In Vitro and In Silico Analyses Explore the Role of Flavonoid Classes in the Antiviral Activity of Plant Extracts Against the Dengue Virus

1. Introduction

Dengue virus (DENV) causes a spectrum of disease ranging from asymptomatic infection to dengue fever and severe dengue shock syndrome [1]. According to the Pan American Health Organization, dengue fever increased significantly in the Americas in 2024 rising by 232% compared to 2023 and 421% compared to the average of the previous five years [2]. There are currently no specific antiviral treatments for dengue, so therapy is primarily supportive [3]. Although two vaccines have been approved in some countries, their use is constrained due to safety concerns and their limited effectiveness against all DENV serotypes [4]. Dengue is a complex disease, making the discovery of effective therapies challenging. Increasing evidence suggests that herbal medicines have the potential as sources for clinical dengue treatments [5]. Administration of such medicines after virus exposure has shown to reduce the risk of severe dengue [5,6]. Thus, analyzing plant extracts has become an important strategy to discover and develop alternative therapies for dengue [5,7].

Plant-derived extracts prepared from a variety of species cultivated in different countries have demonstrated antiviral properties against human viruses, including DENV [5,6,7]. Their antiviral effect varies according to the plant species, chemotype, and phenological stages. The chemical composition of plant extracts depends on both the species and the extraction technique used which, in turn, affects their antiviral activity [7,8,9]. Diverse secondary metabolites, especially phenolic compounds such as flavonoids, have shown the potential to inhibit a variety of human pathogenic viruses, including influenza H1N1, HIV, and Zika [7,8,9]. Further research is necessary to fully elucidate chemical profiles that explain the anti-DENV activity of plant extracts. Studies demonstrate that plant-derived flavonoids have antiviral properties and can inhibit DENV by interfering with viral replication, entry, or specific viral proteins [10,11].

The genus Scutellaria (Lamiaceae) includes about 400 species widely distributed across America, Europe and Asia [12]. S. baicalensis is recognized as a promising natural source of compounds for treating viral diseases [13]. The genus Lippia (Verbenaceae) comprises 150 species, primarily found in the Americas, tropical Africa and India [14]. Lippia alba (Mill.) N.E. Brown was the first plant approved by the French Medicines Agency for inclusion in the French Pharmacopoeia [15]. L. origanoides Kunth was included in the Formulário de Fitoterápicos da Farmacopéia Brasileira due to its strong antibacterial properties [16]. L. alba and L. origanoides are traditionally used to treat viral infections such as influenza, measles and gastroenteritis [17,18]. Extracts prepared from S. baicalensis and L. origanoides have demonstrated inhibitory effect on the replication of DENV and other enveloped viruses [19,20]. We have investigated biological activities of both Lippia species and Scutellaria species cultivated in Colombia, and their extract chemical composition have been documented [20,21,22,23,24].

Dengue has been identified as a significant and persistent public health concern in Colombia, which is among one of the most affected countries in the Americas. From 2012 to 2020, annual incidence rate ranged from 90.7 to 476.2 per 100,000 population, with the most recent outbreak reported in 2019 (465.9 per 100,000 population) [25]. Traditional Colombian herbal remedies are widely used to alleviate dengue symptoms, although few plants have been systematically studied for their therapeutic potential [26]. Research on the anti-DENV activity of plant extracts contributes to discovering alternative therapies to prevent severe dengue. In this study, we evaluated the in vitro anti-DENV activity of extracts from Scutellaria species and L. alba and L. origanoides and investigated the influence of their flavonoid content. In silico analyses were also performed to hypothesize how flavonoid in the extracts might interfere with DENV entry into Vero cells.

2. Results

2.1. Cytotoxicity

Fourteen extracts from Scutellaria, L. alba, and L. origanoides were selected for analysis, including eight prepared in this study and six in previous studies [20,21,22]. The characteristics of the extracts are summarized in Table 1, categorized by plant species, chemotype, and extraction technique. Eleven extracts were prepared using ultrasound-assisted extraction (UAE) at different temperatures (50 °C and 47 °C) and extraction time (60, 23, and 15 minutes). Three extracts from L. origanoides were prepared using supercritical fluid extraction (SFE).

Cytotoxicity of the extracts was assessed in Vero cells uninfected with DENV using crystal violet assays. Assays were carried out at corresponding time points as the antiviral assay to determine non-cytotoxic concentrations for the cell. All extracts were non-toxic at 100 µg/mL, the maximum concentration used for antiviral assays (Figure 1A). Additional assays following the standard protocol determined the 50% cytotoxicity concentration (CC₅₀). Extracts from Scutellaria and L. alba, along with two extracts from L. origanoides (LopSE1 and LocSE2) reduced cell viability with CC₅₀ values ranging from 23 µg/mL to 213 µg/mL (Figure 1B). The extract from S. ventenatii + incarnata (SivSE) and one extract from S. incarnata (SiSE1) exhibited the highest cytotoxicity (CC₅₀: 23 and 50 µg/mL). The remaining six L. origanoides extracts showed no cytotoxicity at the maximum concentration tested, suggesting CC₅₀ values greater than 250 µg/mL.

2.2. Antiviral Effect

Virus-induced cytopathic effect (CPE) is a surrogate measure of virus replication in vitro; therefore, lower DENV-induced CPE indicates stronger antiviral activity [27]. The extracts were tested at six nontoxic concentrations to evaluate their effect on CPE caused by DENV-1 and DENV-2 after treatment during virus adsorption to Vero cells. The presence of DENV in untreated cells was confirmed by detection of non-structural protein 1 (NS1 = 75 ± 1.97 PanBio units) using an ELISA kit. Antiviral effects were classified as: (i) strong, greater than 50% reduction in CPE of both virus serotypes with IC₅₀ values lower than 55 µg/mL; (ii) weak, IC₅₀ ≤ 55 µg/mL for only one serotype or IC₅₀ between 55 and 100 µg/ml for one or both serotypes; and (iii) inactive, no significant CPE reduction in either serotype. Selectivity index (SI) was included in the analysis; extract with IC₅₀ ≤ 55 µg/mL and SI ≥ 3.0 was considered as promising anti-DENV sample. The extracts exhibited variable anti-DENV effect and selectivity (Figure 2 and Table 2).

Extracts from S. coccinea (ScSE) and S. incarnata (SiSE1) exhibited strong anti-DENV effect. ScSE showed IC₅₀ values < 53 µg/mL and SI values > 4.0, indicating promising anti-DENV sample. SiSE1 showed IC₅₀ values < 40 µg/mL, but SI values < 3.0, suggesting reduced antiviral promise. The other S. incarnata extract (SiSE2) was inactive, significant reduction of the virus-induced CPE of both serotypes was not detected. The S. ventenatti + incarnata (SviSE) extract exhibited weak antiviral effect, reduced DENV-2 at IC₅₀ of 65 µg/mL (SI of 0.4), but not DENV-1. Extracts from L. alba exhibited weak antiviral effect. LaiSE (citral chemotype) reduced the replication of both virus serotypes, but at IC₅₀ of 81 µg/mL against DENV-1. LacSE (carvone chemotype) showed antiviral promise against DENV-2 (IC₅₀ of 19 µg/mL and SI of 3.9), but was inactive against DENV-1. All eight extracts from L. origanoides were inactive against both virus serotypes.

2.3. UHPLC/ESI-Q-Orbitrap-MS Analysis of the Extracts Prepared in This Study

The analysis, based on retention times (Rt) and fragmentation patterns, identified only compounds belonging to the flavonoid class, including glycosides (n = 19), aglycones (n = 19) and methylated (n = 3) (Table 3 and Table 4).

Extracts from Scutellaria (UAE, 60 min at 50 °C) had higher concentrations of flavonoid glycosides than aglycones. Nine flavonoids were detected in S. incarnata (SiES2), with glycosides accounting for 64.7% (55-85 mg/g) of the total content. The predominant flavonoids were dihydrobaicalein-glucuronide, scutellarin, baicalin, baicalein, and wogonin. The flavonoid content of the S. ventenatti + S. incarnata (SviSE) extract was similar to that of SiES2, except for the absence of dihydrobaicalein-glucuronide and dihydrobaicalein. Flavonoid glycosides accounted for 55% (32/58 mg/g) of the SviSE content.

L. alba extracts (UAE, 60 min at 50 °C) had high flavonoid glycosides content, exceeding levels reported in extracts of the same species cultivated in Brazil [32], which may be attributed to variations in extraction methodology and cultivation practice of the plant. In the LaiSE (citral chemotype) extract were identified eight flavonoid glycosides accounting for 90.7% (16.6/18.3 mg/g) of total content. Tricin-7-diglucuronide, chrysoeriol-7-diglucuronide, chrysoeriol-7-glucuronide, tricin-glucuronide, and apigenin-7-glucuronide were predominant. Seven flavonoids aglycones were also identified, five from which were below detection limits. The LacSE (carvonal chemotype) extract exhibited a comparable profile, yet with higher amounts of flavonoid glycosides (97.1%, 20.7 / 21.3 mg/g).

L. origanoides extracts had higher flavonoid aglycone content than the other test extracts. Notable differences were observed in extracts from the carvacrol chemotype. LocSE1 (UAE, 23 min at 47 °C) contained twenty-four flavonoids, 82.3% (11.2/13.6 mg/g) of which were aglycones. Eriodictyol, naringenin, luteolin, quercetin, and cirsimaritin were predominant. The LocSE2 extract (UAE, 60 min at 50 °C) contained twelve flavonoids in higher amounts (259 mg/g vs. 13.6 mg/g), with aglycones accounting for 41.7%. The predominant flavonoids were eriodictyol, eriodictyol-7-glucoside, and luteolin-7-glucoside. The thymol chemotype LotSE (UAE, 23 min at 47 °C) extract contained twenty-three flavonoids. Aglycones accounted for 73.1% (42/57.4 mg/g), with eriodictyol, quercetin, and naringenin as the predominant. The phellandrene chemotype LopSE2 (UAE, 60 min at 50 °C) extract contained fourteen flavonoids, with aglycones accounting for 61.5% (228/371 mg/g) with eriodictyol, pinocembrin, and galangin as the predominant.

2.4. Relationship Between Anti-DENV Effect and the Flavonoid Content

Fourteen extracts listed in Table 1 were included in the analysis. The chemical compositions of six extracts that were analyzed in previous studies are shown in Table S1. The relationship between antiviral effect and the flavonoid content was investigated to test the hypothesis that extracts exhibiting anti-DENV effect (strong or weak) possess distinct chemical profiles compared to inactive extracts. Table 5 compares flavonoid classes identified in the fourteen extracts.

Extracts with the strongest anti-DENV effects (ScSE and SiSE1) contained high amounts of flavonoids (278 mg/g and 493 mg/g, respectively). Of these flavonoids, 83.4% and 86.1%, respectively, corresponded to flavone glycosides, with baicalin, dihydrobaicalein-glucuronide, and scutellarin being the predominant flavones. ScSE and SiSE1 also had small amounts of flavone aglycones (10.3% and 5%, respectively) and non-flavonoid compounds (6.4% and 10.5%), including verbascoside and umbelliferone-hexoside-pentoside (Table S1). In contrast, extracts with weak anti-DENV effects (SviSE, LacSE, and LaiSE) had significantly lower amounts of flavonoid (from 18.3 mg/g to 57.9 mg/g), though most of these were flavone glycosides (56.5%, 97%, and 90.7%, respectively). Among the extracts with anti-DENV effect, SviSE contained a significantly higher concentration (44.5%) of flavonoid aglycones, all of them belonging to the flavone class. The S. incarnata SiSE2 extract, lacking anti-DENV effect, had a chemical profile comparable to SviSE, except for the lack of dihydrobaicalein.

Extracts from L. origanoides lacking anti-DENV effect had the lowest concentrations of flavone glycosides (0 to 23.9%) but the highest of aglycones (61.5% to 98.1%), especially flavanone and flavanol + flavanonol, which were absent in extracts showing anti-DENV effects. Furthermore, UAE extracts (LopSE1, LopSe2, LocSE1, and LocSE) exhibited higher levels (17% to 34.4% vs. 0 to 4.6%) of flavonoid glycosides distinct to flavones, while SFE extracts lacked of flavonoid glycosides and contained small amounts of methylated flavonoids. The UAE-LopSE1 extract (phellandrene chemotype) had higher levels of flavonoid glycosides (77.4%) than aglycones (22.2%) compared to other L. origanoides extracts.

A Kohonen’s self-organized map algorithm, using percentages of flavonoid classes as input, grouped the fourteen extracts into two clusters (Figure 3A). Cluster 1 included five anti-DENV active extracts (Scutellaria and L. alba) along with the inactive S. incarnata extract (SiSE2). These extracts shared the highest flavone glycosides content and lowest flavonoid aglycones. Cluster 2 grouped seven inactive L. origanoides extracts marked by elevated flavonoid aglycones levels. The LopSE1 extract from L. origanoides was not included in cluster 2, as would have been expected, due to its distinct chemical profile. Figure 3B and 3C show that CPE values for DENV-1 and DENV-2 were significantly lower in cluster 1 compared to cluster 2 (Wilcoxon test, p < 0.01). This suggests a plausible relationship between the anti-DENV activity and a flavonoid profile characterized by a higher content of flavonoid glycosides, especially flavones, relative to flavonoid aglycones.

2.5. Molecular Interactions Between Flavonoids and DENV and Vero Cell Proteins

DENV infection of Vero cells is initiated by the interaction between its envelope (E) protein and specific cell surface receptors, including Axl receptor tyrosine kinase [34]. The virus enters cells via a non-classical endocytic pathway mediated by the clathrin and dynamin proteins [35,36,37]. To explore the role of flavonoids in the anti-DENV effect, an AutoDock Vina software was used to dock forty-six flavonoids found in fourteen extracts to the DENV-E, Gas6-Axl complex, clathrin, and dynamin. Binding energy values of ≤-7.55 kcal/mol indicate binding affinity between ligand and target. Figure 4 shows the relative concentrations of flavonoids in the extracts and their binding affinities to proteins. As expected, the analysis revealed varied binding affinities to targets. Generally, glycosides exhibited higher binding affinities (energy ranging from -10.22 to -8.01 kcal/mol) compared to aglycones (-8.23 to -7.90 kcal/mol), a difference that can be influenced by the presence of sugar. Most flavonoid glycosides exhibiting the lowest binding energies were the most abundant in extracts with anti-DENV effect.

The DENV-E protein structure consists of three domains [33], DI and DII domains form the DI/DII hinge region containing a conserved fusion loop. Conformational changes in this region lead to the fusion of the viral and cell endosomal membranes [37]. Nineteen glycosides (-9.16 to -7.72 kcal/mol) and fourteen aglycones (-8.06 to -7.55 kcal/mol) were predicted to bind to E. As for flavonoids found in extracts with antiviral effect, all flavone glycosides exhibited affinity (-8.78 to -7.72 kcal/mol) for E, except schaftoside. The lowest binding energies were exhibited by chrysoeriol-7-diglucuronide, chrysoeriol-7-glucuronide, luteolin-7-glucuronide, apigenin-7-glucuronide, and tricin-7-diglucuronide, all of which are predominant in L. alba extracts. All flavonoid aglycones, except cirsimaritin, exhibited affinity for E; the predominant baicalein, dihydrobaicalein, and apigenin showed the lowest energies (-7.96 to -7.84 kcal/mol). As for flavonoids present in extracts lacking antiviral effect (L. origanoides), 50% glycosides non-flavones (-8.08 to -7.73 kcal/mol) and 50% aglycones (-8.05 to -7.73 kcal/mol) exhibited affinity for E. Eriodyctiol-7-glucoside, eriodyctiol, and pinocembrin had the lowest binding energies. Flavonoids formed hydrogen bonds with amino acid residues at three consensus binding sites (Figure 5). Most glycosides were accommodated within the DII domain closer to the dimer interface (A/B), with the remaining glycosides located within the interface DIII/DI domains closer to the fusion loop. Most aglycones were accommodated within the detergent β-octyl glucoside (βOG) site in the hinge region, with the remaining aglycones located within the DII (A/B) interface.

DENV can bind to the Axl receptor via the Gas6 protein, which facilitates virus entry by bridging the viral envelope's phosphatidylserine to Axl [34]. Gas6 has two domains (Lg1 and Lg2) that interact with Axl domains (Ig1 and Ig2). Sixteen flavonoid glycosides were predicted to bind to Gas6 (-8.66 to -7.56 kcal/mol), whereas no aglycones (-7.52 to -6.83 kcal/mol) exhibited binding affinity. All flavone glycosides found in anti-DENV extracts exhibited affinity (-8.69 to -7.56 kcal/mol) for Gas6, except for schaftoside. The lowest binding energies were exhibited by chrysoeriol-7-diglucuronide, tricin-7-diglucuronide, luteolin-7-glucuronide, and luteolin-7-glucoside. In contrast, only two (eriodyctiol-7-glucoside and eriodyctiol-rhamnoside) from six glycosides non-flavones found in inactive extracts were predicted to bind to Gas6 (-7.77 kcal/mol and -7.76 kcal/mol, respectively). Flavonoid glycosides formed hydrogen bonds with amino acid residues at five consensus binding sites (Figure 6). Most flavonoids were located within the interface of Gas6-Lg2/Lg1 domains in which each of the monomers. The remaining flavonoids were accommodated within the Gas6-Lg1 domain, and the interface of the Gas6-Lg1/Axl domains within each of the monomers.

Dynamin possess a GTPase domain that binds to and hydrolyzes guanosine-5'-triphosphate (GTP), causing a conformational change crucial for its function in DENV and cell membranes fusion [39]. Small molecules targeting GTPase can interfere with virus internalization [39]. All twenty-three flavonoid glycosides (-9.96 to -7.68 kcal/mol), except quercetin-3-glucoside, were predicted to bind to GTPase. The glycosides with the lowest binding energies were chrysoeriol-7-diglucuronide, baicalin, isoliquiritin, apigenin-7-glucoside, and chrysoeriol-7-glucuronide. These flavonoids are found in anti-DENV extracts but not in inactive L. origanoides extracts. Seventeen aglycones exhibited binding affinity for GTPase (-8.26 to -7.73 kcal/mol). Baicalein, dihydrobaicalein, luteolin, and scutellarein had the lowest binding energies. The flavonoids formed hydrogen bonds with amino acid residues of five consensus binding sites (Figure 7). Most glycosides were accommodated at the G4 loop extending to the switch1 loop (G4-switch1 loops), while others were accommodated within the P-loop and a few at the switch1 or switch2, closer to the P-loop (switch1-P-loops and switch2-P-loops). Most aglycones were accommodated at the switch2-P-loops and G4-switch1-loops.

Clathrin is a transport protein whose N-terminal domain (CTD) has various peptide-binding sites, including the PWDLW motif (W-box) [38]. CTD has been proposed as a target for antiviral agents that hinder viral entry into the cell [40]. All of the flavonoid glycosides were predicted (-10.01 to -8.08 kcal/mol) to bind to CTD, except for quercetin-3-glucoside. The lowest binding energies were observed for chrysoeriol-7-diglucuronide, triicin-7-diglucuronide, apigenin-7-glucoside, baicalin, and scutellarin. All flavonoid aglycones except wogonin exhibited binding affinity for CTD (-8.22 to -7.75 kcal/mol). Luteolin, eriodictyol, hesperetin, nepetin, and baicalein had the lowest binding energies. Both glycosides and aglycones formed hydrogen bonds with amino acid residues in the W-box motif (Figure 8).

Molecular docking results for the predominant flavonoids of extracts from Scutellaria and L. alba are showed in Table 6. Results for the other flavonoids are shown in Table S1 and Table S2.

3. Discussion

This study analyzed fourteen plant extracts from Scutellaria and Lippia species cultivated in Colombia that were prepared under varying experimental conditions. Chemical analysis indicated that flavonoids constituted over 100% of their content. UAE-extracts exhibited significant differences in their chemical profiles. Retention of flavonoid glycosides was greater with shorter extraction times (5 or 23 minutes) than with longer times (15 or 60 minutes). This trend was evident in comparisons between SiSE1 vs SiSE2 as well as LopSE1 vs LopSE2 extracts. The reduction in flavonoid glycosides by prolonged extraction is likely due to excessive sonication that facilitate thermal, oxidative, and mechanical degradation, which particularly affect thermolabile glycosides and phenolic compounds [41,42]. Time-dependent variation in the extraction process critically influences the chemical profiles, and ultimately, the abundance of bioactive compounds [41]. SFE is a highly efficient technique for extracting aglycone-rich fractions with limited recovery of water-soluble or glycosylated phytochemicals [43]. The SFE extracts of L. origanoides contained high levels of flavonoid aglycones and lacked glycosides, reflecting the principle that nonpolar CO₂ extraction selectively enriches lipophilic compounds while excluding polar glycosides [43]. Chemical profiles of the extracts analyzed in this study differ from those reported for extracts of the same plant [19,28,29,30,32]. These variations can be explained by differences in growth stage and plant part, cultivation practice, and preparation technique.

The potential of crude plant extracts to inhibit DENV replication in vitro has been widely documented [5,6,7]. However, relationships between their chemical composition and anti-DENV effect remain to be elucidated. This study integrates UHPLC/ESI-Q-Orbitrap-MS data with CPE-DENV reduction data showing that higher contents of flavonoid glycosides, predominantly flavones, correlate with greater inhibitory effect on DENV replication in Vero cells. Moreover, increased aglycones contain is associated with lack of antiviral action, as evidenced by the extracts from L. origanoides. The distinction between S. inarnata extracts is notable: SiSE1 showed strong antiviral effect, whereas SiSE2 had none. SiSE2 contained lower flavonoid glycoside content (64.8% vs. 88.6%) and sevenfold higher aglycones (35.2% vs. 5.0%). In a previous study, we compared the anti-DENV effects of LopSE1 (77% glycosides) and LopSFE (no glycosides) extracts from L. origanoids. After treatment during DENV-1 adsorption on human hepatic cells, the level of viral NS1 protein was significantly reduced by LopSE1, but not by LopSFE. The flavonoid content is a determining factor in the antiviral action of plant extracts on virus others than DENV [8,44]. Glycosylation may enhance flavonoid interactions with virus particles or cell surface receptors by increasing polarity or specific binding conformations relative to aglycones [44]. A study revealed that baicalin exhibited higher anti-DENV activity than baicalein in virus-infected cells [45]. Other study revealed that myricetin and its glycosides exhibited higher antiviral activity against HIV-1 virus compared to their aglycone counterparts [46].

While mechanisms underlying plant extract antiviral activity are yet to be fully elucidated, flavonoids are widely accepted as primary agents [10,11]. Certain flavonoids can interfere with viral adsorption to host cells through the attachment and subsequent inhibition of proteins and molecules involved in the virus endocytosis [46,47]. The antiviral assay in this study evaluated the potential of the extracts to interfere with virus adsorption to cells. Reduction of DENV-CPE by Scutellaria and L. alba extracts suggests their flavonoids may block virus-cell membrane interactions. In silico analysis support this. We selected the envelope E protein of DENV, which plays an important role in the process that allows the virus to enter the cell [33]. Flavonoid glycosides from these extracts exhibited the strongest affinity for amino acids of the DII domain that interacts with receptors to promote viral entry into cells [33]. In addition, flavonoid aglycones bound to the βOG biding site, which has been established as a target for developing antivirals for dengue [48]. Numerous flavonoids have been identified as ligands for the DENV-E protein [11,49].

DENV uses a variety of cell receptors to gain entry into Vero cells [35,36,37]. The Gas6-Axl complex was selected for the in silico analysis. Flavonoid glycosides from anti-DENV active extracts showed good binding affinities to Gas6, suggesting potential biological interference with this receptor, whereas none of the flavonoid aglycones bound effectively. Flavonoid glycosides may have prevented DENV virions from adhering to Vero cells by blocking the PtdSer-Gas6 binding step. Disrupting the PtdSer-Gas6-Axl complex has been proposed as a potential therapeutic approach to inhibit DENV replication [36]. Flavonoids may also have entered the cells and affected the expression of Gas6 and Axl. Luteolin can downregulate the expression of these proteins in human cells [50].

Dynamin-dependent and clathrin-mediated endocytosis pathways are involved in DENV entry into Vero cells [33,34,35]. Targeting these pathways has been proposed as a strategy for developing antivirals [40,51]. The N-terminal domain (CTD) of clathrin and the GTPase domain of dynamin were selected for in silico analysis. Once again, flavonoid glycosides present in extracts with anti-DENV effect exhibited the best binding affinities with CTD and GTPase. Baicalein and scutellarin, abundant flavonoid aglycones in Scutellaria extracts, had the best binding affinity to both proteins. The flavonoids may have destabilized the W-box domain of clathrin-CTD and reduced the capacity of dynamin to exchange GDP for GTP. This, in turn, may have affected the functionality of both proteins, consequently impacting the endocytosis process of DENV in Vero cells. Also, it is plausible that the flavonoid treatment altered the fluidity of both the viral envelope and the cell membrane, thereby affecting their fusion process during virus endocytosis into the cell. The capacity of flavonoids to affect cell membrane fluidity (raft-breaking effect) has been documented [52].

4. Materials and Methods

4.1. Reagents

HPLC-grade acetonitrile, HPLC-grade formic acid (FA), isopropanol (98%), ammonium formate (AF, ≥99%), LC/MS-grade methanol, and potassium persulfate (≥98%) were obtained from Merck (Darmstadt, Germany). Standard substances (e.g., eriodictyol-7-glucoside, salvigenin, galangin, scutellarin, apigenin-7-glucuronide) were purchased from Sigma-Aldrich (St. Louis, MO, USA), Chemfaces (Wuhan,China) and Phytolab GmbH (Vestenbergsgreuth, Bavaria, Germany). MO, USA). Type I water was obtained from a Millipore Direct-QTM purification system. Eagle’s Minimum Essential Medium (MEM), fetal bovine serum (FBS), phosphate buffered saline (PBS), and antibiotics were purchased from Gibco (Grand Island, NY).

4.2. Viruses and Cell

Dengue virus type 1 (DENV-1) strain US/Hawaii/1944 and Dengue virus type-2 (DENV-2) New Guinea C strain (NGC) were used. Viruses were propagated titrated in BHK-21 cells as in a previous study [53]. African green monkey kidney (Vero) cell line (ATCC® CCL-81™) was cultured in Eagle’s Minimum Essential Medium (EMEM) containing 10% fetal bovine serum (FBS, Gibco, NY, U.S.A.) at 37 °C in the presence of 5% CO₂.

4.3. Extracts

Plant material was provided by the Colombian government through Contract No. 270 for Access to Genetic Resources and Derived Products between the Ministry of Environment and Sustainable Development and the Industrial University of Santander. The plants used in this study were grown in the experimental plots at the Agroindustrial Pilot Complex of the National Center for Agroindustrialization of Aromatic and Medicinal Tropical Vegetables (CENIVAM) in the Industrial University of Santander (Bucaramanga, Colombia). The plants were dried in the dark, and leaves and stems were crushed and homogenized. Voucher specimens were deposited in the Herbarium of the Industrial University of Santander. Fourteen extracts were included in the analysis of their anti-DENV effect. Eight extracts were prepared in this study using the UAE technique following the protocol used in previous studies [20,22]. Briefly, dried leaves and stems (100 g) were crushed and the material was mixed with ethanol:water (20 mL; 70:30), the mixtures were subjected to ultrasonication (35 kHz, Elmasonic S15H, Singen, Germany) and filtered (Wattman No. 1 paper), and the residue was extracted again with ethanol:water (10 mL). Temperature of 50 ^oC for 60 min was used. The extracts were vacuum roto-evaporated in a Heidolph apparatus, dried in a VirTis AdVantage Plus tray lyophilizer, and stored at 4 ◦C in the absence of light. Six extracts were prepared in previous studies using UAE [20,22] or SFE techniques [21]. Each extract (1x10⁵ µg/mL) was dissolved in 1% dimethyl sulfoxide (DMSO) and stored at -20 ^◦C before analysis.

4.4. UHPLC/ESI-Q-Orbitrap-MS Analysis

UHPLC-Q-Orbitrap-MS/MS analysis was used for identifying and characterizing the chemical components of the eight extracts prepared in this study, in accordance with the protocol used in a previous study [22]. An Ultimate Dionex^TM 3000 UHPLC (Thermo Fisher Scientific, Bremen, Germany) connected to an Orbitrap™ mass detector Exactive Plus, Bremen, Germany), with with heated electrospray ionization (HESI-II; Thermo Fisher Scientific). operated in positive-ion acquisition mode (350 ^oC). Chromatographic separation was performed on 50 mm L × 2.1 mm I.D. Zorbax Eclipse XDB C18 column, 1.8 µm particle size (Sigma-Aldrich, St. Louis, MO, USA). The mobile phase was a mixture of 0.2% formic acid-water (A) and 0.2% formic acid-acetonitrile (B) at a flow rate of 0.3 mL/min. The initial gradient condition was A-100%, which linearly changed to B-100% in 8 min, held for 4 min, returned to A-100% in 1 min, followed by 3 min of equilibration. The injection volume was 2 µL. Mass spectrometry experiments were accomplished on a Q-Exactive Plus Orbitrap mass spectrometer (Thermo Scientific, Bremen, Germany) connected to a the UHPLC and equipped with an electrospray ionization (ESI) source. Samples were analyzed in negative ion modes. The parameters were as follows: mass range, m/z 80 to 1000; electrospray ionization temperature, 350 °C; capillary temperature, 320 ◦C; capillary voltage, 3.5 kV; higher-energy-collisional dissociation cell (HCD), 10- 40 eV range. Data were analyzed using Thermo Scientific™ Dionex™ Chromatography Data System (CDS) software, version 7.2 and Thermo Xcalibur 3.1 software (Thermo Scientific, CA, USA). Extract metabolites were identified by comparing their retention times (tR), exact ion masses, isotopic ratios and fragmentation patterns with those of standard substances and databases [31,54]. All samples were run in triplicates and the data were presented as the mean value ± standard deviation (SD) of mg per g of freeze-dried extract.

4.5. Crystal Violet Assays

The assay evaluated the effect of the fourteen extracts on the viability of non-DEN-infected Vero cells. Two different protocols, each including seven extract concentrations (3.12 to 250 µg/mL), were followed. Protocol 1: cell monolayers were treated with extract for 1.5 h, the extract was removed by washing with PBS, fresh medium MEM with 2% fetal bovine serum was added, and the plates were incubated at 37 ºC; 5% CO₂ for five days. Culture supernatants were discarded and crystal violet solution (0.05%) was added followed by methanol, and then the plates were then analyzed using an ELISA plate reader at a wavelength of 570 nm. Untreated cells and cells treated with DMSO were run in parallel as control negative and positive, respectively. The higher concentration of extract that did not reduce cell viability respect to untreated control was calculated by regression analysis (GraphPad Software, San Diego, CA, USA). Protocol 2: cell monolayers were incubated in culture medium containing extract for 72 h, the extract was then removed by washing, and the cell viability was measured as in Protocol 1. Untreated cells and cells treated with DMSO were included as controls. The CC₅₀ values were calculated by regression analysis as in previous studies [20,53].

4.6. Cytopathic Effect (CPE)-Based Antiviral Assay

Extracts were analyzed in the CPE-based antiviral assay according to the protocol used in a previous study [53]. DENV (1 PFU/cell) was adsorbed on Vero cells grown in 96-weel plate for 1.5 h at 37 ◦C; 5% CO₂ in the presence of plant extract at non-cytotoxic concentrations (3.12 to 100 µg/ml); then, the cells were washed with PBS and fresh media without extract was added to allow virus replication. After five days of incubation at 37 ºC, the culture supernatant was discarded and cell viability measured using the crystal violet assay. Untreated DENV-infected cells and virus-infected / SDS-treated cells were included as negative and positive controls, respectively. DENV CPE (%) was calculated as follows: [(OD₅₇₀ of infected treated cells / OD₅₇₀ of uninfected untreated cells) x 100]. DENV CPE reduction was calculated as follows: DENV CPE reduction (%) = [(OD₅₇₀ of infected treated cells - OD₅₇₀ of negative control) / (OD₅₇₀ of positive control - OD₅₇₀ of negative control) × 100]. The concentration of extract that inhibited DENV-CPE by 50% (IC₅₀) was calculated by nonlinear regression followed by the construction of a concentration-response curve (GraphPad Prism software version 8.0, San Diego, CA, U.S.A.). Selectivity index (SI = CC₅₀ / IC₅₀) values were calculated. Each extract was analyzed in triplicate in three independent assays.

4.7. Molecular Docking Analysis

Three-dimensional structures of target proteins were downloaded from the Protein Data Bank. DENV-2 E protein (PDB ID: 10AN), Gas6-Axl receptor (PDB ID: 2C5D), clathrin (CDT domain, PDB ID: 2XZG), and dynamin (GTPase domain, PDB ID: 2X2E). Structures of flavonoids were retrieved from the PubChem (https://pubchem.ncbi.nlm.nih.gov/ database (accessed on September 2024). The preparation of the target and ligands and molecular docking analyses were carried out using AutoDock Vina (Version 1.5.6, La Jolla, CA, USA), as described in previous studies [53,54]. The optimized protein structure was saved in the PDBQT file format. Default parameters were used, and the search exhaustiveness parameter was set to 100. Twenty five docked conformations were generated for each ligand using docking simulations. Three simulations were performed for each ligand–protein pair using seeds 6, 12, and 18. The ligand–protein interactions was displayed using the Discovery Studio Visualizer v21.1.0.20298 software.

4.8. Statistical Analysis

The relationship between antiviral activity and the flavonoid content was analyzed using two independent parameters: the chemical composition, represented as the absolute percentage (%) of different flavonoid classes (Table 5), and the antiviral effect, DENV-CPE (%) as a unidimensional variable ranging from 0 to 100. The fourteen extracts were clustered based on to their content of flavonoids and other compounds, using an unsupervised self-organized Kohonen Map [55]. One-way ANOVA and a Tukey-Kramer post hoc tests compared DENV-CPE (%) values between clusters, adopting a significance level of 0.05. The analysis was performed using the Matlab^® R2021b software.

5. Conclusions

This study reports for the first time the correlation between the antiviral efficacy of extracts from medicinal plants cultivated in Colombia and their flavonoid content. Integration of antiviral assays, UHPLC analysis, and molecular docking analysis revealed that the content in flavonoid glycosides is a key determinant of the extracts’ anti-DENV effect. Extracts rich in flavonoid glycosides, particularly flavones, demonstrated the greatest antiviral efficacy. In contrast, extracts with lower levels of glycosides or lacking them did not show antiviral activity. Docking analysis suggests that the flavonoids present in the extracts could block the interaction between DENV particles and the cell membrane via different mechanisms. Based on IC₅₀ and SI values from the CPE-based antiviral assay, the UAE-extract from S. coccinea can be classified as a prospective anti-DENV sample. This extract contains high concentrations of flavonoid glycosides with potential to inhibit DENV, and therefore could serve as a starting point for research on herbal medications for dengue treatment. Considerable evidence demonstrates the potential of Scutellaria species as an antiviral agent against various pathogenic viruses, including DENV.