**1. Introduction**

Parasites of the genus *Plasmodium*, which cause malaria, contain an organelle called the apicoplast, and its functioning is essential for the survival in both the erythrocytic and the hepatic phases of development in mammalian hosts [1]. The apicoplast is similar to plastids of plants, as it is thought to be a vestigial plastid derived from endosymbiosis of a red alga by a heterotropic, unicellular eukaryote [2]. It retains plant-like metabolic pathways, which are absent in vertebrate hosts, making the enzymes of these pathways suitable targets for malaria drug discovery [3–6]. Some phytotoxins released by plant pathogenic fungi inhibit metabolic pathways of the plastid [7]. If the compounds produced by plant pathogenic fungi show phytotoxicity and malarial parasite death without causing cytotoxicity towards mammalian cells, it indicates that the mechanism of parasitic death

**Citation:** Kumarihamy, M.; Rosa, L.H.; Techen, N.; Ferreira, D.; Croom, E.M., Jr.; Duke, S.O.; Tekwani, B.L.; Khan, S.; Nanayakkara, N.P.D. Antimalarials and Phytotoxins from *Botryosphaeria dothidea* Identified from a Seed of Diseased *Torreya taxifolia*. *Molecules* **2021**, *26*, 59. https://dx. doi.org/10.3390/molecules26010059

Academic Editors: Valeria Patricia Sülsen and William Setzer Received: 24 September 2020 Accepted: 22 December 2020 Published: 24 December 2020

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/ licenses/by/4.0/).

may be due to the ability of the compounds to inhibit the plant-like metabolic pathways in the apicoplast. As part of our program to search for new antimalarials from plant pathogenic fungi [8–12], we investigated fungi from seeds of a diseased *Torreya taxifolia* Arnott. (Taxaceae). *T. taxifolia*, also known as Florida nutmeg, Florida torreya, stinkingcedar, or gopherwood, is a rare, critically endangered evergreen conifer endemic to three counties in Northern Florida [13–15]. The decline of the native population during the recent past has been attributed to both abiotic and biotic causes, including fungal diseases. Several fungi have been isolated from diseased *T. taxifolia* and some of them have been shown to cause leaf spots and canker disease in healthy plants [15–18]. For this study, seeds of *T. taxifolia* were collected from a tree with disease symptoms cultivated on the Biltmore Estate in Asheville, North Carolina.

From fragments of a surface-sterilized seed, several endophytic fungi were isolated. An EtOAc extract of the broth of one of these fungi grown in potato-dextrose liquid medium showed phytotoxic and antiplasmodial activities. This fungus (UM124) was identified as *Botryosphaeria dothidea* (*Botryosphaeriaceae*) by DNA analysis. Members of the family *Botryosphaeriaceae* (*Botryosphaeriales*, *Ascomycota*) cause leaf spots, fruit and root rots, and cankers in a variety of hosts [19], and *B. dothidea* has specifically been isolated from a large number of diseased and healthy woody plants, including many economically important crops [20]. A *Botryosphaeria* sp. strain has previously been isolated from *T. taxifolia* leaves infected with needle-spot disease [14]. A chemical investigation of endophytic *B. dothidea* has previously been carried out, and a variety of compounds, including a simple α-pyridone, were reported from a solid culture of this fungus [21].

Bioassay-guided fractionation of the active EtOAc extract resulted in the isolation and identification of a mixture of known, isomeric phytotoxins, FRT-A (**1**) [22] and flavipucine (**2**) [23] (or their enantiomers, sapinopyridione [24] and (-)-flavipucine [22]), as well as two new unstable *α*-alkyl-*γ*-lactam alkaloids, dothilactaenes A (**3**) and B (**4**), closely related to epolactaene [25] (Figure 1). While dothilactaene A showed no activity, dothilactaene B was isolated from the active fraction, which showed moderate, selective antiplasmodial activity. Two other components isolated from this fraction had the same molecular formula and similar NMR data indicating that they were diastereomers but due to the lack of sufficient materials their structural investigation was not completed. This is the first report of the isolation of a fungus producing phytotoxins from the seeds of diseased *T. taxifolia. B. dothidea* might play a significant role in decreasing the population of the endangered *T. taxifolia*.

**Figure 1.** Structures of compounds.

### **2. Results and Discussion**

The EtOAc extract of the fermentation broth of a fungus isolated from surface sterilized seed fragments of diseased *T. taxifolia* showed good phytotoxic activity against model plants, a dicot (lettuce, *Lactuca sativa* L.) and a monocot (bentgrass, *Agrostis stolonifera* L.). Furthermore, it showed good antiplasmodial activity against chloroquine-sensitive (D6) and -resistant (W2) strains of *Plasmodium falciparum* (IC50 = 0.86 and 1.3 μg/mL, respectively), with low cytotoxicity (32 μg/mL) to mammalian kidney fibroblasts (Vero cells).

The analysis of the ITS genomic region of UM124 for a closest neighbor with published sequences showed that the highest identity was with various strains of the species *Botryosphaeria dothidea*. Analysis of 18S rDNA of the fungus gave 100% sequence identity to *B. dothidea.* The construction of the phylogenetic tree with different strains of *B. dothidea* involved 29 nucleotide sequences with a total of 488 positions in the final dataset. In addition, for the construction of the phylogenetic tree of UM124 with various taxa of the family *Botryosphaeriaceae*, 92 nucleotide sequences were analyzed with a total of 382 positions in the final dataset (Supplementary Materials: Tables S1 and S2 and Figures S1 and S2).

The EtOAc extract of the culture broth was fractionated by silica gel column chromatography and the active fractions were combined and separated by Sephadex LH-20 gel filtration using MeOH as the mobile phase. A fraction with high phytotoxic activity and no antiplasmodial activity afforded a white precipitate, and its NMR data (Table 1) indicated that it was a 3:1 mixture of two isomeric known 2,4-pyridione epoxides, fruit rot toxin A (FRT-A) (**1**) [22,23] and (+)-flavipucine [24] (or their enantiomers, sapinopyridione [25] and (-)-flavipucine [24,26,27]). FRT-A was previously reported from *B. berengeriana* [23].

**Table 1.** 1H- and 13C-NMR data for mixture of compounds **1** and **2** (3:1) in CDCl3.


<sup>a</sup> Recorded at 100 MHz, <sup>b</sup> Recorded at 400 MHz, <sup>c</sup> Overlapped signals.

The subfraction with antiplasmodial activity (IC50 = 0.68 and 0.78 μg/mL) was further separated by reverse-phase semi-preparative HPLC to afford an inactive and an active fraction (IC50 < 0.523 μg/mL), with no cytotoxicity to Vero cells. The NMR data of the inactive fraction showed that it was more than 85% of a single compound with lipid impurities. Due to instability and limited material, further purification of this compound was not possible and was identified as dothilactaene A (**3**). The active fraction was separated by analytical HPLC to three peaks. The NMR data of the first peak showed it was a single compound with 90% purity with lipid contaminants and was identified as dothilactaene B (**4**).

The molecular weight of dothilactaene A (**3**) was determined as C22H29NO5 by HRES-IMS. The 1H- and 13C-NMR data of **3** (Table 2) indicated the presence of 22 carbon resonances that consisted of five quaternary, three carbonyl, six methine, two methylene, and

six methyl carbons. Analysis of the 1H- and 13C-NMR data indicated that it was an *α*alkyl-*γ*-lactam derivative, related to epolactaene [28] (Figure 1). Comparison of the 1H and 13C-NMR data of these compounds (Table 2) showed that they had the same side chain but differed in the 2-pyrrolidone ring. These differences were attributable to the replacement of the epoxide and the hydroxy groups in the 2-pyrrolidone ring of epolactaene by a double bond and methoxy group, respectively, in **3**. In the HMBC spectrum of **3**, cross-peaks of H3-1 (*δ*<sup>H</sup> 1.72) with C-2 (*δ*<sup>C</sup> 139.9) and C-3 (*δ*<sup>C</sup> 130.3); H-2 (*δ*<sup>H</sup> 6.96) with C-3 (*δ*<sup>C</sup> 130.3) and C-19 (*δ*<sup>C</sup> 167.8); H-4 (*δ*<sup>H</sup> 5.97) with C-21 (*δ*<sup>C</sup> 14.3), C-19 (*δ*<sup>C</sup> 167.8), and C-2 (*δ*<sup>C</sup> 139.9); H3-21 (*δ*<sup>H</sup> 1.62) with C-6 (*δ*<sup>C</sup> 135.6) and C-4 (*δ*<sup>C</sup> 123.1); H2-8 (*δ*<sup>H</sup> 2.32) with C-6 (*δ*<sup>C</sup> 135.6), C-7 (*δ*<sup>C</sup> 128.2), C-9 (*δ*<sup>C</sup> 28.4), and C-10 (*δ*<sup>C</sup> 148.6); and H3-22 (*δ*<sup>H</sup> 1.89) with C-10 (*δ*<sup>C</sup> 148.6) and C-12 (*δ*<sup>C</sup> 191.4) confirmed the structure of the side chain. Further, HMBC correlations of the -OCH3 singlet (*δ*<sup>H</sup> 3.18) with C-15 (*δ*<sup>C</sup> 89.2); the 18-CH3 singlet (*δ*<sup>H</sup> 1.61) with C-15 (*δ*<sup>C</sup> 89.2) and C-14 (*δ*<sup>C</sup> 149.5); and the H-14 olefinic proton singlet (*δ*<sup>H</sup> 6.83) with C-12 (*δ*<sup>C</sup> 191.4), C-13 (*δ*<sup>C</sup> 139.1), and C-17 (*δ*<sup>C</sup> 168.3) confirmed the structure of the 2-pyrrolidone moiety and its link to the side chain. Other COSY and HMBC correlations supported this structure. The (*E*)-configuration of the Δ6(7) double bond was determined by the coupling constant (*J* = 15.2 Hz). ROESY data indicated that the Δ2(3), Δ4(5) and Δ10(11) olefinic bonds were all (*E*) configured.

**Table 2.** 1H- and 13C-NMR spectroscopic data for compounds **3** and **4.**


<sup>a</sup> Recorded at 150 mHz, <sup>b</sup> Recorded at 600 mHz, <sup>c</sup> Overlapped signals.

Dothilactaene B (**4**) had the molecular formula, C24H33NO7, by HRESIMS data. Its NMR spectra showed resonances due to the same side chain as **3,** and COSY and HMBC data provided confirmatory evidence (Figure 2). Comparison of the remaining resonances with those of **3** indicated the absence of the olefinic bond and the methoxy group in the 2-pyrrolidone ring and the presence of three oxygenated carbons due to a glycerol moiety. The COSY and HMBC correlations (Figure 2) of **4** showed, respectively, an oxygenated methine (H-24; *δ*<sup>H</sup> 3.65) coupled with two oxygenated methylenes (H2-23; *δ*<sup>H</sup> 3.45, 3.49, and H2-25; *δ*<sup>H</sup> 3.39, 3.58), and the methylene resonance at *δ*<sup>H</sup> 3.39 (H-25) correlating with

C-23 (*δ*<sup>C</sup> 60.6) and C-24 (*δ*<sup>C</sup> 70.4), confirming the presence of a disubstituted glycerol moiety. Even though it is not common, secondary metabolites with glycerol moieties have previously been isolated from endophytic fungi [29–32]. In the HMBC spectrum, H-14 (*δ*<sup>H</sup> 4.29) had a cross peak with the oxygenated methylene at *δ*<sup>C</sup> 60.7 (C-23) of glycerol, indicating one of its linkages. To satisfy the molecular formula C24H33NO7 and the index of hydrogen deficiency, there should be another ether linkage to the 2-pyrrolidone moiety.

**Figure 2.** HMBC ( ) and COSY ( ) correlations of **4** ROESY ( ) correlations of **4**.

Even though no other HMBC cross peaks were visible between the glycerol moiety and the 2-pyrrolidone ring, this link should most probably be between the remaining primary hydroxy group of the former and OH-15 of the latter. An HMBC cross peak between H-14 (*δ*<sup>H</sup> 4.29) and the C-12 carbonyl (*δ*<sup>C</sup> 196.2) confirmed the link between the 2-pyrrolidone ring and the side chain. Large coupling (*J* = 15.6 Hz) between H-6 and H-7 showed that the Δ6(7) olefinic bond was in an *E* (*trans*) configuration. ROESY correlations (Figure 2) indicated the (*E*) configuration for the other double bonds in the side chain and the cofacial orientation of CH3-18, H-14, and H-24. Comparison of <sup>3</sup>*J*13,14 value of **4** with those reported for fusarin A [33] and lucilactanen [34,35] indicates an *anti* configuration of H-13/H-14. These data permitted assignment of the (13*S*\*, 14*R*\*, 15*R*\*, 24*R*\*) relative configuration. However, the limited sample quantity and the instability precluded electronic circular dichroism (ECD) studies to assign its absolute configuration.

The active fraction afforded two additional components, which had the same molecular formula, C24H33NO7, as that of **4**. NMR data analysis (Supplementary Materials) showed that they have the same gross structure but differ from **4** by the resonances of the 2-pyrrolidone ring and the glycerol moiety suggesting them to be diastereoisomers. Due to the lack of sufficient materials, their structural investigation was not completed.

These diastereomers probably formed from the precursor **6** by Michael addition (Scheme 1) [33]. All the fusarin [33], epolactaene [28], and lucilactaene [34] type compounds so far reported from natural sources have a free hydroxy group at C-15. It is very likely that compound **5** is the precursor of both **3** and **4.** The stereocenter at C-15 undergoes racemization (or epimerization) easily through ring opening, even under mild conditions [33,36]. The formation of methyl or glycerol ethers stabilizes this compound [36], but it is very likely that the product would be racemic in the case of **3** and diastereomeric in the case of **6**. In the chemical synthesis of lucilactaene, the Michael addition occurs spontaneously between the hydroxy group of the hydroxyethyl moiety and Δ13(14) olefinic bond [36]. The glycerol moiety in **6** also has a stereogenic center, and a similar Michael addition with the primary hydroxyl group of the glycerol would create additional stereogenic centers, yielding diastereomers.

In **4**, the NMR resonance of H-13 was very weak and integrated to less than one proton, and the resonance due to C-13 appeared very small. The 2-pyrrolidone ring in this type of compound could undergo keto-enol tautomerism [36,37], and in deuterated protic solvents it is highly likely that H-13 was partially exchanged with deuterium in CDCl3/methanol-*d4* in the presence of traces of water or HOD [38] (Scheme 2). The NMR spectra of fusarin A and lucilactaene have previously been recorded in CDCl3 [33–36].

**Scheme 1.** Possible route of formation of the diastereomers of compound **4**.

**Scheme 2.** Deuterium exchange of the 2-pyrrolidone ring of compound **4**.

Even though a number of fusarins [33], closely related compounds with fully unsaturated sidechains, have been isolated from different fungi, epolactaene [25] is the only other compound with the same sidechain as those of compounds **3** and **4** that has been reported so far. Unlike our previous studies [8–12] where active metabolites were found in fungi at the stationary phase of growth, which was typically after three weeks, antiplasmodial compound (**4**) in *B. dothidea* appeared in the log phase of growth after about two weeks and disappeared rapidly before the stationary phase of growth. These compounds may be biosynthetic intermediates rather than stable end products. Compounds **3** and **4** are the first natural fusarin type compounds with ether linkages at C-15.

Biological Activity

The 3:1 mixture of compounds **1** and **2** showed strong phytotoxic activity for both representative monocots and dicots (Table 3). Others have reported **1**, **2**, or both compounds to be phytotoxic [22,23,25], using bioassays that did not utilize whole plants. However, they were devoid of antiplasmodial activity (Table 4). The mixture of compound **1** & **2** was evaluated for toxicity against tumor cell lines SK-MEL, KB, BT-549, and SK-OV-3, as well as against the kidney epithelial cell line, LLC-PK11. The mixture showed cytotoxicity towards SK-MEL and SK -OV-3 cancer cell lines and kidney epithelial cells LLC-PK11, but no toxicity against KB and BT-549 cancer cell lines (Table 5). Previously, the enantiomers and racemate of compound **2** have shown moderate antibacterial activity against *Bacillus subtilis* and strong cytotoxic activity against human leukemia HL-60 cells that was comparable to the activity shown by the positive control, irinotecan [39].

**Table 3.** Phytotoxic activity of mixture of compounds 1 and 2 a.


<sup>a</sup> Concentration = 1 mg/mL.


**Table 4.** Antiplasmodial activity of compounds **3** and **4**.

<sup>a</sup> this fraction showed 100% inhibition at the lowest concentration tested (0.523 μg/mL). <sup>b</sup> Positive controls, IC50 = concentration causing 50% growth inhibition, NA = not active at the highest concentration tested (4.76 μg/mL), NC = no cytotoxic at the highest concentration tested (4.76 μg/mL), S. I. (selectivity index) = IC50 for cytotoxicity/IC50 for antiplasmodial activity.

**Table 5.** Cytotoxic activity [IC50 (μM)] of **1** and **2** mixture.


<sup>a</sup> Positive control, IC50 = concentration causing 50% growth inhibition, SK-MEL = human malignant melanoma, KB = human epidermal carcinoma, BT-549 = human breast carcinoma (ductal), SK-OV-3 = ovarian carcinoma, LLC-PK11 = kidney epithelial cells.

Compound **3** was found to be inactive in antiplasmodial assays. The fraction containing compound **4** showed moderate in vitro antiplasmodial activity against chloroquinesensitive (D6) and -resistant (W2) strains of *P. falciparum* (IC50 < 0.523 μg/mL) with no cytotoxicity to Vero cells (Table 4). Lack of sufficient material and instability prevented further biological studies on these compounds.

It is interesting that a related compound with a fully unsaturated sidechain, lucilactaene (Figure 1), has also shown potent antiplasmodial activity [35]. It has been found that the tetrahydropyran ring and methylation of the acid group of the sidechain are essential for the activity of this compound. Chemical instability and various other biological activities reported for this class of compounds [33–36] would preclude them from being potential antimalarial agents.
