Antiglycation and Amyloid‐β Inhibitory Activities of Pandanus odorifer Extracts and Isolated Constituents: In Vitro and Molecular Docking Insights

ABSTRACT

Alzheimer's disease (AD) is a neurodegenerative disorder associated with amyloid‐β (Aβ) aggregation, advanced glycation end products (AGEs), and oxidative stress. This study investigated the antiglycation, antioxidant, anti‐amyloidogenic, and predicted drug‐likeness properties of Pandanus odorifer (Forssk.) Kuntze. The methanolic crude extract showed high total flavonoid content (121.7 ± 2.50 mg QE/g), moderate total phenolic content (16.61 ± 1.50 mg GAE/g), and strong AGE inhibition (IC₅₀ = 57.30 µg/mL), with moderate radical scavenging activity. Chromatographic separation yielded syringaresinol (1), dehydrovomifoliol (2), and blumenol C glucoside (3), identified through spectroscopic analysis and comparison with published data. In the Thioflavin T assay, compounds 1–3 reduced Aβ‐associated fluorescence, with compound 3 producing the largest reduction (71.20% ± 4.42%) under the tested conditions. Molecular docking suggested interactions with Aβ42 oligomers and AD‐related enzymes, particularly acetylcholinesterase and β‐secretase, while in silico prediction indicated acceptable drug‐likeness without Lipinski violations. Overall, the findings identify P. odorifer as a source of constituents with antiglycation activity and preliminary Aβ‐associated ThT responses that warrant confirmation using orthogonal biophysical and biological assays.

Pandanus odorifer methanolic extract exhibited strong antiglycation activity, moderate antioxidant capacity, and high flavonoid content. Three isolated compounds reduced Aβ‐associated ThT fluorescence, with blumenol C glucoside showing the greatest effect. Molecular docking supported interactions with Aβ42, acetylcholinesterase, and β‐secretase, highlighting P. odorifer as a promising source of anti‐Alzheimer's lead compounds.

1. Introduction

Alzheimer's disease (AD) is a neurodegenerative disease characterized by a chronic decline in cognitive function, memory, and behavior. The clinical syndrome arises from a multifactorial pathobiology rather than from a single molecular lesion. Amyloid deposition, tau‐associated pathology, synaptic and cholinergic dysfunction, oxidative imbalance, neuroinflammation, mitochondrial impairment, and disruption of protein homeostasis interact across the course of the disease [1, 2, 3, 4, 5, 6]. One of the major pathological features of AD is the accumulation of amyloid‐β (Aβ) peptides generated through amyloidogenic processing of amyloid precursor protein. In this pathway, β‐site amyloid precursor protein‐cleaving enzyme 1, commonly referred to as BACE1 or β‐secretase, performs the initiating cleavage, after which γ‐secretase releases Aβ peptides of different lengths. Aβ42 has a relatively high propensity to misfold and self‐associate into soluble oligomers, protofibrils, and β‐sheet‐rich fibrils [7]. Soluble oligomeric assemblies are considered particularly relevant to early synaptic dysfunction because they may disrupt neuronal communication before extensive plaque deposition occurs. Amyloid pathology nevertheless represents only one component of AD, and current drug‐discovery strategies increasingly evaluate complementary targets associated with cholinergic dysfunction, oxidative injury, neuroinflammation, and proteostasis [2, 3, 4, 5, 6, 8]. Recent studies of AChE and BChE inhibition further illustrate the continued therapeutic interest in the cholinergic system [2, 6], while cholinesterase alterations have also been examined in other age‐related degenerative conditions [9, 10].

Oxidative stress, broadly characterized by an imbalance between oxidant generation and antioxidant or repair capacity, is another process implicated in AD. Excess reactive oxygen species can oxidatively modify membrane lipids, proteins, and nucleic acids, impair mitochondrial bioenergetics, and activate redox‐sensitive inflammatory signaling [11, 12, 13]. These responses are context‐dependent; studies in other physiological systems have similarly linked oxidative imbalance with metabolic, inflammatory, and post‐transcriptional remodeling [14, 15]. Plant‐derived phenolics and other natural products may exhibit radical‐scavenging, metal‐chelating, carbonyl‐trapping, or other redox‐modulating activities, although the magnitude and biological meaning of these effects depend on chemical structure, concentration, and assay conditions [13]. Consequently, cell‐free antioxidant measurements should be interpreted as biochemical screening endpoints rather than direct evidence of mitochondrial protection, suppression of inflammation, or neuronal protection. These activities are strongly dependent on chemical structure, concentration, bioavailability, and assay conditions, and phenolic compounds may exhibit antioxidant or redox‐modulating effects depending on the experimental context. Integrated phytochemical studies therefore increasingly combine constituent profiling with biological assays and computational analyses rather than assigning activity solely from total phenolic or flavonoid measurements [13, 16, 17].

Natural products capable of reducing radical‐associated signals, AGE formation, or Aβ‐associated aggregation are frequently investigated as preliminary candidates in AD‐oriented screening [17]. However, activity in these cell‐free assays should not be equated with neuroprotection or disease modification. The present assays establish biochemical activity under defined experimental conditions but do not demonstrate regulation of endogenous antioxidant enzymes, mitochondrial protection, anti‐inflammatory activity, preservation of neuronal viability, or efficacy in living systems [17, 18]. These mechanisms require validation using target‐specific assays, neuronal models, pharmacokinetic studies, and in vivo experiments.

Among the plant genera frequently studied for their medicinal value is Pandanus (commonly known as “pandan”), a member of the Pandanaceae family comprising approximately 500 species distributed across tropical and subtropical regions [19]. For AD‐oriented natural‐product screening, Pandanus is relevant not merely because of its broad medicinal reputation, but because several reported activities of selected species converge on mechanisms directly associated with AD pathology, including oxidative stress, Aβ aggregation, AGE formation, and cholinergic signaling. Several studies have also demonstrated the pharmacological potential of various Pandanus species, including their antioxidative [20, 21, 22], anticancer [20], antimicrobial [20, 21, 22], antidiabetic [23], and anti‐amyloidogenic activities [24, 25, 26]. More importantly for the present study, crude extracts of P. clementis and P. amaryllifolius have been reported to inhibit Aβ aggregation and protect Aβ‐treated SH‐SY5Y neuronal cells, while alkaloids from P. amaryllifolius inhibited AGE formation and Aβ aggregation [24, 25, 26]. Extracts and constituents from P. amaryllifolius [26] and P. tectorius [27] have also shown acetylcholinesterase inhibitory activity, free radical scavenging, and inhibition of Aβ aggregation, which are mechanisms directly relevant to AD. Thus, the available literature suggests that selected Pandanus species may contain structurally diverse metabolites capable of acting on overlapping AD‐related processes rather than on a single biological endpoint alone. Phytochemical analyses have revealed the presence of essential oils, tannins, alkaloids, and glycosides in different parts of the plant, contributing to its medicinal efficacy. Traditionally, pandan leaves from Southeast Asia, including the Philippines, have been used to treat headaches, arthritic pain, stomachache, and wound infections [28, 29].

Pandanus odorifer (Forssk.) Kuntze (syn. Pandanus odoratissimus) is a mangrove‐associated species commonly found along coastal habitats, including shorelines and estuarine forests [28, 29]. Crude extracts from various parts of P. odorifer have demonstrated antioxidant, antidiabetic [23], and antimicrobial properties [22]. Several bioactive constituents have also been isolated and identified, such as the antibacterial steroid stigmast‐5,22‐dien‐3β‐ol from the stem bark of P. odorifer [22], and the antioxidative lignans pinoresinol and 3,4‐bis(4‐hydroxy‐3‐methoxybenzyl) tetrahydrofuran from its roots [30, 31]. However, these studies do not yet sufficiently establish the relevance of P. odorifer to AD‐associated amyloidogenic and proteotoxic mechanisms. In particular, its antiglycation and anti‐Aβ aggregation potential remains underexplored, despite the presence of phenolic and lignan‐associated chemistry that may be mechanistically relevant to oxidative stress, AGE formation, and amyloid aggregation.

In line with our continuing investigation of Pandanus metabolites with AD‐relevant biochemical activities, the present study evaluated the total phenolic and flavonoid contents, radical‐scavenging activity, and AGE‐formation inhibitory activity of a methanolic extract prepared from P. odorifer leaves. Syringaresinol (1), dehydrovomifoliol (2), and blumenol C glucoside (3) were subsequently isolated and evaluated for their effects on Aβ42‐associated aggregation. To the best of our knowledge, this is the first study to evaluate these three compounds isolated from P. odorifer leaves for Aβ42 aggregation‐modulating activity. Molecular docking and in silico pharmacokinetic analyses were used as hypothesis‐generating approaches to prioritize the isolated compounds for subsequent experimental validation.

2. Results and Discussion

Oxidative stress, protein glycation, and amyloid aggregation are interconnected biochemical processes involved in AD pathology [8, 11, 12]. Accordingly, the DPPH, ABTS, and AGE‐formation assays were used to characterize complementary aspects of the activity of the P. odorifer methanolic extract. Total phenolic and flavonoid contents were first determined, followed by assessment of radical‐scavenging and antiglycation activities. AGE formation is particularly relevant because glycation promotes oxidative imbalance, protein cross‐linking, and structural modification of proteins, all of which can favor amyloidogenic processes [32]. Thus, the extract‐level antiglycation assessment and compound‐level Aβ aggregation assay provide complementary evidence that different preparations from P. odorifer act on distinct biochemical processes associated with AD pathology.

2.1. Estimation of TPC and TFC

The total phenolic content of the P. odorifer crude methanolic extract was determined from the gallic acid calibration curve (y = 0.0040x + 0.0975, R ² = 0.9981; Figure S1) and was calculated as 16.61 ± 1.50 mg GAE/g extract. The total flavonoid content, determined using the quercetin calibration curve (y = 0.0036x + 0.0254, R ² = 0.9950), was 121.7 ± 2.50 mg QE/g extract. These findings indicate that the extract contains both phenolic and aluminum‐reactive flavonoid‐associated constituents that may contribute to its biological activity. However, because TPC and TFC are expressed using different reference standards and the aluminum chloride assay does not identify individual compounds, the values should be interpreted as complementary estimates of phytochemical composition rather than direct evidence that flavonoids are the predominant or principal active constituents.

2.2. Screening of DPPH•, ABTS•+, and AGEs Inhibition Activity

The P. odorifer methanolic extract showed weaker radical‐scavenging activity than ascorbic acid and Trolox, with IC₅₀ values of 2.08 mg/mL for DPPH and 0.713 mg/mL for ABTS (p < 0.0001; Table 1). In contrast, it exhibited pronounced AGE‐formation inhibitory activity, with an IC₅₀ of 57.3 µg/mL, lower than that of quercetin.

TABLE 1.

IC₅₀ values of the P. odorifer methanolic extract and assay‐specific reference compounds in DPPH, ABTS, and AGE‐formation inhibition assays.

Samples	DPPH• (mg/mL)	ABTS•+ (mg/mL)	AGE (µg/mL)
PoM extract	2.08 ± 0.11	0.713 ± 0.20	57.3 ± 1.23
Standard a	0.0298 ± 0.06	0.245 ± 0.34	79.3 ± 2.42
p b	<0.0001	<0.0001	<0.0001

^a Standards used for DPPH•, ABTS•+, and AGE assays were ascorbic acid, Trolox, and quercetin, respectively.

^b p < 0.05 indicates a significant difference between the standard and the crude extract.

The strong antiglycation activity despite moderate TPC suggests that the response was influenced more by the chemical identity and combined effects of the extract constituents than by total phenolic abundance alone. Potential contributors include phenolic lignans, norisoprenoid derivatives, glycosides, and other minor metabolites that may interfere with carbonyl‐mediated glycation, protein cross‐linking, metal‐dependent reactions, or AGE‐associated fluorescence [33]. Although the relatively high TFC‐equivalent value indicates the presence of aluminum‐reactive constituents, the activity cannot be assigned specifically to flavonoids without targeted identification and quantification.

Differences between phenolic content and biological activity have also been reported in other Pandanus extracts, where nonphenolic constituents and solvent‐dependent extraction influenced antioxidant responses [19, 28]. These findings indicate that the pronounced AGE inhibition of P. odorifer may arise from a chemically diverse and potentially interactive metabolite profile, supporting subsequent compound isolation, bioassay‐guided fractionation, and compound‐level antiglycation evaluation.

2.3. Isolation and Identification of Compounds From P. odorifer

The methanolic crude extract of P. odorifer was subjected to solvent‐solvent partition using hexane, CHCl₃, and BuOH to yield 0.16, 1.73, and 3.63 g extract, respectively. Repeated chromatographic separations (Figure S2) of the CHCl₃ extract led to the isolation and identification of three known compounds (Figure 1), namely syringaresinol (1) [34], dehydrovomifoliol (2) [35], and blumenol C glucoside (3) [36]. Their identities were confirmed by comparing with published literature (Tables S1–S3).

FIGURE 1.

Compounds isolated from the crude base extract of P. odorifer.

Notably, syringaresinol (1) was isolated from fruit extracts of P. tonkinensis [37], P. kaida [38], and P. tectorius [39, 40, 41]. Similarly, dehydrovomifoliol (2) was also reported in leaf extracts of P. simplex [42] and P. utilis [43]. Last, blumenol C glucoside (3) has been isolated from P. tectorius fruit and leaf extract [39], as well as the leaf extract of P. simplex [42].

Syringaresinol (1) is a phenolic lignan, whereas dehydrovomifoliol (2) and blumenol C glucoside (3) are norisoprenoid derivatives; none of the isolated compounds is a flavonoid. Their isolation confirms that the P. odorifer methanolic extract contains chemically distinct metabolite classes beyond those represented by bulk TPC and TFC measurements. However, because compounds 1–3 were not quantified in the crude extract or evaluated individually in the DPPH, ABTS, and AGE assays, their contributions to the extract‐level activities cannot yet be established.

2.4. Thioflavin T Assay

Compounds 1–3 were evaluated for their effects on Aβ aggregation using the Thioflavin T assay at 100 µg/mL (Figure 2). ThT fluoresces strongly upon binding to β‐sheet‐rich amyloid fibrils and is widely used to monitor amyloid formation [44]. Blumenol C glucoside (3) showed the highest inhibition at 71.20% ± 4.42%, followed by syringaresinol (1) at 66.10% ± 1.28% and dehydrovomifoliol (2) at 64.46% ± 5.13%. Compounds 1 and 2 were significantly less active than morin and phenol red, whereas compound 3 did not differ significantly from either reference compound (p > 0.05). Thus, compound 3 was the most active isolate under the tested conditions, although concentration–response studies are needed to compare their relative potencies.

FIGURE 2.

Effects of compounds 1–3 on Aβ‐associated ThT fluorescence. Morin, phenol red, and compounds 1–3 were evaluated at 100 µg/mL. Data are presented as mean ± SD from three independent experiments. Different letters indicate statistically significant differences based on one‐way ANOVA followed by Tukey's multiple‐comparison test at p < 0.05. The letter a indicates not significantly different from both controls, while b indicates significantly different from both controls at p < 0.05.

These findings extend previous reports of reduced Aβ‐associated ThT fluorescence or Aβ aggregation inhibition by extracts and alkaloids from P. clementis and P. amaryllifolius [24, 25, 26] and further support Pandanus as a source of structurally diverse amyloid‐modulating metabolites. The activity observed for the phenolic lignan syringaresinol and the norisoprenoid derivatives dehydrovomifoliol and blumenol C glucoside also indicates that this effect is not restricted to flavonoid scaffolds. However, because compounds 1–3 were not quantified in the crude extract or evaluated individually in the DPPH, ABTS, and AGE assays, their contribution to the extract‐level antioxidant and antiglycation activities cannot yet be assigned. Concentration–response analysis and orthogonal confirmation of Aβ aggregation will be needed to establish their relative potency and exclude ThT‐associated interference.

2.5. Molecular Docking Studies

To complement the ThT‐based screening, compounds 1–3 were docked against Aβ42 pentameric and dodecameric models and two AD‐associated enzymes, acetylcholinesterase (AChE) and β‐secretase 1 (BACE1). The calculations were used to compare predicted ligand poses, residue contacts, and AutoDock Vina scores within the selected protein models and docking regions (Table 2 and Figures 3, 4, 5). Because direct binding measurements and AChE or BACE1 inhibition assays were not performed, the docking results are interpreted as hypothesis‐generating predictions rather than evidence of target engagement, enzyme inhibition, or neuroprotective activity.

TABLE 2.

Binding energies and interactions between test ligands and target proteins implicated in AD pathophysiology.

Target proteins	Binding energy (BE, kcal/mol)	Interactions
Syringaresinol (1)
Acetylcholinesterase (AChE)	−8.0	Phe338, Tyr337, Trp86, Ser125 (C─H bond), Trp286, Trp86 (pi–pi stacked), Tyr341, Tyr72, His445, Val294, and Trp86 (alkyl/pi–alkyl)
β‐secretase (BACE1)	−7.4	Thr293, Leu324, Thr133 (H‐bond), Lys382 (alkyl/pi–alkyl), Gly291, and Gln134 (C─H bond)
Amyloid‐β42 pentamer	−5.6	LeuA17 (H‐bond), GlyB38 (C─H bond), PheA19 (pi–pi stacked), ValA18, ValB40, ValA40, LeuB34, LeuA34, AlaA21, ValA36, and ValB36 (alkyl/pi–alkyl)
Amyloid‐β42 dodecamer (fibrillar structure)	−7.4	GlyB33 (H‐bond), GlyC33, IleC32 (pi–pi stacked), ValB12, LeuC34, IleB32, HisB14, LeuC17, HisD14, LeuE17, HisE14, and IleE32 (alkyl/pi–alkyl)
Dehydrovomifoliol (2)
Acetylcholinesterase (AChE)	−7.6	Ser125, Tyr124 (H‐bond), Phe338, Tyr337, Trp86 (alkyl/pi–alkyl), Tyr341, Leu130, Tyr133, Gly126, Gly120, Asp74, Pro88, Val73, Tyr72, Asn87, Gly121, Ser203, Gly122, His447, and Gly448 (C─H bond)
β‐secretase (BACE1)	−6.0	Gln134 (H‐bond), Thr133 (C─H bond), Ile179, Phe169, and Tyr132 (alkyl/pi–alkyl)
Amyloid‐β42 pentamer	−4.7	ValA36, PheA19 (alkyl/pi–alkyl), ValB40, ValA40, GlyA38, GlyB38, ValA39, GlyA37, ValB36, AlaA21, and PheA20 (C─H bond)
Amyloid‐β42 dodecamer (fibrillar structure)	−5.2	ValD12, ValC12, HisC14, LeuD17, HisD14, IleD32, LeuE17 (alkyl/pi–alkyl), HisE14, LeuC34, LeuD34, GlyC33, and GlyD33 (C─H bond)
Blumenol C glucoside (3)
Acetylcholinesterase (AChE)	−7.7	Asp74 (H‐bond), Tyr124, Phe338, Tyr337, Trp86, His447 (alkyl/pi–alkyl), Tyr337, Trp86, and His447 (C─H bond)
β‐secretase (BACE1)	−6.5	Arg189, Tyr259, Thr133 (H‐bond), Tyr132, Ile287 (alkyl/pi–alkyl), Tyr259, and Ile187 (C─H bond)
Amyloid‐β42 pentamer	−5.4	AlaA21, ValA36, ValB36, PheA19 (alkyl/pi–alkyl), and AlaA21 (C─H bond)
Amyloid‐β42 dodecamer (fibrillar structure)	−6.2	GlyE33 (H‐bond), LeuD34, ValD12, IleE32, HisD14 (alkyl/pi–alkyl), and HisC14 (C─H bond)
Cognate ligand or co‐crystallized ligand/inhibitor
Acetylcholinesterase (AChE)	−9.4 a	Ser203, Tyr337, Glu202 (H‐bond), Asp74 (attractive charge), Trp86, Phe338, His447, Trp236, Phe297, Phe295 (alkyl/pi–alkyl), and Ser125 (C─H bond)
β‐secretase (BACE1)	−11.8 b	Thr133, Gln134, Gly95, Gly291, Thr293, Asn294 (H‐bond), Asp286, Asp93 (salt bridge/attractive charge) Tyr132 (pi–pi), The 292 (pi–sigma), Ile187 (alkyl), Thr293, Gly72, and Asp289 (C─H bond)
Amyloid‐β42 pentamer	−5.6 c	MetD35 (alkyl/pi–alkyl), MetE35, GlyD37, ValD36, MetC35, GlyC37, ValC36, MetB35, ValB36, GlyB37, ValB39, ValC39, and ValD39 (C─H bond)
Amyloid‐β42 dodecamer (fibrillar structure)	−7.3 c	GlyA33 (H‐bond), HisC14 (pi–pi), ValB12, LeuB34, ValA12, IleC32 (alkyl/pi–alkyl), and HisB14 (C─H bond)

^a Galantamine.

^b N‐((1S,2R)‐3‐(((1S)‐2‐(cyclohexylamino)‐1‐methyl‐2‐oxoethyl)amino)‐2‐hydroxy‐1‐(phenylmethyl)propyl)‐3‐(ethylamino)‐5‐(2‐oxo‐1‐pyrrolidinyl)benzamide [co‐crystallized inhibitor].

^c Morin.

FIGURE 3.

Docking poses and binding interactions between syringaresinol (1) versus (A) AChE, (B) BACE1, (C) Aβ₄₂ pentamer, and (D) Aβ₄₂ dodecamer.

FIGURE 4.

Docking poses and binding interactions between dehydrofomifoliol (2) versus (A) AChE, (B) BACE1, (C) Aβ₄₂ pentamer, and (D) Aβ₄₂ dodecamer.

FIGURE 5.

Docking poses and binding interactions between blumenol C glucoside (3) versus (A) AChE, (B) BACE1, (C) Aβ₄₂ pentamer, and (D) Aβ₄₂ dodecamer.

Compounds 1–3 produced more negative docking scores against the Aβ42 dodecamer model than against the pentamer model. Syringaresinol (1) yielded scores of −7.4 and −5.6 kcal/mol against the dodecamer and pentamer, respectively; dehydrovomifoliol (2) yielded −5.2 and −4.7 kcal/mol; and blumenol C glucoside (3) yielded −6.2 and −5.4 kcal/mol (Table 2; Figures 3C,D, 4C,D, and 5C,D). Soluble Aβ42 oligomers, including dodecameric assemblies, have been associated with synaptic dysfunction and fibril nucleation in AD [45, 46, 47]. However, the differences observed here represent relative scoring within the selected static models and search spaces and do not demonstrate preferential experimental binding, oligomer selectivity, or disruption of Aβ assembly.

The retained pose of syringaresinol (1) in the dodecamer model included a predicted hydrogen bond with GlyB33 and additional contacts within the oligomeric interface (Table 2 and Figure 3D). Dehydrovomifoliol (2) formed predominantly hydrophobic and weak polar contacts involving residues from several peptide chains (Table 2 and Figure 4D). For blumenol C glucoside (3), the glucoside moiety formed a predicted hydrogen bond with GlyE33, together with contacts involving LeuD34, ValD12, IleE32, and HisD14 (Table 2 and Figure 5D). The polar functionality of glycosides can support hydrogen‐bonding contacts in modeled ligand–target complexes [48].

Against AChE, syringaresinol (1), dehydrovomifoliol (2), and blumenol C glucoside (3) yielded docking scores of −8.0, −7.6, and −7.7 kcal/mol, respectively (Table 2; Figures 3A, 4A, and 5A). The predicted pose of compound 1 included contacts with Trp86 and Trp286, residues associated with the catalytic and peripheral anionic regions of AChE [49]. Compounds 2 and 3 also formed predicted contacts with residues lining the AChE gorge, including Trp86, Tyr337, Phe338, Asp74, and His447 (Table 2; Figures 4A and 5A) [49, 50]. These residue contacts indicate that the compounds could be accommodated within the selected docking region, but they do not demonstrate experimental AChE inhibition. For BACE1, syringaresinol (1) produced the most negative score among the isolated compounds at −7.4 kcal/mol, followed by blumenol C glucoside (3) at −6.5 kcal/mol and dehydrovomifoliol (2) at −6.0 kcal/mol (Table 2; Figures 3B, 4B, and 5B). The predicted pose of compound 1 included contacts with Thr133, Gln134, Gly291, and Thr293, residues located within or near regions involved in substrate and inhibitor recognition [51, 52]. None of compounds 1–3 produced a more negative docking score than the corresponding co‐crystallized reference ligand for AChE or BACE1 (Table 2). Among the isolated compounds, syringaresinol (1) showed the most consistently favorable ranking across the AChE, BACE1, and Aβ42 models. This ranking supports prioritization of compound 1 for subsequent enzyme‐based and biophysical testing but does not establish multitarget activity, selectivity, or biological efficacy.

Overall, the docking analysis revealed a consistent interaction profile for compounds 1–3 across the selected Aβ42, AChE, and BACE1 models (Table 2 and Figures 3, 4, 5). Syringaresinol (1) showed the most favorable and consistent docking scores across the target panel, supporting its prioritization for subsequent multitarget evaluation. Blumenol C glucoside (3), meanwhile, produced the greatest reduction in Aβ‐associated ThT fluorescence, highlighting a complementary activity profile among the isolated compounds. Together, these findings provide a rational basis for advancing compounds 1 and 3 to more detailed enzyme‐based, biophysical, and cell‐based studies, while recognizing that the predicted interactions require experimental confirmation.

2.6. Drug‐Likeness Predictions Based on Lipinski's Rule of Five

The preliminary drug‐likeness of compounds 1–3 was evaluated using Lipinski's rule of five. Syringaresinol (1), dehydrovomifoliol (2), and blumenol C glucoside (3) had molecular weights below 500 g/mol, fewer than ten hydrogen‐bond acceptors, fewer than five hydrogen‐bond donors, MlogP values below 5, and no Lipinski violations (Table 3). These findings indicate that the compounds satisfy commonly applied rule‐based criteria for oral drug‐likeness. However, Lipinski compliance does not establish gastrointestinal absorption, metabolic stability, toxicity, or central nervous system exposure. Thus, the present analysis supports only preliminary drug‐likeness, while CNS‐relevant pharmacokinetic and toxicity properties require separate computational and experimental evaluation.

TABLE 3.

Drug‐likeness predictions of compounds 1–3 based on Lipinski's rule of five.

	MW<500	#H‐bond acceptors < 10	#H‐bond donors < 5	Lipophilicity MlogP < 5	Lipinski violations	Drug‐likeness
Syringaresinol (1)	418.44 g/mol	8	2	0.56	0	Yes
Dehydrovomifoliol (2)	222.28 g/mol	3	1	1.05	0	Yes
Blumenol C glucoside (3)	372.45 g/mol	7	4	0.53	0	Yes

3. Conclusions

The methanolic crude extract of P. odorifer exhibited stronger antiglycation activity than radical scavenging activity, indicating that its bioactivity may involve mechanisms beyond conventional antioxidant effects. Although the extract showed moderate total phenolic content, its marked inhibition of AGE formation suggests the possible contribution of structurally diverse constituents, including lignans, norisoprenoid derivatives, glycosides, minor unidentified metabolites, or synergistic interactions among extract components. Phytochemical investigation of the extract led to the isolation of three known compounds, syringaresinol (1), dehydrovomifoliol (2), and blumenol C glucoside (3), whose identities were confirmed through spectroscopic analysis and comparison with published data. In the Thioflavin T assay, compounds 1–3 reduced Aβ‐associated fluorescence, with blumenol C glucoside (3) producing the largest response under the tested conditions. Because compound‐associated fluorescence interference and dye displacement were not independently excluded and no orthogonal method was performed, these findings remain preliminary and do not by themselves establish inhibition of Aβ fibril formation. Molecular docking further suggested possible interactions of the isolated compounds with Aβ42 oligomeric assemblies and AD‐related enzymes, particularly AChE and BACE1, with syringaresinol (1) showing the most favorable predicted binding profile across several targets. A preliminary Lipinski‐based assessment showed that compounds 1–3 satisfied the evaluated rule‐of‐five parameters, including molecular weight, hydrogen‐bond donor and acceptor counts, lipophilicity, and absence of Lipinski violations. However, these computational results should be interpreted only as predictive and hypothesis‐generating, rather than as direct evidence of binding affinity, selectivity, enzyme inhibition, or experimental drug‐likeness. Overall, this study supports P. odorifer as a promising source of antiglycation and anti‐amyloidogenic constituents relevant to AD‐related mechanisms. Further bioassay‐guided fractionation, compound‐level AGE inhibition assays, molecular dynamics simulations, binding free‐energy calculations, in vitro cholinesterase and β‐secretase inhibition assays, pharmacokinetic validation, and cell‐based Aβ‐toxicity models are needed to validate the biological significance of these findings.

4. Experimental Section

4.1. Materials and Instruments

Optical rotation data were obtained on a JASCO P‐2200 polarimeter (with a Na lamp). The UV and IR spectra were recorded on a JASCO V‐560 and a JASCO FT/IR‐4700 spectrophotometer, respectively. NMR spectra were recorded on a JEOL Ltd. JNM ECZ‐600 FT NMR spectrometer in CDCl₃ at 600 MHz for ¹H NMR and 150 MHz for ¹³C NMR, respectively. Preparative HPLC was carried out using a Shimadzu system equipped with a CBM‐20A communication bus module, DGU‐20A5R degassing unit, SPDM2OA diode array detector, FRC‐10A fraction collector, and FCV‐20AH2 valve unit. The chromatographic separation utilized an Inertsil Diol column (5 µm, 14 mm × 250 mm) with an Inertsil Diol PREP Guard Cartridge (30 mm × 7.6 mm) (GL Sciences Inc.). Additionally, an Osaka Soda CAPCELL Pak C18 column (5 µm, 20 mm I.D. × 150 mm) was used in conjunction with a Shimadzu Shim‐pack GIS(G) C18 guard column (10 µm, 20 mm × 50 mm). For open column chromatography, silica gel 60 (Merck, 230–400 mesh and 70–230 mesh ASTM) and Chromatorex NH (100–200 mesh) (Fuji Silysia Chemical, Ltd.) were employed. TLC was performed using Merck precoated Silica gel 60 F₂₅₄ aluminum‐backed plates.

4.2. Collection and Extraction of the Plant Material

The plant material was authenticated morphologically as P. odorifer (Forssk.) Kuntze by Prof. Dr. Cecilia Moran, Curator of the University of Santo Tomas Herbarium (USTH). A voucher specimen was deposited at USTH under Registry No. USTH‐014470.

4.3. Preparation of the Plant Material

Fresh P. odorifer leaves were air‐dried, weighed, and ground to obtain powdered samples (2 kg). The resulting powder was subjected to methanol (2.5 L) extraction using a percolator over a period of five days. The combined extracts were filtered and concentrated under reduced pressure to yield 80.3 g of crude methanolic extract. A portion of this extract was reserved for evaluation of antioxidant and anti‐AGE activities.

4.4. Determination of Total Phenolic Content (TPC) and Total Flavonoid Content (TFC)

The total phenolic content (TPC) of the methanolic extract was determined by the modified Folin–Ciocalteu method [53, 54]. Briefly, 80 µL of the sample or the gallic acid standard was mixed with 80 µL of distilled water in a microplate well. Then, 12 µL of 10% (w/v) Folin–Ciocalteu's phenol reagent and 125 µL of 7.5% (w/v) Na₂CO₃ were added sequentially to the mixture. Samples were incubated for 90 min at 25°C in a dark room and then subjected to spectrochemical analysis. The absorbance of the samples was measured at 750 nm against a blank sample. A calibration curve was constructed using gallic acid at concentrations of 0–1000 µg/mL, and TPC was expressed as mg gallic acid equivalents (GAE) per gram of dry extract. The sample extracts were evaluated at 1000 µg/mL.

Total flavonoid content (TFC) of the methanolic extract was assessed using a modified aluminum trichloride colorimetric assay [55]. A 50 µL extract solution or quercetin standard (0–1000 µg/mL conc) was mixed with 10 µL of 10% (w/v) AlCl₃ in methanol, 10 µL of 1 M potassium acetate, and 200 µL of distilled water. The samples were then incubated for 30 min at 25°C, followed by absorbance reading at 415 nm against the blank. Sample extracts were evaluated at a final concentration of 1000 µg/mL. The TFC was calculated from the quercetin calibration curve and expressed as mg quercetin equivalents (QE) per gram of dry extract, based on triplicate analysis.

4.5. DPPH Free Radical Scavenging Assay

The antioxidant activity of the methanolic extract was determined using the 2,2‐diphenyl‐1‐picrylhydrazyl (DPPH) radical scavenging assay, with slight modifications from the method described by Clarke et al. [56]. A 0.1 mM DPPH solution in methanol (180 µL) was mixed with the 20 µL of extract solutions, pre‐diluted in DMSO to concentrations ranging from 0.5 to 4.0 mg/mL. The reaction mixtures were incubated in the dark at room temperature for 15 min, after which the absorbance was measured at 540 nm using a microplate reader. DMSO was used as the blank, and ascorbic acid served as the reference standard. All assays were conducted in triplicate.

The percentage of DPPH radical scavenging activity was calculated using the equation:

$$\%\mathrm{DPPH}\mathrm{radical}\mathrm{inhibition}=\left[\frac{A_{\mathrm{c}}-A_{\mathrm{s}}}{A_{\mathrm{c}}}\right]\times 100$$

where A _c and A _s are the absorbance of the control and the sample extract. The half‐maximal inhibitory concentration (IC₅₀) was determined from the dose‐response curve generated by plotting % inhibition against extract concentration.

4.6. ABTS Radical Scavenging Activity

The antioxidant activity of the crude methanol extract was determined using the ABTS radical cation decolorization assay, following the method of Re et al. [57], with slight modifications. To generate the ABTS⁺• radical solution, 500 µL of 7 mM ABTS was mixed with 5.05 µL of 245 mM potassium persulfate and incubated in the dark at room temperature for 12–16 h. The resulting ABTS⁺• solution was diluted with methanol until the absorbance at 734 nm reached 0.70 ± 0.02. For the assay, 190 µL of the ABTS⁺• working solution was added to 10 µL of the extract (2 mg/mL in DMSO), Trolox standard, or control. After incubation at room temperature for 6 min, the absorbance was measured at 734 nm against a methanol blank. All measurements were performed in triplicate.

The percentage of inhibition of ABTS⁺• was calculated using the following equation:

$$\%\mathrm{ABTS}\mathrm{cation}\mathrm{radical}\mathrm{inhibition}=\left[\frac{A_{\mathrm{C}}-A_{\mathrm{s}}-A_{\mathrm{b}}}{A_{\mathrm{C}}}\right]\times 100$$

where A _c is the absorbance of the negative control (ABTS⁺• and methanol), A _s is the absorbance of the sample, and A _b is the absorbance of the blank (methanol). The half‐maximal inhibitory concentration (IC₅₀) was determined from the dose‐response curve generated by plotting % inhibition against extract concentration.

4.7. Advanced Glycation End (AGE) Products Inhibition Assay

The glycation reaction was carried out following the method described by Sharma et al. [58], with slight modifications. Briefly, the reaction mixture consisted of 100 mM phosphate buffer (pH 7.4) containing bovin serum albumin (BSA, 50 mg/mL), 0.5 M D‐glucose monohydrate, and 0.02% sodium azide. Extracts were incubated with this mixture at 37°C for 14 days in the dark. Aminoguanidine was used as the positive control, and quercetin was used as the reference compound. After the incubation period, advanced AGE formation was assessed by measuring fluorescence using a microplate multimode reader (Synergy‐H1 BioTek, Agilent Technologies, USA) at an excitation wavelength of 370 nm and an emission wavelength of 440 nm. All measurements were performed in triplicate. The percent inhibition of AGE formation was calculated using the following equation:

$$\%\mathrm{Inhibition}=\left(\frac{C-T}{C}\right)\times 100$$

where C is the fluorescence intensity of the control (without sample), and T is the intensity in the presence of the test sample. The half‐maximal inhibitory concentration (IC₅₀) was determined from the dose‐response curve generated by plotting % inhibition against extract concentration.

4.8. Solvent Extraction

The MeOH extract (80.3 g) was subjected to liquid‐liquid extraction using hexane (800 L), chloroform (1300 L), and butanol (600 L) to yield 0.16, 1.73, and 3.63 g extract, respectively (Figure S1). TLC analysis of the extracts revealed that the CHCl₃ extract contained both polyphenols and alkaloids, as indicated by color development with vanillin‐sulfuric acid and Dragendorff's reagents. The CHCl₃ extract was then fractionated on a silica gel open column using a stepwise gradient of CHCl₃–EtOAc (9:1→ 4:1→1:1→0:1), followed by an EtOAc–MeOH gradient (9:1→ 7:3→1:1→0:1), affording 13 fractions (PoCA–PoCM). Fraction PoCD (94.5 mg) was further separated on a silica gel open column using a hexane–EtOAc gradient (4:1 and 1:1), yielding 10 sub‐fractions (PoCD1–PoCD10). Fraction D7 (2.7 mg) was purified isocratically on a silica gel pencil column using 50% EtOAc in hexane to yield compound 1 (2.5 mg). Fraction D1 (49.0 mg) was chromatographed isocratically on a silica open column with 20% EtOAc in hexane to obtain fractions D1a (4.5 mg) and D1b (21.7 mg). Further isocratic elution (10% EtOAc in hexane) of D1b resulted in two sub‐fractions (Fr. D1b1–D1b2). Fraction D1b2 (15.7 mg) was purified on a silica pencil column using 20% EtOAc in petroleum ether, which afforded compound 2 (5.1 mg). Separation of fraction PoCL (349.2 mg) was subjected to NH open column using a CHCl₃–MeOH gradient (1:0→49:1→9:1→4:1→3:2→0:1), resulting in nine fractions (PoCL1–PoCL9). Fraction L4 was purified by preparative HPLC on a C₁₈ column (2 mL/min, H₂O–MeCN gradient, 3:2→0:1) to obtain three fractions (PoCL4a–PoCL4c). Finally, purification of fraction L4b was achieved by preparative HPLC on a diol column (2 mL/min, hexane–isopropanol gradient, 4:1→3:2→3:1→0:1) to obtain compound 3 (1.3 mg).

4.9. Compound Characterization

Compounds 1 and 2 were identified using UV, IR, ¹H and ¹³C NMR, and ESI‐MS data, together with comparison with published spectroscopic values. Compound 3 was assigned as blumenol C glucoside based on UV, IR, ¹H NMR, mass spectrometric data, and comparison with published data.

4.9.1. Syringaresinol (1)

White crystalline solid; UV (MeOH) λ _max (log ɛ): 206.5 (4.51), 274.5 (3.33); IR (ATR): 3408, 1610, 1514, 1458, 1426, 1327, 1211, 1107 cm⁻¹; ¹H and ¹³C NMR (600 and 150 MHz, respectively, CDCl₃), see Table S1 and Figures S3–S5, ESI‐MS: [M─H]⁻ 417.

4.9.2. Dehydrovomifoliol (2)

Colorless oil; UV (MeOH) λ _max (log ɛ): 232 (3.74); IR (ATR): 3426, 2963, 2931, 1660, 1362, 1249, 1178, 1122, 1023, 980 cm⁻¹; ¹H and ¹³C NMR (600 and 150 MHz, respectively, CDCl₃), see Table S2 and Figures S6–S8, ESI‐MS: [M+Cl]⁻ 257.

4.9.3. Blumenol C Glucoside (3)

Yellow amorphous solid; UV (MeOH) λ _max (log ɛ): 236 (3.56); IR (ATR): 2931, 1720, 1642, 1621, 1476, 1228, 1136, 1101, 1072, 1030, 974, 913, 885 cm⁻¹; ¹H NMR (600 MHz, CD₃OD), see Table S3; Figures S9 and S10, ESI‐MS: [M+H]⁺ 373.

4.10. Thioflavin T Assay

The effects of compounds 1–3 on Aβ‐associated Thioflavin T (ThT) fluorescence were evaluated using the SensoLyte Thioflavin T β‐Amyloid (1–42) Aggregation Kit (AnaSpec, Fremont, CA, USA; catalog no. AS‐72214), following the manufacturer's protocol with the indicated test‐compound conditions [59]. The kit contained pretreated monomeric human Aβ1–42 peptide, assay buffer, ThT, morin, and phenol red. Aβ1–42 was reconstituted by adding 1 mL of cold assay buffer to a vial containing 0.25 mg peptide, allowed to hydrate, mixed by inversion, and centrifuged at 10 000 rpm for 5 min at 4°C to remove insoluble material. The fibrillation reactions were prepared in nonbinding black 96‐well plates with a final volume of 100 µL per well. Each reaction contained 85 µL of the prepared Aβ1–42 solution, 10 µL of 2 mM ThT, and 5 µL of test compound, reference inhibitor, or vehicle. This corresponded to final concentrations of approximately 47 µM Aβ1–42 and 200 µM ThT. Compounds 1–3 and the reference compounds morin and phenol red were evaluated at 100 µg/mL. The plate was maintained at 37°C, and fluorescence was measured every 5 min with 15 s of shaking between readings. Fluorescence signals were recorded using a PerkinElmer VICTOR Nivo 3F multimode microplate reader at excitation and emission wavelengths of 440 and 484 nm, respectively. Control wells included Aβ1–42 without inhibitor, Aβ1–42 with morin or phenol red, vehicle controls, and assay‐buffer blanks. Percentage reduction in Aβ‐associated ThT fluorescence was calculated as follows:

$$\%\mathrm{inhibition}=\left[\left(1-I_{Fi}/I_{Fc}\right)\times 100\%\right]$$

where I _Fi (with inhibitor) and I _Fc (without inhibitor) are the fluorescence signals after correcting for the background of the ThT solution.

4.11. Computational Simulations

4.11.1. Ligand and Protein Preparation

The SMILES notations of the ligands or test compounds were generated using ChemDraw version 18.1 and subsequently optimized in Avogadro version 1.95 [60] to obtain MOL2 files. Protein structures acetylcholinesterase (PDB ID: 4EY6), BACE1 (PDB ID: 2XFJ), Aβ pentamer (PDB ID: 2BEG), and dodecamer (PDB ID: 2MXU), were obtained from the Protein Data Bank (PDB) (https://www.rcsb.org/). Post‐processing steps, including the removal of co‐crystallized ligands and energy minimization, were performed in UCSF Chimera version 1.14 [61] using the default parameters of the steepest descent algorithm [62].

4.11.2. Molecular Docking Simulations

Molecular docking simulations were performed using UCSF Chimera version 1.14 coupled with AutoDock Vina [63, 64]. Missing hydrogen atoms were added to the ligand and protein structures, and Gasteiger charges were assigned using the Antechamber module of Amber. The docking search spaces were defined using the grid‐center coordinates and dimensions reported in Table S4. Ligands were treated as flexible, allowing translational, rotational, and torsional sampling within the predefined active‐site grids, whereas the protein structures were maintained as rigid receptors. Ten binding modes were generated for each ligand–target combination using the Broyden–Fletcher–Goldfarb–Shanno optimization algorithm. The exhaustiveness and other parameters not otherwise specified were retained at the default AutoDock Vina settings. Molecular docking was used to compare predicted ligand orientations, residue contacts, and Vina docking scores within the selected target regions. The pose with the most negative Vina score was selected for interaction analysis using BIOVIA Discovery Studio version 4.1 [4, 65, 66, 67]. The molecular docking validation for acetylcholinesterase and BACE1 was performed following the same protocol used for ligand docking. The co‐crystallized ligand was first extracted from the crystal structure and then redocked into the active site of acetylcholinesterase under identical docking parameters. The resulting docked pose was subsequently superimposed with the original co‐crystal structure. Root‐mean‐square deviation (RMSD) between the heavy atoms of the two conformations was calculated using PyMOL. For both acetylcholinesterase and BACE1, the RMSD values were less than 2.0 Å For Aβ42 pentameric and dodecameric structures, ligand‐free oligomeric assemblies for comparative docking were used; therefore, classical native‐ligand redocking is not applicable to these models. All PDB structures and docking parameters used in this study were based on previous studies of Tan et al. [26] and Castro et al. [59].

4.11.3. Drug‐Likeness Predictions

The drug‐likeness profiles of the isolated compounds were evaluated through in silico prediction [68]. The SMILES notations of syringaresinol (1), dehydrovomifoliol (2), and blumenol C glucoside (3) were submitted to SwissADME to predict drug‐likeness based on Lipinski's rule of five [69].

4.12. Statistical Analysis

The percentage inhibition values were expressed as means ± standard deviations from triplicate experiments (n = 3). Statistical analyses were conducted using GraphPad Prism 8 version 8.0.2 (GraphPad Software Inc., San Diego, CA, USA). One‐way ANOVA followed by Tukey's multiple comparison test was performed, observing a statistical difference of p‐value < 0.05.

Author Contributions

Patricia Marie P. Oliva: methodology, data curation, software, investigation, formal analysis, and writing – original draft. Sarleen G. Castro: methodology, data curation, software, investigation, formal analysis, and writing – original draft. Joe Anthony H. Manzano: data curation, software, formal analysis, and writing – review and editing. Hayato Ishikawa: supervision, funding acquisition, and writing – review and editing. Mario A. Tan: conceptualization, data curation, formal analysis, supervision, funding acquisition, and writing – review and editing. All authors have read and approved the final manuscript.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File 1: cbdv71641‐sup‐0001‐SuppMat.pdf.

Data Availability Statement

The data that support the findings of the study are available in the Supporting Information.