activity (Kulh ankov a et al. 2013;Cortes et al. 2015;Zhan et al. 2017). Yet there are no previous reports on the alkaloid profile and acetylcholinesterase inhibitory activity of Crinum calamistratum and C. natans aquatic species nor for Z. rosea.
In this paper we adopted a strategy to discover novel sources of alkaloids for potential use in the treatment of AD. An integration of the chemical composition by GC/MS and the enzyme inhibitory activity of AChE by Ellman method (Ellman et al. 1961), followed by multivariate analysis of the GC/MS data and in silico techniques, revealed the principal bioactive alkaloids related to the acetylcholinesterase inhibitory activity. The phytochemical investigation and bioactivity of totally immersed aquatic species is very scarce. This prompted us to compare the metabolic alkaloid profile of some terrestrial Amaryllidaceae species and their corresponding aquatic ones.

Results and discussion
GC/MS has proven to be a useful method for the detection, identification and quantification of Amaryllidaceae alkaloids (Guo et al. 2014;Torras-Claveria et al. 2014). This technique was used to detect 37 compounds, 30 of them identified from their mass spectra and retention indices (Table S1) in the extracts of different organs of six Amaryllidaceae species grown in Egypt. The alkaloid metabolomics profiling of aquatic Amaryllidaceae species is discussed here for the first time.
The GC/MS analysis of the different organs of aquatic C. calamistratum Bogner & Heine and C. natans Baker resulted in the identification of 11 different alkaloids, 7 of them belonging to the lycorine-, crinane-and galanthamine-type (Table S1).
The aquatic species C. calamistratum Bogner & Heine produced two major alkaloids. Lycorine was the main alkaloid in all organs and notably yielded more than 9 ng/mg in roots (dried material ratio in terms of lycorine). The tentatively identified alkaloid from crinane series 8-O-demethylmaritidine was the second one detected in great extent by GC/MS analysis, showing a yield of more than 4 ng/mg from roots as well. The GC/MS technique is not suitable for defining the aor b-ethano bridge from crinane derivatives, yet, the Crinum species are well-known to produce crinane alkaloids from a-series preferably, although some haemanthamine compounds have already been described (Bastida et al. 2006). For the purpose of this work, based on GC/MS approach, the 5,10b-ethano phenantridine representatives detected by GC/MS will be referred to as compounds tentatively identified taking into account the mass fragmentation pattern and retention index, even though the GC/MS technique cannot be used for describing absolute configuration in this case.
Bulbs and roots of the aquatic species C. natans Baker produced galanthamine derivatives (galanthamine, N-demethylgalanthamine and epigalanthamine) even though as minor components while only the bulbs of C. natans Baker produced vittatine (crinine) which was absent from the roots. Lycorine was the main alkaloid from bulbs and roots and 2-O-acetyllycorine was also detected despite as traces. The leaves of C. natans Baker showed no signal of alkaloids in this approach. On average, the alkaloid content was higher in the roots and bulbs than in the leaves and lycorine was the main alkaloid found in both species.
A total of 32 Amaryllidaceae alkaloids were identified in the different organs of the Zephyranthes and Habranthus species (Table S2). Alkaloids were detected in all organs except, leaves and roots of aquatic Z. candida (Lindl.) Herb, roots of Z. rosea Lindl. and leaves of preflowering Habranthus robustus Herb (synonym to Z. robusta Herb). The most commonly found individual alkaloid was the well-known pretazettine artefact tazettine (Wildman and Bailey, 1967), which was present in all the extracts except Z. candida (Lindl.) Herb. leaves and is quantified as tazettine under GC-MS conditions (Bastida et al. 2006). The next most prevalent alkaloids detected in the samples were trisphaeridine, lycoramine, and lycorine. Lycorine-type alkaloids were predominant in terrestrial Z. candida (Lindl.) Herb. organs while tazettine-type alkaloids were predominant in the aquatic species. In general, the alkaloid content of the terrestrial plant is much higher than the aquatic one both qualitatively and quantitatively (Table  S2). We found significant differences between the metabolic profile of aquatic and terrestrial Z. candida (Lindl.) Herb. species. These differences are not only related to the presence or absence of a specific alkaloid but also in the distribution of structural types of the alkaloids present. For instance, lycorine-type alkaloids were the major compounds of the terrestrial Z. candida (Lindl.) Herb. organs whereas tazettine -type alkaloids were predominating in the aquatic species.
In the terrestrial plant, alkaloids of the narciclasine-, galanthamine-, crinane-, lycorine-, tazettine, and lycorenine-types were identified, while the aquatic species contained alkaloids of the narciclasine-, lycorine-and tazettine-types only. Quantitatively, the terrestrial bulbs showed higher amounts of the narciclasine-type (0.405 ng/mg) compared to the aquatic bulbs which contained only 0.047 ng/mg organ weight. This was even more significant in case of lycorine-tyoe alkaloids where the terrestrial bulbs contained 19.60 ng/mg while the aquatic one contained 0.060 ng/mg. meanwhile, the difference in the total alkloi content of the tazettine-type alkaloid was less pronounced where the terrestrial bulbs contained 1.739 ng/mg while the aquatic one contained 0.836 ng/mg.
The most important structural group in the total alkaloid content of Z. rosea Lindl. was the crinane-type. It is noteworthy that the alkaloids were detected only on very minute amounts in the roots and were located mainly in the bulbs. Crinane-type alkaloids were again prevalent in the extracts of Habranthus robustus (Z. robusta Herb) Six alkaloids were unique to certain organs of some species. 5,6-dihydrobicolorine, 2-dehydroxylycorine, 2-O-acetyllycorine and assoanine were only detected in Z. candida (Lindl.) Herb. roots, while sternbergine and 3-epi-macronine were exclusively detected in the bulbs of Z. candida (Lindl.) Herb. bulbs.
The bulbs and root extracts of Z. candida (Lindl.) Herb. stand out in showing Lycorenine-type alkaloids which were lacking from all other Zephyranthes species extracts. Narciclasine-type alkaloids were only absent from Habranthus robustus (Z. robusta Herb) in the preflowering stage while galanthamine-type alkaloids were only absent from Z. candida (Lindl.) Herb. leaves and aquatic Z. candida (Lindl.) Herb.
In most cases, the alkaloid structural type represented by the greatest variety of compounds was the lycorine-type (Z. candida (Lindl.) Herb.) and crinane-type (Z. rosea Lindl. and Habranthus robustus (Z. robusta Herb)) ( Table S2). The diversity of the other types of alkaloids found in the different extracts decreased in the order galanthaminetype > narciclasine-type > lycorenine-type > tazettine-type alkaloids.
Meanwhile, the highest amount of total alkaloids was found in Z. candida (Lindl.) Herb. roots followed by Z. candida (Lindl.) Herb. bulbs (Table S2) which both contained the highest amounts of galanthamine-type, narciclasine-type and lycorine-type alkaloids. The relatively high amount of lycorine alkaloid in Z. candida (Lindl.) roots and bulbs was remarkable.
The skeleton types of alkaloids reported from Z. candida (Lindl.) Herb. include lycorine, homolycorine, crinine and haemanthamine, tazettine, pancratistatin, galanthamine types (Luo et al. 2012;Singh B 2015;Zhan et al. 2016), yet there are no previous reports on their quantitative distribution within the different organs of the plant.
When comparing the different extracts of Habranthus robustus (Z. robusta Herb) at the flowering and preflowering stage, the main striking difference between them is the lack of narciclasine-type alkaloids from the later. Generally speaking, the organs of the preflowering stage accumulate greater amounts of alkaloids than the flowering one except for galanthamine-type alkaloids which are mainly concentrated in the roots of the flowering stage. Meanwhile, the unusual alkaloid galanthindole was detected in Z. candida (Lindl.) Herb. roots and bulbs and the preflowering bulbs of Habranthus robustus (Z. robusta Herb).
Galanthine was the major alkaloid present in Z. rosea Lindl. leaves and bulbs followed by haemanthamine. The amount of alkaloids present on the bulbs were realtively higher than those present in the leaves. Meanwhile, the highest amount of alkaloids were present in Habranthus robustus (Z. robusta Herb) bulbs followed by the roots and finally the leaves. The main alkaloids present in H. robustus bulbs were vittatine, trisphaeridine and tazettine.
The variation in alkaloid composition of the bulbs and leaves within the same species could be explained by ontogenic (seasonal) variability, since it has been observed that Amaryllidaceae alkaloids can vary among tissues during the biological cycle (Viladomat et al. 1986). It is noteworthy that crinane-type alkaloids were predominantly found in Z. rosea Lindl. and Habranthus robustus (Z. robusta Herb), while galanthamine-type alkaloids were found mainly in Z. candida (Lindl.) Herb. roots and bulbs in addition to Habranthus robustus (Z. robusta Herb) roots.
The AChE inhibitory activities were calculated as the percentage inhibition compared to the blank. Galanthamine was used as a positive control (IC 50 : 1.39 lg/mL).
Chemometric modeling using PLSR was applied to rapidly screen bioactive markers from the studied Amaryllidaceae plants. The alkaloids galanthamine, lycoramine, caranine and tazettine appeared in cluster trend with the acetylcholinesterase inhibitory activity data points, suggesting they were most relevant with the AChE inhibitory activity. In detail, the first latent variable explained almost half of the variation (56.6%), followed by 18.5% variances of the latent variable. Meanwhile, the correlation coefficients of each component to the activity are listed in Table S4. It is clearly shown that the values of galanthamine, lycoramine, caranine and tazettine were much larger than other peaks. Furthermore, the importance of the X-variables for the model was expressed by variable importance for the projection (VIP) values. The VIP results were in good accordance with the above given findings (Table S4, Figure S2A) were the alkaloids galanthamine, lycoramine, caranine and tazettine had the highest VIP values. The S-plot results ( Figure S2B) suggest that the most relevant variables are attributed to galanthamine, lycoramine, caranine and tazettine. These results are in accordance to the PLSR model results. It can thus be concluded that these alkaloids were the principal bioactive ACHE inhibitory alkaloid markers. Table S5 shows the results of the molecular-docking simulation performed using the structures of all detected alkaloids. The best conformation of galanthamine in active-site structures have estimated values for the energy of protein-ligand interaction (À8.10 kcal/mol). The simulated molecular docking showed that the alkaloid Caranine (Lycorine-type) theoretically possesses stronger inhibitory activity toward hAChE (gscore ¼ À8.79 kcal/mol) while N-demethylgalanthamine (galanthamine-type) and lycoramine (galanthamine-type) theoretically have slightly lower inhibitory activity toward hAChE than that of galanthamine. Meanwhile, the alkaloid tazettine (Tazettinetype) showed considerable activity with gscore of -6.69 Kcal/Mol. The inhibitory behaviors of the alkaloids were compared against the behavior of galanthamine since its experimental results are available (IC 50 1.39 mg/mL) in this paper. Structural representations of the best conformation of the complexed active site of the hAChE enzyme is presented in Figure 1A.
The hAChE-caranine complex ( Figure 1B) shows that amino acids Tyr124, Tyr133, Trp86 and Phe338 make a hydrogen bond with Tyr133, all located at PAS in the active site of hAChE (Dvir et al. 2010). Furthermore, the charged nitrogen atom of forms two salt bridge with Trp86. A classical p-p interaction was also observed between Try124 and the aromatic ring of caranine, The hAChE-lycoramine complex ( Figure 1C) differs from that of galanthamine complex only by the salt bridge with residue Ser125 as depicted in Figure 1C. Considering the results of the molecular docking experiments and the presence of galanthamine and lycoramine in all active extracts, we may suggest that these compounds are responsible for most of the hAChE-inhibitory activity.