Antifungal Activity of Extracts, Fractions, and Constituents from Coccoloba cowellii Leaves

Coccoloba cowellii Britton (Polygonaceae, order Caryophyllales) is an endemic and critically endangered plant species that only grows in the municipality of Camagüey, a province of Cuba. A preliminary investigation of its total methanolic extract led to the discovery of promising antifungal activity. In this study, a bioassay-guided fractionation allowed the isolation of quercetin and four methoxyflavonoids: 3-O-methylquercetin, myricetin 3,3′,4′-trimethyl ether, 6-methoxymyricetin 3,4′-dimethyl ether, and 6-methoxymyricetin 3,3′,4′-trimethyl ether. The leaf extract, fractions, and compounds were tested against various fungi and showed strong in vitro antifungal activity against Cryptococcus neoformans and various Candida spp. with no cytotoxicity (CC50 > 64.0 µg/mL) on MRC-5 SV2 cells, determined by a resazurin assay. A Candida albicans SC5314 antibiofilm assay indicated that the antifungal activity of C. cowellii extracts and constituents is mainly targeted to planktonic cells. The total methanolic extract showed higher and broader activity compared with the fractions and mixture of compounds.


Introduction
Fungal infections represent a major health problem, with mortality rates comparable to those of tuberculosis or malaria, estimated at 1.5 million individuals per year [1]. The growing number of immunocompromised individuals due to the increase of organ transplantation, the prevalence of cancer and AIDS patients, and the ageing population have all contributed to this situation [2]. The predominant etiological agents of systemic fungal infections are species of Candida, Aspergillus, and Cryptococcus, representing over 90% of mycotic deaths [3]. Candida sp. stand out as the most common in the intensive care units, affecting individuals that undergo invasive clinical procedures and/or have experienced significant traumas requiring prolonged treatments [4]. Candida albicans is the most common (50-70%), producing more infections than all other Candida species combined [5]. C. glabrata is the second most dangerous species, with an increasing number of invasive candidiasis over the past several years [4].
Despite the negative impact that these fungi have on human health, currently, there are only three classes of antifungal agents available to treat serious Candida infections:   At the same time, we intended to evaluate the selectivity of the antifungal activity using the cytotoxicity on human foetal lung fibroblasts (MRC-5 SV2 cells). None of the fractions showed cytotoxicity except for the n-hexane fraction (nH-F). Thus, the selectivity index of the total extract ranged from 160 (for C. glabrata) to 3 (for C. tropicalis), depending on the microorganism susceptibility. The indexes were consequently lowest for the MeOH90-F fraction, with values of 27 and 5 for C. glabrata and C. parapsilosis, respectively,

UHPLC-HRMS Characterization
On the basis of the results obtained in the biological assays, the MeOH90-F fraction of C. cowellii was qualitatively analysed for its chemical composition using UHPLC-ESI-QTOF-MS in negative ionization mode. The base peak intensities (BPI, peaks 1 to 30 corresponding to Table 1) in negative ionization mode are shown in Figure 2.
The analysis and interpretation of the MS E data allowed the identification of 21 phytochemical compounds from a total of 30 peaks ( Table 2). The data from formerly identified phytochemicals in the Coccoloba genus and/or the Polygonaceae family were also utilised in the identification when applicable.
The fragment nomenclature for flavonoid glycosides was applied according to Vukics and Guttman [9]. The nomenclature used for lignin oligomers and fragments was taken from Morreel et al. [10]. The MS spectra are shown in the Supplementary Material ( Figure S1a

UHPLC-HRMS Characterization
On the basis of the results obtained in the biological assays, the MeOH90-F fraction of C. cowellii was qualitatively analysed for its chemical composition using UHPLC-ESI-QTOF-MS in negative ionization mode. The base peak intensities (BPI, peaks 1 to 30 corresponding to Table 1) in negative ionization mode are shown in Figure 2. The analysis and interpretation of the MS E data allowed the identification of 21 phytochemical compounds from a total of 30 peaks ( Table 2). The data from formerly identified phytochemicals in the Coccoloba genus and/or the Polygonaceae family were also utilised in the identification when applicable.

Flavonoid Glycosides/Glucuronides
MS spectral data from peaks 1-3 matched with those previously reported for the total extract [8]. Peak 1 data fragmentation suggested (see Table 2) the presence of myricetin-Oglucuronide. The main fragment corresponded to the ion at m/z 317 [Y 0 ] − as a consequence of glucuronide loss, followed by a retro Diels-Alder (RDA) fragmentation generating the secondary fragments at m/z 287 [Y 0 -H-CO-H] − and m/z 179 ( 1,2 A − ). On the other hand, the peak 2 and 3 data implied the presence of a quercetin-O-pentoside following similar fragmentation behaviour to compound 1, but with the particularity that both homolytic and heterolytic loss of the sugar could be documented by the presence of ions at m/z 300 [Y 0 -H] −· and m/z 301 [Y 0 ] − , respectively. It was impossible to differentiate between these possible isomers; therefore, they were labelled as quercetin-O-pentoside 1 and 2, respectively.

Lignin Oligomers
Peaks 4, 5, 7-11, and 19 all showed similar spectra and fragments. The fragmentation patterns matched the ones reported by Morreel et al. [10,11] for lignin oligomers. Table 3 shows the product ions of 4, 5, and 7-11, identified as trilignols. The fragment ion of m/z 195 (present in all the compounds) indicates that the 8-phenolic end (A − fragment) corresponds to a G unit (guaiacyl, a unit derived from coniferyl alcohol) [12]. The fragmentation patterns ( Figure S2a) were in correspondence with isomers of type G(8-O-4)X(8-5)X and G(8-O-4)X(8-8)X-containing trimers, with X being either an S unit (syringyl, a unit derived from sinapyl alcohol) or the previously mentioned G unit. The relative intensity of the product ions compared with the base peak is given in parentheses.  [13]. Table 4 shows the product ions of peaks 14, 15, 18, and 21-27, and the different losses were characteristic of polymethoxyflavonoids with at least three methoxy groups, except for peaks 14 and 18, which only contained one methoxy group.
Flavonoid glycosides and glucuronides with quercetin and myricetin aglycon moiety have already been reported in relatively high concentrations in the total extract of C. cowellii leaves [8] and in other species of the genus [14]. In fact, compounds 1-3 from this study matched the ones previously identified in the total extract [8]. The other compounds identified in the methanol 90% fraction were reported in this species for the first time.
The harsh growing conditions of C. cowellii can be associated with the presence of compounds unique to this species. This plant, strictly endemic to the savannas of north Camagüey, Cuba, only grows on serpentine soils and is subjected to almost constant drought and high levels of solar radiation ( Figure S3). Its leaves are quite hard, with a Pharmaceuticals 2021, 14, 917 9 of 17 lignified cuticula to prevent water loss. Therefore, these relatively rare compounds may be synthesised as a way to adapt to such severe adverse conditions [26].
These results portray a complex panorama: while the highly active total extract is rich in quercetin and myricetin glucosides/glucuronides and proanthocyanidins, its only active fraction is mainly comprised of lignans and methyl and methoxy derivatives of quercetin and myricetin. Only the aforementioned compounds 1-3 were common between both active tested extracts. In addition, and as can be seen in Figure 1, the lowest yield of extractable substances was obtained for the methanol 90% fraction, and therefore, active compounds must be present at very low concentrations in the crude extract. In fact, this could be an explanation for why such lignanoids and methoxyflavonoids were not detected during the study of the total extract. Despite these unfavourable conditions, biofractionation was pursued, aided by flash chromatography.
This second fractionation rendered 24 subfractions, which were all evaluated for their antifungal activity. Only one subfraction, M-6, showed some activity against C. neoformans and C. glabrata (Table 5). Therefore, this subfraction was selected for the isolation and characterization of its components through a semi-preparative HPLC-DAD-MS.

Isolated Compounds
The semi-preparative HPLC-DAD-MS procedure rendered three main isolates (M-6A, M-6B, and M-6C) with yields of 2.6, 4.0, and 1.3 mg, respectively ( Figure 1). The analysis of the 1 H NMR spectra revealed that M-6A, M-6B, and M-6C were mixtures of two compounds in different ratios (Figure 3a). The low yields obtained did not allow further purification. Nevertheless, it was possible to determine the structures of both the major and minor compounds, except for the minor compound of M-6A. Employing the results obtained from SMART 2.1 (Tables S1-S5, Supplementary Information) and the molecular weight derived from the m/z value of the [M-H] − peaks, a preliminary structure was drawn.
M-6A was identified as a mixture of quercetin (major, compound I) and an unidentified impurity (minor) (amorphous yellow powder, 2.6 mg). The NMR data ( Figure S4) were similar to those previously reported in the literature for quercetin [27].
M-6B, and M-6C) with yields of 2.6, 4.0, and 1.3 mg, respectively (Figure 1). The analysis of the 1 H NMR spectra revealed that M-6A, M-6B, and M-6C were mixtures of two compounds in different ratios (Figure 3a). The low yields obtained did not allow further purification. Nevertheless, it was possible to determine the structures of both the major and minor compounds, except for the minor compound of M-6A. Employing the results obtained from SMART 2.1 (Tables S1-S5, Supplementary Information) and the molecular weight derived from the m/z value of the [M-H] -peaks, a preliminary structure was drawn.  M-6A was identified as a mixture of quercetin (major, compound I) and an unidentified impurity (minor) (amorphous yellow powder, 2.6 mg). The NMR data ( Figure S4) were similar to those previously reported in the literature for quercetin [27].
For the structure elucidation, NMR data of compounds with a similar chemical backbone were consulted for comparison [31][32][33]. Table 5 shows the activities of subfraction M-6 and the mixtures of compounds defined as M-6A, M-6B, and M-6C. As can be seen, the antifungal activity of the binary mixture M-6C against C. glabrata and C. neoformans was approximately the same as the activity of the methanol 90% fraction, but the activity remained lower than that of the total extract. Furthermore, it was noted that the M-6 fraction and the three mixtures of compounds were active against C. glabrata specifically. The increasing resistance of this Candida species against azole compounds [4] and echinocandins [34] necessitate the search for novel compounds that can be used to treat infections caused by this fungus.
Methoxyflavonoids (specifically, derivatives of the flavonols quercetin and myricetin) seem to be responsible for the observed activity. Antifungal activity has previously been reported for these types of compounds [35,36]. The analysis of Limonium caspium (Plumbaginaceae) showed that the compound 5-methylmyricetin exhibited good antifungal activity against C. glabrata, with an IC 50 value of 6.79 µg/mL [37]. From Combretum zeyheri (Combretaceae), the compound 5-hydroxy-7,4 -dimethoxyflavone was found to be active against C. albicans using the broth dilution method. These substances showed synergistic activity when combined with miconazole, completely inhibiting C. albicans growth after only 4 h of incubation [38]. The study of the plant Kaempferia parviflora (Zingiberaceae) allowed the identification of 3,5,7,4 -tetramethoxyflavone and 5,7,4 -trimethoxyflavone as acceptable antifungal agents against C. albicans, with respective IC 50 values of 39.71 and 17.63 µg/mL [39].

Antibiofilm Screening Assay
Plant extracts and/or their isolated compounds can act as antimicrobials via different mechanisms. Biofilm disruption is one of the most explored, and the active extracts/fractions/compounds were tested for this mode of action. The biofilm screening assay was realised by employing the C. albicans SC5314 strain [40]. The formation of fungal biofilms decreased overall susceptibility from both host defences and antimicrobial therapies [41]. Natural products have been reported to demonstrate antibiofilm activity, which is relevant because developing resistance to these kinds of molecules is rare [42]. The results showed ( Table 6) that only the total extract showed low activity against the C. albicans biofilms, while the rest of the samples showed no effect at the tested concentrations (0.25 to 64 µg/mL to all the samples except for M-6C due to the low amount).  This assay indicated that the antifungal activity of the total extract of C. cowellii on Candida species is mainly targeted to planktonic cells and has rather low activity against biofilm colonies, at least in the conditions established in these experiments. The mechanism(s) of action of the total extract and active fractions could thus be related to inducing the death of free-living cells and not the disruption of cell-to-cell communication and biofilm association.

Discussion
The bioassay-guided fractionation performed on C. cowellii leaves led to the isolation and tentative identification of at least 21 new compounds in this species; nevertheless, the isolation of the compounds responsible for high antifungal activity shown by the total extract was not successful. Unfortunately, this is not an uncommon situation. Bioassayguided fractionation of plant extracts is not always effective. This procedure can lead to failures in the isolation of active compounds and losses of activity [43]. The probable degradation of the compounds during the purification process, the difficulty related to isolating bioactive compounds present in low concentrations, and/or the loss of other substances responsible for potential synergistic effects are some of the causes referred to in the literature [44]. In any case, the higher and broader activity of the total extract of C. cowellii compared with the fractions and mixtures of compounds can be associated with any of these aforementioned events.
According to the previous analysis, flavonoid glycosides or glucuronides as well as proanthocyanidins are the main compounds of the total extract of C. cowellii leaves [8]. These compounds have a broad spectrum of biological activities, including antifungal activity [45,46]. This mixture of different kinds of polyphenols can contribute to the overall antifungal activity, considering that the different groups can have different underlying mechanisms of action [35]. Proanthocyanidins have shown synergic effects with vari-ous commercial antifungal agents. Catechin and epigallocatechin gallate have shown synergism with fluconazole. These compounds induce the activation of the phospholipid phosphatidylserine, which inhibits fatty acid synthase [47], supporting in this way the action mechanism of fluconazole. In a preclinical study of disseminated candidiasis, epigallocatechin-O-gallate administered with amphotericin B showed a synergistic interaction against C. albicans. The results of the assay showed that epigallocatechin-O-gallate exclusively inhibits the hyphal formation and ergosterol synthesis of the fungi [48]. The synergic effect of these main compounds, together with the effect of the methoxyflavonoids (see Table 3), is a plausible explanation for the high antifungal activity identified in the total extract of C. cowellii leaves. In any case, the extract is a promising candidate for the treatment of fungal diseases, either alone or mixed with commercial antifungals as a way to increase their effectiveness and/or decrease the required doses. Confirmatory assays will be necessary to corroborate these hypotheses.

Leaf Extraction and Bioassay-Guided Fractionation
The plant material was processed, and the total extract was obtained as previously described [8]. The total extract (40.00 g) was dissolved in acidic (pH < 3) methanol 50% and partitioned with dichloromethane. The dichloromethane residue was concentrated and partitioned between n-hexane and methanol 90%. Next, the extract was basified with concentrated ammonia until it reached a pH > 9, and partitioning with ethyl acetate and then n-butanol was performed. The yield of all the fractions was 2.65 g for the n-hexane fraction (nH-F), 1.89 g for the methanol 90% fraction (MeOH90-F), 13.04 g for the ethyl acetate fraction (EtOAc-F), 10.14 g for the n-butanol fraction (nBut-F), and 7.02 g for the residual fraction (Re-F). The total yield of the fractionation was 86.9%. nH-F, MeOH90-F, EtOAc-F, and nBut-F were concentrated under reduced pressure at 40 • C and then stored at −20 • C until further use.

Antifungal Assay
The microdilution method with resazurin (redox indicator) in sterile 96-well microplates was the assay used to determine antifungal activity. This was performed according to the protocols of the Laboratory of Microbiology, Parasitology, and Hygiene (LMPH) as previously reported [8,49]. Miconazole was used as a positive control.

Microorganisms and Dilutions
The microorganisms used in the study were obtained from the culture collection of the Laboratory of Microbiology, Parasitology, and Hygiene (LMPH of the University of Antwerp). The strains of Candida albicans ATCC B59630 (azole-resistant), Candida glabrata ATCC B63155, Candida parapsilosis ATCC B66126, Candida tropicalis ATCC CDC44 as well as Aspergillus fumigatus ATCC B42928 and Cryptococcus neoformans ATCC B66663 were used for in vitro screening of antifungal activity. The LMPH protocols established for the culture of the microorganisms and the dilutions of the samples were followed, as performed and reported in previous publications [8,49]. Tested sample concentrations ranged from 0.25 to 128 µg/mL.

Antibiofilm Screening Assay
The antibiofilm assay was performed following the LMPH protocols previously reported [50]. A Candida albicans SC5314 overnight culture, grown in RPMI, was diluted to an optical density (OD) of 0.04-0.05 in RPMI medium, and 95 µL of this suspension was added to a 96-well plate. Then, 5 µL samples and control (miconazole) were added (in a final concentration range from 0.25 to 128 µg/mL). The plate was wrapped in parafilm and placed in a styropor box, a cup of MilliQ water was added, and the box was placed in a shaking incubator at 37 • C and 25 rpm. After 24 h of incubation, the medium was carefully removed with a vacuum pump, avoiding contact of pipette tips with the biofilms. Finally, the biofilms were washed and quantified by adding 150 µL of resazurin solution (1/10 in PBS) to each well. The plate was wrapped in aluminium foil and incubated in the dark at 37 • C for 1 h. The fluorescence was measured with a microplate reader (TECAN GENios, Männedorf, Switzerland) at a λ ex of 550 nm and a λ em of 590 nm.

Cytotoxicity Assay
Human foetal lung fibroblasts (MRC-5 SV2 cells) were purchased from the ATCC (American Type Culture Collection). For their culture and assay, the protocols of the Laboratory of Microbiology, Parasitology, and Hygiene (LMPH) were followed, as reported in previous publications [8,49]. The 50% cytotoxic concentration (CC 50 ), resulting from the % reduction of cell growth/viability compared to control wells, is reported. Tamoxifen was used as a reference drug.

Isolation of Constituents from Active Fractions
In the fractionation of the MeOH90-F fraction (1.7 g), selected as the most active, flash chromatography was applied on a GraceResolv 80 g silica column using a RevelerisiES system (Columbia, MD, USA). A gradient program consisting of dichloromethane (A), ethyl acetate (B), and methanol (C) as mobile phases and a flow rate of40 mL/min were used ( Figure S7). These solvents were applied as follows: 0 to 6 min 100% A, 0% B, and 0% C; 6 to 40 min linear changing until 0% A and 100% B; 40 to 45 min 0% A, 100% B, and 0% C; 45 to 81 min linear changing until 0% B and 100% C. An evaporative light scattering detector (ELSD) and UV absorption at 254 and 366 nm were used as the detection methods. According to their TLC profile, all collected subfractions were pooled for a total of 24 subfractions (M-1 to M-24).
On the basis of its chromatographic profile, fraction size, and biological activity, subfraction M-6 (85 mg) was selected for further purification by semi-preparative HPLC-DAD-MS (Waters, Milford, MA, USA). A C18 Luna (250 mm × 10.0 mm, particle size 5 µm) from Phenomenex (Utrecht, The Netherlands) was used as a column. As in previous experiments, H 2

Statistical Analysis
GraphPad Prism V8 Software for Windows (GraphPad, San Diego, CA, USA) was employed for all statistical analyses. The results were analysed and expressed as the means ± standard deviation (SD) of three different replicates.

Conclusions
In this study, five secondary metabolites were isolated and characterized from the MeOH90-F fraction of the total extract of C. cowellii by means of a combined methodology of NMR and MS analysis. All five are reported here for the first time for both the plant and the genus. Another 16 compounds were tentatively characterized employing UHPLC-HRMS. C. cowellii extract was confirmed to have good antifungal activity against a second fungal/yeast panel, while fractions and mixtures of compounds obtained from the bioassayguided fractionation showed acceptable activity specifically against C. glabrata and C. neoformans. These results highlight the possible use of this plant as a natural antifungal and contribute to a better understanding of the phytochemistry and biological activities of the genus Coccoloba. At the same time, they suggest the probable synergistic effect that the combination of different types of polyphenols may show in inhibiting fungal growth.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript or in the decision to publish the results.