Cytotoxic Activity of Amaryllidaceae Plants against Cancer Cells: Biotechnological, In Vitro, and In Silico Approaches

Cancer is a major cause of death and an impediment to increasing life expectancy worldwide. With the aim of finding new molecules for chemotherapeutic treatment of epidemiological relevance, ten alkaloid fractions from Amaryllidaceae species were tested against six cancer cell lines (AGS, BT-549, HEC-1B, MCF-7, MDA-MB 231, and PC3) with HaCat as a control cell line. Some species determined as critically endangered with minimal availability were propagated using in vitro plant tissue culture techniques. Molecular docking studies were carried out to illustrate binding orientations of the 30 Amaryllidaceae alkaloids identified in the active site of some molecular targets involved with anti-cancer activity for potential anti-cancer drugs. In gastric cancer cell line AGS, the best results (lower cell viability percentages) were obtained for Crinum jagus (48.06 ± 3.35%) and Eucharis bonplandii (45.79 ± 3.05%) at 30 µg/mL. The research focused on evaluating the identified alkaloids on the Bcl-2 protein family (Mcl-1 and Bcl-xL) and HK2, where the in vitro, in silico and statistical results suggest that powelline and buphanidrine alkaloids could present cytotoxic activity. Finally, combining experimental and theoretical assays allowed us to identify and characterize potentially useful alkaloids for cancer treatment.


Introduction
Cancer is a major cause of death and a severe impediment to increasing life expectancy worldwide [1]. Breast, prostate, stomach, and uterine cancers are the main types of cancer in terms of the incidence and mortality cases produced [2]. Breast cancer was the most frequently diagnosed malignancy among women worldwide in 2020, and its burden has been growing with a total of 2.3 million new cases (11.7%) [3,4]. By 2040, the burden from this type of cancer is predicted to increase to over 3 million new cases [2]. Among all new cases of all cancers combined (19.3 million), the burden of prostate, stomach, and corpus uteri cancer corresponds to 7.3%, 5.6%, and 2.2%, respectively [3]. In addition, almost 10.0 million cancer-related deaths occurred in 2020 [3,5]. Stomach (0.8 million, 7.7%), female breast (0.7 million, 6.9%), prostate (0.4 million, 3.8%), and corpus uteri (0.1 million, 1.0%) cancer contribute a significant number of cancer-related deaths [3]. Although stomach cancer is a highly heterogeneous disease, it is considered the third cause of death among oncological patients [3,6]. Additionally, an estimate suggests one in six cancer deaths in women was due to breast cancer [2]. Although prostate cancer incidence and mortality rates and trends are decreasing compared to other types of cancer [3,5], this illness remains Table 1 shows the identification by genus and species of the materials and their collection origin.

Establishment of In Vitro Culture
The species C. subedentata, C. tenera, E. caucana, and E. formosa were collected in the field and a bulb of each was selected for the micropropagation process. Figure 1 shows the results obtained in the in vitro multiplication of these four materials, and significant differences were found. E. caucana showed the lowest multiplication rate corresponding to 1.5 shoots/explant, while the other three species did not show significant differences with a multiplication rate of 2.8 shoots/explant. In the rooting phase, two to three roots were obtained per shoot.

Plant Material
The specimens were explored, collected, and herborized. Subsequently, the identification of the material was made using specialized bibliography and comparison with the reference collection deposited in the herbarium of the University of Antioquia (HUA). Table 1 shows the identification by genus and species of the materials and their collection origin.

Establishment of In Vitro Culture
The species C. subedentata, C. tenera, E. caucana, and E. formosa were collected in the field and a bulb of each was selected for the micropropagation process. Figure 1 shows the results obtained in the in vitro multiplication of these four materials, and significant differences were found. E. caucana showed the lowest multiplication rate corresponding to 1.5 shoots/explant, while the other three species did not show significant differences with a multiplication rate of 2.8 shoots/explant. In the rooting phase, two to three roots were obtained per shoot. and naphthaleneacetic acid (NAA, 0.2 mg/L) on the formation of in vitro shoots in C. subedentata, C. tenera, E. caucana, and E. formosa explants. According to Tukey's test, different letters indicate significant differences at the level of (p ˂ 0.05).

Cytotoxic Activities of the Different Tested Alkaloid Fractions
The cytotoxic effects of the ten alkaloid fractions of Amaryllidaceae species (Table 2) were assessed on breast cancer cell lines BT-549, MCF-7, and MDA-MB 231, gastric cancer cell line AGS, prostate cancer cell line PC3, and uterine cancer cell line HEC-1B. Results According to Tukey's test, different letters indicate significant differences at the level of (p < 0.05).

Cytotoxic Activities of the Different Tested Alkaloid Fractions
The cytotoxic effects of the ten alkaloid fractions of Amaryllidaceae species (Table 2) were assessed on breast cancer cell lines BT-549, MCF-7, and MDA-MB 231, gastric cancer cell line AGS, prostate cancer cell line PC3, and uterine cancer cell line HEC-1B. Results were expressed as the percentage of cell viability at 30 µg/mL, considering the American National Cancer Institute (NCI) guidelines for a promising crude extract, which states that a crude extract with IC 50 less than 30 µg/mL is a candidate for chemical characterization [29,30]. As illustrated in Table 2, except for E. formosa, the alkaloidal fractions did not affect the viability in the control cell line HaCat, as the percent cell viability is greater than 90%, showing their low toxicity to the control line. A percentage equal to or below 50% of cell viability was taken as a control point. . It is important to clarify that although E. formosa presented anti-proliferative activity against AGS and MDA-MB 231, it was the cell line that presented the lowest viability for the HaCat control line (<90%). However, the best results (lower cell viability percentages) were obtained for C. jagus (48.06 ± 3.35%), C. subedentata (52.15 ± 2.10%), and E. bonplandii (45.79 ± 3.05%) in the AGS cell line. Therefore, using this cell line, IC 50 values were determined through the MTT assay for some Amaryllidaceae alkaloids such as haemanthamine (13.18 ± 0.47 µg/mL or 43.74 ± 1.56 µM) and lycorine (4.17 ± 0.18 µg/mL or 14.51 ± 0.62 µM). In addition, doxorubicin (5.73 ± 0.80 µg/mL) was used as a positive control.
In the case of the PC3 and HEC-1B cell lines, none of the alkaloidal fractions showed significant cell death results, with survival percentages above 70%. These results suggest that under the conditions in our laboratory, and with an acute treatment for 24 h at a concentration of 30 µg/mL, some alkaloidal fractions are selective and present cytotoxic potential against certain types of cancer cell lines, without compromising the safety in this case, of immortal keratinocytes (HaCat cell line). Interestingly, 80% of the alkaloidal fractions that were active are native/endemic species from Colombia, showing the chemotherapeutic potential of the chemistry of Colombian Amaryllidaceae in molecular targets of pharmacological interest.

Alkaloid Profile of the Different Alkaloid Fractions
All examined alkaloid fractions were chemically characterized by the presence of their different profiling of Amaryllidaceae alkaloids using GC/MS. The analysis of alkaloid identification was performed by comparing the fragmentation pattern reported in the literature and database of Amaryllidaceae alkaloids. Table 3 shows that distribution and abundance depend on both genus and species. A total of 30 alkaloids derived from crinane, galanthamine, and lycorine-type alkaloids were identified. The highest alkaloid content was found in C. subedentata (3.925, 1.824, and 1.225 mg/g DW, respectively) and Z. carinata (2.980, 2.628, and 2.307 mg/g DW, respectively), with 8-O-demethylmaritidine, haemantamine, hamayne, galanthamine, and lycorine alkaloids being the major contributors in C. subedentata. The species C. jagus is characterized by the production of an exclusive group of crinane-type alkaloids corresponding to crinine acetate, buphanidrine, crinamine, and powelline, which were identified only in this species. According to the Eucharis genus, it was found that E. caucana produces a total of 14 alkaloids followed by E. bonplandii with 12 and E. formosa with ten. This last species is the major producer of crinane, galanthamine, and lycorine-type alkaloids with values of 0.649, 0.933, and 2.400 mg/g DW, respectively. E. caucana produces the greatest diversity of alkaloids with the exclusive presence of 6-Omethylpretazettine, narwedine, and N-formylnorgalanthamine. A total of nine alkaloids were identified in Phaedranassa plants, the species with the lowest diversity and production of alkaloids being crinane, galanthamine, and lycorine-type with values of 0.015, 0.019, and 0.119 mg/g DW for P. lehmannii and 0.054, 0.015, 0.013 for P. ventricosa, respectively. In the alkaloid analysis of Zephyranthes species, fourteen alkaloids were identified in Z. carinata and eight in Z. puertoricensis, with the exclusive presence of galanthine, lycoramine, aulicine, norlycoramine, and assoanine (identified only in Zephyranthes species).

ER REVIEW
8 of 21

Multivariate Analyses of the Cytotoxic Activities of the Alkaloid Fractions
The PCA was performed using information from the Amaryllidaceae species and their alkaloidal composition ( Figure 2). The PCA analysis score plot showed that principal components 1 and 2 explain 51.0% of the total variance, and it was possible to satisfactorily differentiate Amaryllidaceae species according to the alkaloid profile. The PCA loading plot ( Figure 2) shows that the most important contributors for C. subedentata, C. tenera, E. caucana, P. ventricosa, and Z. carinata correspond to the crinane-type alkaloids: crinine (X1), haemanthamine (X6), hamayne (X10), and deacetylcantabracine (X11), followed by typelycorine alkaloids: 11,12-dehydroanhydrolycorine (X19), 5,6-dihydrobicolorine (X13) and galanthine (X30). Similarly, galantamine-type alkaloids such as sanguinine (X14), lycoramine (X27), and galanthamine (X12) are differentiating molecules for these species. These results show that multivariate statistical analysis allows the grouping of Amaryllidaceae species concerning their alkaloid content.  Table 3 describes the codes assigned to the alkaloids.
The screening of alkaloid fractions in cancer cell lines showed that the most promising species were C. Jagus, C. subedentata, E. bonplandii, and E. formosa in gastric cancer cell  The screening of alkaloid fractions in cancer cell lines showed that the most promising species were C. Jagus, C. subedentata, E. bonplandii, and E. formosa in gastric cancer cell line AGS and breast cancer cell line MDA-MB-231, with higher cytotoxic potential in AGS cells. Therefore, the resulting cytotoxic data expressed as percentage values of the cell growth relative to the control were used to develop multivariate statistical analyses; in particular, the supervised partial least square discriminant analysis (PLS-DA) was utilized to establish a correlation between the alkaloid profile and cell viability in gastric cancer cell line AGS (Figure 3). The maximum negative effect on cell growth was caused for crinane-type alkaloids such as crinine (X1), crinane acetate (X3), buphanidrine (X5), crinamine (X7), and powelline (X9) identified in C. Jagus, followed by crinine (X1) and 8-O-demethylmaritidine (X2) present in C. subedentata and E. bonplandii. Finally, 8-O-demethylmaritidine (X2) was identified in E. formosa. Thus, crinane-type alkaloids could be related to possible cytotoxic effects on gastric cancer cells.  Table 3 describes the codes assigned to the alkaloids.

Molecular Docking Analysis
Molecular docking analysis is frequently a useful tool for predicting the affinity of small compounds toward potential molecular targets through the study of binding orientation. Table 4 shows the results of the in silico inhibition of the Bcl-2 family (Mcl-1, Bcl-xL, and Bcl-2) and HK2 enzymes by all the alkaloids identified in the most active fractions. Therefore, the ability of the alkaloids to inhibit cell survival through interaction with these proteins is theoretically explained. The affinity scores obtained for selected compounds showed that the proteins have an affinity for compounds with a binding energy between −3.31 to −8.80 kcal/mol. According to the results of the estimated free energy of binding of both controls, the alkaloids with the lowest interaction energy concerning Mcl-1 protein are 11,12-dehydroanhydrolycorine, deacetylcantabricine and norlycoramine. Structural representations of the best conformation of the complexed Mcl-1 with the three tested al-  Table 3 describes the codes assigned to the alkaloids.

Molecular Docking Analysis
Molecular docking analysis is frequently a useful tool for predicting the affinity of small compounds toward potential molecular targets through the study of binding orientation. Table 4 shows the results of the in silico inhibition of the Bcl-2 family (Mcl-1, Bcl-xL, and Bcl-2) and HK2 enzymes by all the alkaloids identified in the most active fractions. Therefore, the ability of the alkaloids to inhibit cell survival through interaction with these proteins is theoretically explained. The affinity scores obtained for selected compounds showed that the proteins have an affinity for compounds with a binding energy between −3.31 to −8.80 kcal/mol. According to the results of the estimated free energy of binding of both controls, the alkaloids with the lowest interaction energy concerning Mcl-1 protein are 11,12-dehydroanhydrolycorine, deacetylcantabricine and norlycoramine. Structural representations of the best conformation of the complexed Mcl-1 with the three tested alkaloids are shown in Figure 4. Similarly, the alkaloids with the lowest interaction energies with Bcl-xL and HK2 proteins are shown in Figures 5 and 6, respectively.

In Vitro Propagation
Amaryllidaceae is one of Colombia's rarest and most striking plant families and is endangered due to habitat destruction [31]. In vitro plant tissue culture techniques are an alternative for conserving and propagating endangered species [32]. Although in vitro tissue cultures are scarce for Caliphruria and Eucharis species, the reports on the micropropagation of E. grandiflora by direct organogenesis showed that from a mother bulb, it is possible to obtain 3.8 shoots/explant and 133 plants in 5 months [33]. In this study, an adequate multiplication rate was achieved for micropropagated species C. subedentata, C. tenera, E. caucana, and E. formosa, which allows obtaining more than 1000 plants in 6 months, allowing the initiation of habitat enrichment programs with these species. Additionally, it is necessary to develop protocols to stimulate the production of alkaloids with biological potential, which are necessary for many species of Amaryllidaceae.

Alkaloid Fractions Differed in Alkaloid Profiles and Cytotoxic Activities
Naturally-derived compounds may be an important source of novel effective anticancer drugs [19,20,23]. These new chemotherapeutic agents ought to be seen as necessary due to the growing incidence and mortality of the different types of cancer and the development of resistance to conventional anti-cancer drugs [1-3]. Recently, Amaryllidaceae alkaloids have been isolated and tested as anti-proliferative agents [28]. However, searching for the types of Amaryllidaceae alkaloids with the greatest anti-cancer potential is convenient. In this work, the crinane, galanthamine, lycorine, and miscellaneous type alkaloids were identified in ten species of Amaryllidaceae. The alkaloid profile of C. jagus shows that this species produces an exclusive group of crinane-type alkaloids in combination with lycorine-type alkaloids, which agrees with previous phytochemical reports [34][35][36][37]. E. bonplandii, E. caucana, P. lehmannii, and Z. carinata were shown to produce several crinane, galanthamine, and lycorine-type alkaloids [35,38,39], the last with the highest chemodiversity in alkaloids of Amaryllidaceae, which is also in agreement with previous studies [40][41][42]. Furthermore, and to the best of our knowledge, this is the first report of phytochemical data for C. tenera, E. formosa, P. ventricosa, and Z. puertoricensis. Figure 2 shows the separation of the ten species of Amaryllidaceae into four groups. In the PCA loading plot, the alkaloids crinane acetate (X3), buphanidrine (X5), crinamine (X7), powelline (X9), anhydrolycorine (X15), deacetylcantabricine (X11), assoanine (X20), norlycoramine (X29) and aulicine (X28) had the greatest contributions to differentiation and could be considered as characteristic compounds of the species C. jagus and Z. puertoricensis. In general, all these species were characterized by presenting a notable diversity of alkaloids, the vast majority with all types of alkaloids identified, including miscellaneous-type alkaloids. Therefore, the wide diversity of Amaryllidaceae alkaloids (30 alkaloids identified, Table 3) can be considered in the search for potential anti-cancer drugs.
According to the results, gastric cancer AGS cells were the most susceptible to the cytotoxic action of alkaloid fractions, followed by MDA-MB 231 breast cancer cells ( Table 2). The purpose of new chemotherapeutic treatments is selectivity against the affected cells, without affecting healthy cells, which is one of the main problems of oncological treatments [43]. Thus, the AGS cells were subjected to further analysis, due to its higher selectivity index, compared to human keratinocytes HaCat cells (control). The GC/MS analysis of the alkaloid fractions with higher bioactivity showed that some of the identified alkaloids were present in most of these alkaloid fractions at different concentrations. These alkaloids were crinine, hamayne, 5,6-dihydrobicolorine, 11,12-dehydroanhydrolycorine, anhydrolycorine, and lycorine. The correlation between the cytotoxic effect and the relative presence of alkaloids in the active fractions could suggest that the cytotoxic activity obtained is due (at least in part) to the presence of crinane and lycorine-type alkaloids, compounds previously reported as potential anti-cancer agents [44,45]. In this study, the potential of crinane and lycorine-type alkaloids against AGS cells was confirmed with the evaluation of haemantamine and lycorine alkaloids against this cell line, with relevant IC 50 values of 13.18 ± 0.47 µg/mL and lycorine 4.17 ± 0.18 µg/mL, respectively. The suggestion is also supported by the results of PLS-DA analysis (Figure 3), especially for crinane-type alkaloids, the main contributors to the toxicity observed in gastric cancer cells.
In this study, an attempt was made to determine the possible interaction of the Amaryllidaceae alkaloids at the molecular level within the active sites of some molecular targets in gastric cancer, as well as breast, prostate, stomach, and uterine cancers. This in-silico approach should explain, at least in part, the in-vitro assays, which were based on the quantification of cell viability. For this reason, some of the molecular targets highly expressed in a variety of cancer cells and related to this process are those belonging to the Bcl-2 family (Mcl-1 and Bcl-xL) [46], anti-apoptotic proteins, and hexokinase 2 (HK2), which plays a central role in the cellular uptake and utilization of glucose [47]. The anti-apoptotic Bcl-2 proteins protect against cell death by apoptosis and also allow senescent cells to survive [48], and it has been suggested that they might appear in the early phase of carcinogenesis in the stomach [49]. While HK-2 is an important molecular target in gastric cancer because it is involved in regulating accelerated glucose uptake during aerobic glycolysis [50], and these cells rely preferentially on aerobic glycolysis for energy generation rather than on mitochondrial respiration [51].
Concerning the interaction with HK2 protein, 6-O-methylpretazettine is stabilized by six hydrogen bridge-type interactions with Gln608, Val655, Phe602, Phe604, Pro513 and Ser603 ( Figure 6). Additionally, this compound shows π-alkyl interactions with Pro605 and Cys606 residues, as well as a π-cation interaction with the amino acid Glu708. Stabilization of pseudolycorine occurs due to the presence of six hydrogen bonds with Ser603, Phe604, Asp657, Asn656, Asn683, and Gln739. In addition, there are two π-cation interactions with the amino acids Glu708 and Glu742. Molecular docking analyses for this system did not detect π-π stacking or π-alkyl interactions. Concerning buphanidrine, HK2 protein active site stabilization occurs as follows: five hydrogen bonds with Phe604, Phe602, Ser603, Asn706, and Pro513, three π-alkyl type interactions with Leu512, Cys606, and Pro605; and finally a π-cation interaction with Glu708. Powelline is also stabilized by the presence of two hydrogen bonds with Phe604 and Phe602 amino acids and two π-alkyl interactions with Pro605 and Cys606 residues. Obatoclax is stabilized in the active site of the protein by the amino acid Cys606, through hydrogen bonding with different parts of the structure. Additionally, π-π stacking and π-alkyl interactions are formed by Leu 512 and Lys 510, while π-cation interactions are established through amino acid Glu 708. Finally, doxorubicin stabilization in the active site is achieved through hydrogen bonding interactions with Gln 739, Cys 606, Ser 603, and Val 655, and two π-cation interactions with Glu 708 and Glu 742 residues.
Considering the results of molecular docking and the PLS-DA analysis, the alkaloids buphanidrine and powelline present in C. jagus belong to the group of alkaloids that could be responsible for the cytotoxic potential in gastric cancer AGS cells (Figure 3) due to the formation of hydrogen bridges and π interactions with the amino acid residues of HK-2 and Bcl-xL enzymes (Figures 5 and 6). Of course, other minor alkaloids (identified or not yet identified) could also act synergistically with cytotoxic activity. However, although these data are theoretical, it has been reported that powelline has antitumor and anticancer activity in in vitro models [54]. Several mechanisms of interest to explain the cytotoxic effect of Amaryllidaceae alkaloids have been explored. However, their ability to selectively induce apoptosis (or programmed cell death) in cancer cells has been found to be a key finding for many cytotoxic Amaryllidaceae alkaloids [27,44,45]. Surprisingly, this mechanism has not been addressed in many Amaryllidaceae alkaloids [55]. However, these reports show the potential of Amaryllidaceae alkaloids for the treatment of several types of cancer. More experiments are required to clarify the activity related to other alkaloids such as buphanidrine. To our knowledge, there are still no reports on its cytotoxic activity in gastric cancer. Therefore, this first theoretical insight provides a guide for crinane-type alkaloids which have emerged as interesting targets for cytotoxicity-based [25].

Plant Material
The plant material of the wild species of Amaryllidaceae was collected in Colombia, in the departments of Antioquia, Cauca, Chocó, Nariño, Quindío, and Tolima. The cultivated species were donated by the botanical garden "José Negret" located in Popayán, Colombia, and micropropagated by plant biotechnology at the Plant Biotechnology Unit, Faculty of Engineering, Universidad Católica de Oriente. Rionegro, Colombia (Table 1). One specimen of each species was deposited in the Herbarium of the University of Antioquia (HUA), Medellín, Colombia. The species studied were collected with authorization from the Ministry of Environment with a genetic resource access contract #328.

Micropropagation of Amaryllidaceae Species
The species corresponding to C. subedentata, C. tenera, E. caucana, and E. formosa are determined as critically endangered and their availability is minimal; for this reason, they were propagated by using in vitro plant tissue culture techniques, while the other species were propagated under nursery conditions using bulb culture. Standardized laboratory techniques were used for in vitro propagation, which consisted of harvesting the bulbs, which were washed with detergent and running water. Then, under aseptic conditions, they were disinfected with 70% ethanol (Sigma-Aldrich, St. Louis, MO, USA) for 2 min and then with 1% sodium hypochlorite (commercial grade, Medellin, Colombia) for 20 min. The modified "twin-scales" technique [56] was used to obtain the explants.

Alkaloid Analyses by GC/MS
Alkaloid analyses were carried out by GC/MS, according to the methods described by Cortes et al. [40]. For the analysis, we used an Agilent 7890 A GC equipment (Agilent, Wilmington, NC, USA) operating in EI mode at 70 eV with autosampler PAL3 Control SW, detector MS 5975C (SCAN analysis by a quadrupole). A capillary column HP-1 (30 m × 0.250 mm × 0.25 µm, Agilent J&W, Palo Alto, CA, USA) PDMS (phenyl polydimethylsiloxane) was used. The oven temperature was initially set at 80 • C (2 min), increased from 80 • C to 210 • C to 15 • C/min, 210-260 • C to 8 • C/min, 260-300 • C to 15 • C/min, and 4 min at 300 • C. The injection of 1 µL of the sample was performed at 280 • C in splitless mode. The flow of carrier gas (helium) was 1 mL/min. The process of identification and quantification of alkaloids was done by spectral deconvolution using the software Agilent MassHunter Qualitative and Quantitative Analysis version B. 07. 00.

Data Processing and Analysis
The alkaloids were identified by comparison of fragments of the mass spectra of the Amaryllidaceae alkaloid spectrotheque of the AgroBio Institute (Sofia, Bulgaria) and information from scientific databases. The compounds were also identified by comparing the mass spectral fragmentation of the compounds with standard reference spectra from the NIST 17 database (NIST Mass Spectral Database, Gaithersburg, MD, USA). The compounds' Kovats Retention Index (RI) was recorded with a standard calibration mixture of n-hydrocarbons (C7-C40, Sigma-Aldrich, St. Louis, MO, USA). The percentage of TIC (total ion current) was estimated for each alkaloid. The abundance of each compound was calculated using codeine (50 µg/mL) as an internal standard. The area of GC/MS peaks depends on the concentration of the related compounds and the intensity of their mass spectral fragmentation.

ATP Quantification by Bioluminescence
Cell viability was determined by bioluminescence assay, based on the quantification of ATP present, as an indicator of metabolically viable cells [58]. The cells were seeded on 96-well plates with a cell density of 2 × 10 4 cells/well for 24 h. After incubation, the medium was changed, and the cells were treated with the alkaloid fraction at a concentration of 30 µg/mL, according to the cytotoxic activity criteria for a crude extract as established by the American National Cancer Institute (NCI) in IC 50 < 30 µg/mL. This criterion for a crude extract has recently been used in research as a screening parameter [29,30], making it possible to reduce the number of experiments in preliminary tests. After 24 h of incubation, 100 µL of CellTiter-Glo ® cell viability reagent (Promega, Madison, WI, USA) was added to each well. The plate was left agitating for 5 min to induce cell lysis, and the plate was incubated at room temperature for 15 min to stabilize the luminescent signal. Relative luminescence units were measured using the GloMax 96 microplate luminometer (Promega, Madison, WI, USA). Cell viability was expressed as a percentage of control values set at 100%.

In-Silico Analysis
The molecular docking analysis of the Amaryllidaceae alkaloids has been carried out using the Autodock 4.2 program [60]. The tridimensional chemical structures of Amaryllidaceae alkaloids and the positive control, obatoclax, and doxorubicin, were downloaded from the PubChem database and were edited using the Maestro program (Schrödinger Release 2022-3: Maestro, Schrödinger, LLC., New York, NY, USA, 2021) available at https://www.schrodinger.com/citations#Maestro (accessed on 1 March 2023), belonging to the Schrodinger suite. In this process, the hydrogen atoms were added, and the protonation states were checked for a pH between 7.0 ± 2.0. Three-dimensional (3D) experimentally known structures of the anti-apoptotic Bcl-2, Bcl-xL, and Mcl-1 proteins were obtained from the Protein Data Bank (PDB), pdb codes: 2W3L [61], 2YXJ [62], and 3KZ0 [63], respectively. Likewise, the three-dimensional structural coordinates of the HK2 protein were retrieved from PDB (PDB ID: 5HEX) [64]. The readers can find theoretical analysis of these structures in the literature [65][66][67][68][69]. The protein structures have been prepared using the Maestro program. In this process, the water molecules, ions, and ligands included in the crystallography pdb file were deleted. Additionally, the bond orders were assigned, hydrogen atoms added, missing side chains included, and amino acid protonation states checked. In the molecular docking experiment, the first step corresponded to computing a set of pre-calculated grids of affinity potentials via AutoGrid, to find suitable binding positions for a ligand on a given macromolecule. In this step, a grid box with dimensions of 60 × 60 × 60 Å and centered in the xyz coordinates for each of the proteins: BCL-2 (40.98, 27.23, −13.55), BCL-XL (−12.28, −17.44, 12.14), MCL-1 (−15.66, 16.19, −1.62) and HK2 (87.0, 15.99, −102.0) were selected. The second stage in the docking experiment corresponded to obtaining the best orientation of a ligand into the active site of a protein, this one treated or selected as a rigid body, through the Lamarckian Genetic Algorithm (LGA) [70]. For this protocol, a population size of 5000 individuals and 50 LGA runs were selected. The best ligand-protein complexes were analyzed according to the potential intermolecular interactions such as hydrogen bonding and the cation-π, π-π stacking. The 2D ligand-protein diagrams were drawn using the software discovery studio visualizer (Dassault Systèmes BIOVIA. Discovery Studio Modeling Environment, Release 2017. Dassault Systèmes; San Diego, CA, USA: 2017). All the alkaloids identified have been submitted to a re-docking process, removing the ligand molecule from the receptor model and then docking it back.

Statistical Analysis
The results of cell viability are shown as the mean ± SD. The statistical significance between the control group and the treatments with the alkaloid fraction were evaluated by analysis of variance (one-way ANOVA) followed by the Newman-Keuls multiplecomparison test, using GraphPad Prism version 5.00 (GraphPad Software, San Diego, CA, USA) data analysis system. In addition, multivariate principal component (PCA) and partial least squares (PLS-DA) statistical analyses were performed using the MUMA package at RStudio (Posit, Boston, MA, USA). The differences were considered significant for p ≤ 0.05 and with respect to the control.

Conclusions
In summary, the present study explored the cytotoxic activity of ten Amaryllidaceae species, characterized by the content of their alkaloids, against six cancer cell lines. The best results (lower cell viability percentages) were obtained for Crinum jagus and Eucharis bonplandii against the gastric cancer cell line AGS. Our results suggest that the cytotoxic activity might be associated with lycorine and crinine-type alkaloids, which is consistent with previous studies. In addition, this is the first report of phytochemical data for C. tenera, E. formosa, P. ventricosa, and Z. puertoricensis. The research focused on evaluating the identified alkaloids in the Bcl-2 protein family (Mcl-1 and Bcl-xL) and HK2, where the in vitro, in silico, and statistical results suggest that powelline and buphanidrine alkaloids could present cytotoxic activity. More tangible evidence is required based on testing the activity of isolated compounds. Finally, combining experimental and theoretical assays allow us to identify and characterize potentially useful alkaloids for cancer treatment.

Conflicts of Interest:
The authors declare no conflict of interest.