Characterization and enhanced antibiofilm activity of Annona muricata extract in combination with fluconazole against Candida albicans

Introduction

Seventy-five percent of human microbial infections are caused by the growth and persistence of biofilms, which are surface-attached microbial populations encased in a self-synthesized polymeric matrix (1). The ability of Candida albicans to form biofilms is a crucial component of its pathogenicity since it can increase tolerance to the host immune system and traditional antifungal medication (2). Most severe and recurring infections caused by C. albicans are linked to the development of biofilms on natural or synthetic surfaces (3). C. albicans biofilm has been known to cause persistent infections of organs and tissues invasively via seeding disseminated bloodstream infections, known as candidemia (4,5). Interestingly, secreted aspartyl proteinases (Saps) are among the hydrolytic enzymes that contribute significantly to the pathogenicity of the opportunistic pathogen C. albicans (6). It is believed that mucosal infections are associated with Sap 1-3, while systemic infections are associated with Sap 4-6 (7).

Herbal remedies have been utilized for millennia to cure a wide range of illnesses, both in Africa and elsewhere. This is mostly due to the perception that medicinal herbs are more accessible, more affordable, and more effective than Western medications (8). Annona muricata Lin. (Custard-apple, Annonaceae) is endemic to the West Indies and Central America, where it grows widely at elevations below 900 meters above sea level. It is also known by the names guanabana, paw-paw, soursop, and graviola. It grows in nations such as India, Angola, Puerto Rico, Brazil, Costa Rica, Colombia, and Venezuela that have tropical or subtropical climates (9). It has proven possible to isolate >200 bioactive chemicals from the A. muricata. Terpenoids, phenolics, and alkaloids are the most identified secondary metabolites of A. muricata (9-11). In Africa, India, and tropical America, A. muricata is frequently used as a folk remedy for a variety of human illnesses, including diabetes, rheumatism, cancer, and parasitic infections (12). The antibacterial, antifungal, anticancer, anticonvulsant, sedative, antiparasitic, and cardio-depressant properties of A. muricata leaves are just a few of its many potential uses (13,14). According to Rustanti and Fatmawati, A. muricata leaves ethanol extract showed antifungal activity, particularly against C. albicans (13). However, the literature presently has inadequate proof of the use of A. muricata leaf extract on biofilms of C. albicans.

Thus, this work aims to examine the antifungal activity and synergistic potential of the methanol extract of A. muricata leaf when combined with an antifungal agent, fluconazole, against C. albicans biofilm. Also, the existence of putative chemical components of A. muricata leaf extract, as well as the possibility of an interaction between identified compounds of A. muricata leaf and the Sap3 C. albicans, will be investigated to find new inhibitor candidates.

Experimental

Extraction procedure

Briefly, properly washed, cleaned, and dried (21 days at room temperature), leaves were ground into fine powder by using an electrical grinder. Powdered leaves (1000 g) were extracted successively (2.5 L × 3) by macerating with n-hexane, dichloromethane, ethyl acetate, methanol, and water at room temperature for 72 hours to obtain four respective extracts. The extracts were filtered through cotton wool and then dried in a rotary evaporator set at 40°C. To remove any remaining moisture, the extracts were moved to sample bottles and put in a desiccator with anhydrous sodium sulfate. The dried extracts were then refrigerated at −20°C in t-stopped bottles for further analysis (15). According to the studies, polar solvents are better than non-polar solvents, most likely because they offer a higher phytocompound recovery yield (16,17). Therefore, the adoption of methanol extract for additional research was motivated by the increased extraction yield.

LC-MS parameters for secondary phytochemical analysis

An ABSCIEX 4000 QTRAP hybrid triple quadrupole ion trap mass spectrometer was used to analyze the samples, and its front end was a Shimadzu HPLC stack. The software Analyst 1.5 (AB SCIEX) was used for all data processing and collecting. A 20 μL sample was separated using a 10-minute gradient that started with 5% solvent A (H₂O containing 0.1% ammonium formate) and gradually progressed to 10% solvent B (MeOH with 0.1% ammonium formate). This separation was conducted on a C18 column (150 × 4.6 mm, Discovery, Supelco) at a flow rate of 0.5 mL/min. The solvent gradient was gradually increased to 30% B over the next 10 minutes, then to 50% B for an additional 10 minutes, and finally to 95% B over the subsequent 15 minutes. The total runtime for the experiment was 60 minutes, which allowed for column re-equilibration. Analytes that were eluted were electrosprayed into the TurboV ion source at 500°C to evaporate any excess solvent. The system used a nebulizer gas pressure of 30 psi, a heater gas pressure of 30 psi, and a curtain gas pressure of 20 psi. The ion spray voltage was adjusted to 5500V in positive ionization mode and −4500V in negative ionization mode. With fixed declustering and entrance potentials of 40 and 10V, respectively, the eluting analytes were mass measured on the mass spectrometer in Q1 scan mode, covering a range of 100 Da to 2000 Da for a 3-second cycle period.

Multiple optimizations and scans were done to reduce the background noise and come up with the best baseline stabilizations for accurate compound identifications.

The compounds were identified by comparing the outcomes of ESI-MS/MS observations articulated as tR and fragmentation configurations with those disclosed by investigations obtained from various database systems, such as the National Library of Medicine, Mass Bank of Europe, and related literature reviews. MS spectra, TIC, and possible identified compounds with their corresponding structures were retrieved from the library as well as the PubChem Compounds database.

Docking Studies

Using the Lamarckian genetic algorithm (LGA) as the scoring function, the PyRx virtual screening program was utilized to virtually screen ligands, including AutoDock and AutoDock Vina. Molecular docking analysis was used to determine the chemical compounds’ binding affinity with the examined 2H6T protein’s active site amino acids. The SDF format of Sap3 (PDB ID: 2H6T) protein 3D structure was downloaded from the protein data bank (PDB). The AutoDock tool was used to import the proteins with access codes 2H6T. After the water molecules attached to the structures were eliminated, hydrogen atoms were inserted. Similarly, LCMS-identified compound structures were also downloaded from the PubChem database. The ligands were docked using the active site dimensions, which were defined as a grid size based on the XYZ axis. A maximum exhaustiveness of 10 was computed for every ligand. Prior to the docking process starting, AutoDock Vina applied charges to the structures of proteins and ligands (15,18). The 2D visualization of the ideal binding pose—which is the binding energy with the lowest value for each compound-protein combination—was studied using BIOVIA Discovery Studio software. Hence, there is a great chance that the ligands produced will be employed as medication candidates.

Antibiofilm activity

XTT assay of biofilms

On YM agar plates, C. albicans cells were cultured and subsequently incubated for 24 hours at 30°C. Following incubation, 20 mL of yeast nitrogen base (YNB) glucose medium (10 g/L glucose, 6.7 g/L YNB) was filled with a loop full of cells, and the mixture was incubated for 24 hours at 30°C.

Using an Eppendorf centrifuge 5430R, cells were harvested and washed (3075 g × 3 for 5 minutes at 4°C) with phosphate-buffered saline (PBS) and resuspended in 10 mL of RPMI-1640 media (Sigma-Aldrich, UK) at a concentration of 1x10^–6 cells/ml. Aliquots of 200 µL of the cell suspension, including A. muricata methanol extract (reconstituted in sterile water) (AM), fluconazole (FLU) and a combination of the extract and antifungal drug (AM + FLU) at final concentrations ranging from 15-240 µg/ml was dispensed into a 96-well microtiter plate (Corning Incorporated, Costar®, U.S.) and incubated for 48 hours at 37°C to allow biofilm formation. Standardized cells with RPMI-1640 media were used as the negative control. To get rid of non-adherent cells, wells were twice cleaned with 200 µL PBS after 48 hours. Kuhn et al.’s (19) method of assessing the mitochondrial metabolic activity of the biofilms was used for the reduction of 2,3-bis(2-methoxy-4-nitro-5-sulfophenyl)-5-[(phenylamino)carbonyl]-2H-tetrazolium hydroxide (XTT) (Sigma Aldrich, UK) in order to assess the viability of the yeast.

In the cellular supernatant, XTT was converted to colored, diffusible, water-soluble formazan, whose optical density was easily measured at 492 nm. Three biological duplicates of each three technical duplicates were used for this experiment. %Inhibition was calculated according to the following formula:

After calculating the average and standard deviations, the student’s t-test was performed to assess the significance of the data sets, with P < 0.05 being deemed significant. IC₅₀ values were determined by performing non-linear fitting analysis using GraphPad Prism 8.0.2 software.

Scanning electron microscopy (SEM)

Biofilm was prepared as described above in a flat bottom 6-well plate (Greiner Bio-One, Germany) in 2 mL of RPMI-1640 medium containing either FLU (15 µg/mL), AM (15 µg/mL) or a combination of FLU (7.5 µg/mL) and AM (7.5 µg/mL) on sterile polymer discs (Isopore 0.2 µm hydrophilic polycarbonate membrane disc, Merck, Germany). The polymer discs were removed aseptically after incubation and left overnight in a primary fixative solution of 3% (v/v) glutardialdehyde (Merck, Germany) in phosphate buffer (pH 7.0). The biofilms underwent two PBS washes before being fixed for two hours at room temperature using 1% (v/v) osmium tetroxide (Merck, Germany) as a secondary fixative. This was followed by a second wash phase. The biofilms were air dried in a desiccator after being dehydrated in an ethanol series (50% for 20 minutes, 70% for 20 minutes, 95% for 20 minutes, and 100% for one hour – twice). Then, they were subsequently coated with gold using SEM coating equipment (EM ACE600, Leica, Austria) and exposed to critical point drying (Samdri®-795 Critical Point Dryer, Tousimis, United States of America) for 30 minutes. Biofilms were analyzed using a JSM-7800F field emission SEM (ZEISS, Germany).

Results and Discussion

LCMS profile of A. muricata methanol extract

A total of 17 phytochemicals were detected by LC/MS and 14 among them were identified (Table 1). These included alkaloids (asimilobine 2, swainsonine 3, xylopine 7, (+)-4’-O-methylcoclaurine 10), flavonoids (isovitexin 4, vitexin 5, rutin 8, quercetin-3-O-D-glucuronide 9, kaempferol 3-O-rutinoside 11, eriocitrin 15, apigenin 6,8-di-C-glucoside 17) and acetogenins (annonisin 13, annomuricin A 14, montanacin B/C 16) (Fig. 1) from AM (Fig. 1). A potential identification of the chemicals was predicated on contrasting the fragmentation patterns and retention times (tR) obtained from ESI-MS/MS experiments with those documented in research gathered from various databases.

The first-order mass spectrum of compound 3 showed an intense protonated molecular ion [M+H]⁺ at m/z 174.3 and also showed a single loss of one hydroxyl group as water [M+H-H₂O]⁺ at m/z at 156.2, which was identified as swainsonine (20-22). A similar trend was observed for compound 2, where a protonated molecular ion [M+H]⁺ was observed at m/z 268.6. The fragments at m/z 250.6 and 236.5 were assigned to the loss of the hydroxyl group as water [M+H−H₂O]⁺ and methoxy group [M+H−OCH₃]⁺, respectively, and this led to the identification of compound 2 as asimilobine and this was consistent with the results of Lima et al. and Mohanty et al. (23,24). Compound 7 was identified as xylopine based on the typical fragment ions at m/z 279.4 ([M + H−OH]⁺) and 249.7 ([M−CH₂O₂]⁺) (22,25,26). For compound 10, the main fragment ion appeared at m/z 300.4 [M+H]⁺ and another main fragment at m/z 283.9 [M−OH]⁺, which corresponded to the loss of the hydroxyl group. Moreover, other fragments at m/z 251.8, 178.3, and 121.1 were observed, which led to the identification of compound 10 as (+)-4’-O-Methylcoclaurine (27).

Peak	Compound name	tR (min)	m/z	Ionization mode	Molecular formula	References
1	Unidentified	3.585	161.0	[M-H]–	–	–
2	Asimilobine	4.749	268.6	[M+H]+	C8H15NO3	(24)
3	Swainsonine	9.286	174.3	[M+H]+	C17H17NO2	(20-22)
4	Isovitexin	12.663	431.0	[M-H]–	C21H20O10	(32)
5	Vitexin	13.832	431.0	[M-H]–	C21H20O10	(32)
6	Unidentified	16.112	407.0	[M-H]–	–	–
7	Xylopine	18.298	296.2	[M+H]+	C20H21NO4	(22,25,26)
8	Rutin	17.381	609.7	[M-H]–	C27H30O16	(30)
9	Quercetin-3-O-D-glucuronide	17.429	477.1	[M+H]–	C21H20O13	(33)
10	(+)-4’-O-Methylcoclaurine	16.889	300.4	[M-H]+	C18H21NO3	(27)
11	Kaempferol 3-O-rutinoside	19.900	595.3	[M+H]+	C27H30O15	(28,29)
12	Unidentified	22.840	517.4	[M+H]–	–	–
13	Annonisin	30.812	611.3	[M+H]+	C35H62O8	(36)
14	Annomuricin A	31.356	613.3	[M+H]+	C35H64O8	(38,39)
15	Eriocitrin	33.166	595.7	[M+H]–	C27H32O15	(31,34)
16	Montanacin B	38.016	611.2	[M+H]+	C35H62O8	(37)
17	Apigenin 6,8-di-C-glucoside	38.177	593.4	[M+H]–	C27H30O15	(31)

The mass spectrum fragmentation of compound 11 appeared at m/z 595.3 [M+H]⁺. The product ion spectrum of the compound 11 ion (m/z 595.3) showed fragment ion at m/z 577.4 [M+H−H₂O]⁺ due to loss of water molecule. The fragments at m/z 449.4 and 287.4 indicated the loss of rhamnose and glucose sugar units, and therefore, compound 11 was identified as kaempferol 3-O-rutinoside (28,29). Similarly, a deprotonated molecular ion [M-H]^– was observed at m/z 609.7 for compound 8. The spectra showed the presence of the main fragments at m/z 463.2 [M-H-146]^– and 301.1 [M - H-146-162]^–, indicating the loss of two sugar units consisting of rhamnose and a pyranose, and this led to the identification of compound 8 as rutin (30). Compound 17 showed a pseudo molecular ion peak at m/z 593.4 [M−H]^-. MS showed fragment ions at m/z 431.1 [M−H-162] and 269.0 [M−H-162], corresponding to the presence of the hexose nature of C-glycoside flavone. Moreover, from the observed base peak fragment at m/z 269.0, it could be concluded that compound 17 was apigenin 6,8-di-C-glucoside (31). Similarly, compounds 4 and 5 were assigned as isovitexin and vitexin, respectively (32).

Compound 9 gave a pseudo-molecular ion at m/z 477.1 [M-H]^–. The MS/MS spectrum featured characteristic ion of m/z 301.1 [M-H-178]^–, derived from the loss of glucuronide, and this led to the identification of compound 9 as quercetin 3-O-glucuronide (33). For compound 15, the LC-MS chromatogram yielded an intense peak at m/z 595.7 [M-H]^–. Moreover, another main fragments at m/z 449.2 [M-H-146]^– and 287.0 [M-H-162]^– were observed, which implies the loss of deoxyhexose and hexose as rhamnose and a pyranose, and therefore, compound 15 was identified as eriocitrin (31,34).

For compounds 13 and 16, the LC-MS/MS chromatogram showed a molecular ion peak at m/z 611.2 [M+H]⁺, which corresponds to the molecular formula C₃₅H₆₂O₈. Due to consecutive losses of water (18 Da), the mass fragments at m/z 575.4 [M+H-H₂O]⁺, 557.4 [M+H-2H₂O]⁺, 539.5 [M+H-3H₂O]⁺, and 521.7 [M+H-4H₂O]⁺ show the existence of four hydroxyl groups. This was in agreement with Gu et al. (35), who reported that fragment ions, indicative of the multihydroxylated structures of the acetogenins, were generated from consecutive losses of H₂O (three to five molecules). This, therefore, led to the identification of compounds 13 and 16 as annonisin and montanacin B, respectively (36,37). Compound 14 gave a molecular ion peak at m/z 613.35 [M+H]⁺, which corresponds to the molecular formula C₃₅H₆₄O₈ (38). Similarly to compound 13 and 16, compound 14 also had the same fragments at 577.4 [M+H-H₂O]⁺, 559.2 [M+H-2H₂O]⁺, 541.5 [M+H-3H₂O]⁺, 523.6 [M+H-4H²O]⁺ and 505.6 [M+H-5H₂O]⁺ due to successive losses of water (18 Da) which indicates the presence of five hydroxyl groups. These results were in agreement with the literature (38,39), and compound 14 was identified as annomuricin A.

Molecular docking

In order to produce docking scores that show the bio-affinity of the docked molecules, the molecular docking algorithm simulates ligand interactions in the target proteins’ or receptors’ active regions (40). C. albicans is reported to have Sap 1-10 genes. The Sap proteins are one of the classic pathogenic factors whose expression is controlled by numerous parameters such as pH levels, temperature, location of infection, and physicochemical ambient conditions. Since Sap3 is implicated in mucosal infections, the development of inhibitors targeting Sap3 is a promising strategy for addressing infections caused by C. albicans (7).

The molecular docking calculations presented in Table 2 indicate that the enzyme Sap3 (2H6T) exhibited the utmost binding affinities with eriocitrin (-9.7 kcal/mol). The 2D visualization (Fig. 2A) displayed that eriocitrin established strong hydrogen bonds with GLY 34, ASP 86, SER 36, and ASN 192 at the protein active site. However, it also used a distinct pi bond to engage with ASP 218 at the catalytic pocket of the enzyme. In addition, the 2H6T macromolecule interaction with apigenin 6,8-di-C-glucoside showed the second-highest binding energy (−8.9 kcal/mol) through strong hydrogen bonds with GLY 220, THR 222, THR 300, and TYR 225 along with ASP 86 (π bond), GLY 85 (π-π stacked), as well as a noncovalent (amide-π stacking) interaction with TYR 225 (Fig. 2B). The strength of binding between receptors and ligands is determined by the number of hydrogen bonds produced. Also, binding energy influences the efficiency of ligand binding to enzymes (7).

Another molecule, xylopine, also formed bonds with GLY 220 (H-bond), GLY 85 (π stacking), and TYR 225 (π-σ) with a binding energy of −8.5 kcal/mol (Fig. 2C). Whereas isovitexin-2H6T showed same binding energy as xylopine, by interacting with GLU 193, THR 88, ASP 86, THR 222 and VAL 12 amino acids through hydrogen bonds and with LEU 216 by π-alkyl interactions (Fig. 2D). Intriguingly, the 2D viewpoint in Fig. 2C-D confirmed that the ASP 218 and ILE 305 amino acids are the brush border enzymes involved in π-anion and π-alkyl bond formation with both compounds, respectively. A noteworthy interaction has also been seen between rutin-2H6T molecular docking (binding energy of –8.4 kcal/mol). In such interactions, conventional hydrogen bonding consisted of ASP 37, ASP 218, SER 36, GLY 85, GLY 34, and THR 221, while alkyl bond formation occurred with ILE 82 amino acid (Fig. 2E). Vitexin interacted with 2H6T at active sites via ASN 192, SER 36, 35, 81, ILE 82, GLU 83, 132, 193, LEU 194, 216, GLY 34, HIS 131, THR 130, ASP 37, and LYS 129 amino acids with binding energy of −8.1 kcal/mol (Fig. 2F) whereas asimilobine interacted with same binding energy thru ASP 32, 86, 218, THR 221, 222, ILE 123, 305, TYR 84, 225, GLY 34, 85, 220, SER 35, LEU 216 amino acids interaction (Fig. 2G). Kaempferol 3-O-rutinoside interacted with 2H6T protein active sites through ASN192, THR 130, TYR 84, 128, SER 35, 36, 81, GLU 83, 132, 193 LYS 129, GLY 34, 85, ASP 37, HIS 131, LEU 216, ILE 82, ARG 195, and TYR 303 amino acids with binding energy of –8.0 kcal/mol (Fig. 2H). Other possible identified compounds such as (+)-4’-O-methylcoclaurine, quercetin-3-O-D-glucuronide, annonisin, annomuricin A, montanacin B, and montanacin C showed binding energy < –8.0 kcal/mol (Table 2) by interacting with several amino acids as shown in Fig. 2I-N.

In addition, with a binding value of –5.6 kcal/mol, swainsonine likewise showed strong attraction for the 2H6T catalytic site by interacting with GLU 193, THR 33, LEU 216, 194, GLY 34, 127, ASN 192, TYR 128, SER 35, 36 amino acid residues (Fig. 2O). All the ligands are docked inside the active site. As previously documented, the phytochemicals found in A. muricata exhibit potential antifungal activity against C. albicans (12,38,41,42). Thus, it is possible to infer that the inhibition of the Sap3 enzyme by the extract of A. muricata leaves (Fig. 3) is caused by the phytochemicals in the plant influencing the activity of the enzymes.

Effects of AM and AM+FLU on C. albicans biofilm formation

To determine if the predicted interaction may indeed cause an effect on biofilm formation of C. albicans, an in vitro antibiofilm assay was performed. The methanolic extract of A. muricata leaves showed a dose-dependent effect over C. albicans biofilm formation, as shown in Fig. 3.

The maximum level of inhibition observed was ~60% at 240 mg/mL. The biofilm inhibitory activities of a clinical antifungal drug used for the treatment of invasive fungal infections, fluconazole, were also evaluated in this study as positive control and showed a similar dose-dependent response, with ~70% inhibition observed at 120 mg/mL. Interestingly, at lower concentrations, the combined effect of the extract and fluconazole achieved synergistically increased levels of inhibition. At 15 mg/mL of the combination, biofilm formation was inhibited in excess of 70%. Our findings support Campos et al.’s 2023 investigation, which found that an ethanolic extract from A. muricata leaves has antifungal properties against a multidrug-resistant strain of C. albicans (38,42).

This synergism is also supported by the IC₅₀ values (Suppl. Table 1). As expected, FLU (IC₅₀ = 18.33 μg/mL) had a lower IC₅₀ value than the AM (IC₅₀ = 128.70 μg/mL). Importantly, the IC₅₀ value of the combination (IC₅₀ = 0.83 μg/mL) is two orders of magnitude lower than fluconazole, demonstrating a pronounced synergistic activity. A low IC₅₀ value indicates that the drug is effective at lower concentrations, resulting in reduced systemic toxicity when administered to patients (43). Identically, it has been seen that combined treatment is a proposed approach to overcome the issues associated with single-agent therapy; numerous studies have shown that combination therapies are superior to single-agent therapy (44,45). Accordingly, flavonoids (such as quercetin, catechin, epigallocatechin gallate, etc.) are a promising synergistic agent when combined with fluconazole and are an effective antifungal agent against C. albicans biofilms (45-47). Therefore, the phytocomponents of A. muricata leaves may play a role in their promising antifungal properties and possibly produce a synergistic effect with fluconazole.

Effect of AM and AM + FLU on the cell surface of C. albicans

Scanning electron micrographs of the control and treated (AM and AM+FLU) biofilms showed that untreated cells formed normal biofilms consisting of hyphae with smooth surfaces (Fig. 4A). As expected, the biofilms grown in the presence of FLU were less dense and hyphal formation was inhibited (Fig. 4B). Biofilms exposed to AM alone, were inhibited compared to the controls, but still displayed hyphal formation (Fig. 4C). In Fig. 4D, details of the biofilm exposed to the combination of fluconazole and extract can be seen. These biofilms were very sparse, and hyphal formation was inhibited. The release of cellular material can be seen. Studies have reported that A. muricata leaf extract causes modifications to the surface cells in biofilm cells by targeting the cell envelope of C. albicans. This effect may result from several identified compounds in the plant extract (38,42). We identified many of the same compounds reported previously (38), and there is a possibility that these compounds are responsible for generating a synergistic effect by interacting with fluconazole when AM + FLU is given as a combined treatment.

Conclusion

LC-MS analysis of methanol extract demonstrated that A. muricata contains several chemical constituents, 14 of which were identified. As indicated by molecular docking, some of these may interact with the secreted protease Sap3, which is an important hydrolytic enzyme of C. albicans. Our findings regarding the antibiofilm activity of AM leaf extract corroborate recent studies conducted by Campos and co-workers (38,42), who demonstrated that the ethanolic extract of A. muricata leaf is a promising anticandidal agent and can reduce fungal infection (in vitro and in vivo). In addition, we established that the combination of A. muricata methanol extract and fluconazole shows synergistic increased inhibition of biofilm formation, which may be advantageous in combating fluconazole resistance in C. albicans by lowering the effective dose required to inhibit biofilm formation. These results suggest that more research is necessary to fully understand the underlying mechanisms of action of AM + FLU and explore its potential as an anticandidal agent in vivo.

Disclosures

Conflict of interest: The authors declare no conflict of interest.

Financial support: This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Data availability statement: The research data associated with this article are included within the article and in the supplementary material of this article.

Author’s contribution: Conceptualization: APM, CHP; Data Curation: APM, MY, JN; Formal Analysis: LKS, GK, APM, CHP, MY; Investigation: APM, LKS, MY, GK; Methodology: APM, LKS, MY, GK; Project Administration: APM, MGM, CHP; Supervision: CHP, MGM; Writing Original Draft: APM; Writing – Review & Editing: APM, GK, CHP.