*Review* **Structures and Bioactive Properties of Myrtucommulones and Related Acylphloroglucinols from Myrtaceae**

#### **Rosario Nicoletti 1,2, Maria Michela Salvatore 3, Pasquale Ferranti <sup>2</sup> and Anna Andolfi 3,\***


Academic Editors: Francesco Vinale and Maria Luisa Balestrieri Received: 2 December 2018; Accepted: 17 December 2018; Published: 19 December 2018

**Abstract:** Myrtaceae are a group of plants that include a number of renowned species used in ethnomedicine in many areas worldwide. Their valuable therapeutic properties have stimulated a fruitful research activity addressed to the identification of the bioactive components of their extracts yielding a great diversity of terpenes; polyphenols; and other exclusive products. Among the latter, starting with the discovery of myrtucommulone A from myrtle (*Myrtus communis*), a series of structurally-related acylphloroglucinol compounds have been characterized from several species that represent the basic active principles to be considered in view of possible drug development. Aspects concerning chemical and biological properties of these products are reviewed in the present paper.

**Keywords:** myrtucommulone; acylphloroglucinols; Myrtaceae; plant extracts; biological activities

#### **1. Introduction**

Myrtle (*Myrtus communis*) is a typical shrub of maquis and coastal bushes native of the Mediterranean area and Western Asia. It is well-known in traditional medicine, and for centuries its leaves and berries have found ethnomedical application in the treatment of several disorders of the digestive apparatus, as well as pulmonary and skin diseases [1,2]. More recently, experimental studies have provided an indication for a broader spectrum of pharmacological and therapeutic effects based on antioxidant, antiviral, antibiotic, antitumor, antidiabetic, hepatoprotective, and neuroprotective properties of extracts from this plant [2–4]. Many assorted compounds are considered as the possible bioactive components within the myrtle metabolome, such as terpenes occurring in the essential oils, α-tocopherol, anthocyanes, flavanols, and a series of acylphloroglucinols related to the myrtucommulones [4–10]. Particularly, the number of the latter compounds is continuously increasing as a result of a recent fruitful research activity carried out on other plant species belonging to the Myrtaceae in several independent laboratories worldwide. Structural aspects and the valuable bioactive properties that are pointed out for these compounds are reviewed in the present paper.

#### **2. Biological Sources**

The family Myrtaceae includes approximately 6000 species in 132 genera, with a wide distribution in tropical and warm-temperate regions of the world [11]. Until recently, these species were grouped in the two subfamilies of the Myrtoideae, including species with fleshy fruits and the Leptospermoideae, whose members produce dry capsules; however, this traditional arrangement has been disrupted in the current taxonomic schemes based on phylogenetic analysis resulting from DNA sequencing [12].

Historically relevant and well-known for its already-introduced multiple usage, *M. communis* represents the type species, and probably the best studied with reference to its biochemical properties. However, investigations in the field are developing more and more, enlarging the panorama of secondary metabolites produced by the Myrtaceae [2,8,13–15]. Particularly considered are aspects concerning the antibiotic properties, which have even led to proposing the use of some species in soil sanitization [16]. In this respect, a leading position pertains to the myrtucommulones, originally extracted from myrtle leaves [5,17]. More refined analytical studies have later shown their presence in fruits [7], which preludes the possible dietary intake of these products following the use of berries in gastronomy and in the preparation a digestive liquor typical of Sardinia [18,19].

After the pioneering reports concerning *M. communis*, compounds belonging to this class have been reported in species from other genera of Myrtaceae spread in the Australasian supercontinent. More particularly, species of *Callistemon*, which are now incorporated in the Linnaean genus *Melaleuca* [20], *Angophora, Baeckea*, *Corymbia*, *Eucalyptus*, *Kunzea*, *Lophomyrtus*, *Rhodomyrtus*, and *Syncarpia*, while *Myrciaria dubia* is endemic to the neotropical region (Table 1).

#### **3. Structures and Chemical Properties**

Within the large family of phloroglucinols [21], a series of alike natural products have been reported from plant species belonging to the Myrtaceae which are characterized by a molecular structure that is built on a phloroglucinol nucleus coupled with one or more syncarpic acid residues (Table 1). Myrtucommulone A (**1**), 4,4 -[(2,4,6-trihydroxy-5-isobutyryl-1,3-phenylene)bis (2-methylpropylidene)]bis(5-hydroxy-2,2,6,6-tetramethylcyclohex-4-ene-1,3-dione), represents the founder product of this series [17].


*Molecules* **2018** , *23*, 3370

3


**Table 1.** *Cont.*


**Table 1.** *Cont.*



6

#### *Molecules* **2018** , *23*, 3370

Despite the use of an inconsistent nomenclature based on the plant species from which the different compounds were originally extracted, all of these structures are clearly related to the myrtucommulone skeleton, and they most likely are assembled through common biosynthetic pathways. A plausible scheme considers the residues of isobutyrylphloroglucinol and isobutylidenesyncarpic acid deriving from the intramolecular Claisen reaction of a polyketidic intermediate obtained from three residues of malonyl CoA and one of isobutyryl CoA [45,53]. In particular, the cyclic intermediate 2-isobutyrylcyclohexane-1,3,5-trione is thought to form isobutyrylphloroglucinol through a double geminal dimethylation, passing through the formation of flavesone [45]. In this hypothesis, the syncarpic acid residue itself is derived from acylphloroglucinol. On the other hand, it has been previously reported that tetramethylation of the polyketidic chain is followed by intramolecular cyclization to form isobutylidenesyncarpic acid [54]. The latter standpoint is supported by what has been assessed for the biosynthesis of several well-known compounds, such as 5-methylorsellic acid [53].

Acylphloroglucinol and substituted syncarpic acid (one or two residues) components are probably coupled via the Michael reaction to form dimeric and trimeric structures [45]. This reaction might be non-enzymatic because the carbon between these residues may lead to the formation of a couple of enantiomers (without further chiral centers) or epimers (with further chiral centers). In fact, in a few cases the enantiomeric composition of myrtucommulones and related compounds is not fully defined. More thorough assessments in this respect are also desirable in view of ascertaining whether or not the bioactivity of myrtucommulones depends on chirality of the molecules.

By means of derivatization, CD spectroscopy and enantiomeric analysis, it has been shown that natural myrtucommulone A is a mixture of three stereoisomers, a racemate, and a meso form in a 1:1 ratio c.a., while myrtucommulone B (**2**) and nor-semimyrtucommulone (**34**) are both racemates. Furthermore, many isomeric forms may occur due to a tautomeric equilibrium of the enolized syncarpic moiety [42].

Further intramolecular reactions could involve the phenolic or ketonic group to form, respectively, pyranic or furanic rings. Prenylation has also been reported for compounds from this class [25,27,45,49,52].

On account of such a wide extent of structural variation, a convenient discussion on the properties of oligomeric acylphloroglucinols should be based on their grouping in subclasses, depending on the number of structural units and the eventual occurrence of additional cycles, such as furan and pyrane. In this respect we distinguished eight homogeneous groups whose members further differ upon the variation in the acyl functionalities occurring in the different moieties (e.g., isobutyryl, isovaleryl, methylbutyryl, and hexanoyl substituents).

Compounds in the first subclass (Figure 1) present a dimeric structure, modeled on semimyrtucommulone (**6**) and two closely related compounds (**3**–**4**) reported from *Kunzea* species. Other compounds in this group present additional cyclic structures. In particular, the bullataketals (**8**–**9**) have a phenyl-oxabycyclooctane system, myrtucommulone J (**21**) is characterized by a dipyrancyclopentanone moiety, while myrtucommuacetalone (**26**) contains an unprecedented bridged furochromene moiety.

**Figure 1.** Structures of compounds of the dimeric type.

The dimeric-monopyrane skeleton is shared by over one-third of the compounds examined in this review, which mainly differ in the assortment of the acyl functionalities (Figure 2). This group includes compounds that exhibit a methylated phenolic group on the phloroglucinol residue. Interestingly, myrtucommulone M (**25**) consists in two myrtucommulone B (**2**) moieties that are linked together through a methylene bridge to form a symmetrical structure.

**Figure 2.** Structures of compounds of the dimeric-monopyrane type.

The founder compound, myrtucommulone A (**1**), and the related myrtucommulones F (**14**) and H (**16**), presenting a hexanoyl residue on the phloroglucinol ring, are characterized by a trimeric structure (Figure 3). This kind of skeleton can be modified by additional cyclization, with the formation of mono and dipyranic analogues that are separated in the following subclasses.

**Figure 3.** Structures of trimeric compounds. Compound **1** is reported as (*R*,*R*)-stereoisomer.

Myrtucommulone C (**10**), with a trimeric-monopyrane structure (Figure 4), was isolated as single stereoisomer presenting an isobutanoylic residue. Eucalyptone G (**13**) is also a member of this subclass.

**Figure 4.** Structures of compounds of the trimeric-monopyrane-type.

In products of the trimeric-dipyrane-type (Figure 5), the presence of a pentacyclic structure can be observed where cycles may have an orientation that is similar to pentacene in the case of tomentosone C (**42**), while it is similar to pentaphene for other compounds (**11**, **12**, **15**, **17**, **20**, **30**, **31**). The different orientation is due to diverse phenolic and enolic/ketonic groups that are involved in the formation of pyranic rings of the trimer intermediate.

**Figure 5.** Structures of compounds of the trimeric-dipyrane type.

Rhodomyrtosone A (**18**) from *R. tomentosa* is the first natural product possessing a bisfurane fused ring (Figure 6). The dimeric compound **3** may represent its possible biosynthetic precursor, based on oxidation of the isobutyl side chain followed by the formation of benzofuran via cyclization and dehydration. Afterwards, several related compounds (**32**, **43**, **45**, **46**) have been characterized from other species in the Myrtaceae, indicating a possible wider occurrence of this peculiar structure.

**Figure 6.** Structures of compounds of the dimeric-bisfurane type.

Tomentosones A and B (**23**, **24**) are two epimers possessing a novel hexacyclic ring system (Figure 7) whose structures present a bisfuranic group and a hexacyclic ring.

**Figure 7.** Structures of compounds of the trimeric-bisfurane-pyrane-type.

Callistrilones (**40**, **41**, and **54**–**59**) represent the first syncarpic-phloroglucinol-monoterpene compounds that were isolated from a natural source. These compounds are characterized by the presence of a residue similar to phellandrene, which is fused through a furan ring to the phloroglucinol unit (Figure 8). Other compounds belonging to this class are the baefrutones, four of which (**64**–**67**) show the presence of an iridane skeleton, while **68** and **69** are sesquiterpene adducts.

**Figure 8.** Structures of compounds of the terpene-adduct type.

The available literature concerning Myrtaceae also reports the existence of products that are not classifiable as acylphloroglucinol oligomers, and are hence not included in this review. Particularly, monomers of either acylphloroglucinol (e.g., callisalignene A–C [25], xanchryone A–D [55], operculatol A–B [56]) or syncarpic acid (e.g., myrtucommulone K [35,57], callistiviminene A–O [58]), and flavonoids conjugated to a syncarpic acid residue (e.g., kunzeanones A–C [59], myrtocummunines A–D [51]).

A huge laboratory activity has been carried out on the synthesis of phloroglucinol compounds [60]. As an answer to the rising interest for pharmaceutical applications of myrtucommulones and related compounds, in the last decade several independent approaches have been developed in order to synthesize compounds belonging to this class. In particular, myrtucommulone A was first obtained from commercially available precursors [61], and later through stereoselective synthesis [62,63]. Other analogs of the series have been synthetically obtained (Table 2), particularly in the last couple of years, which possibly preludes further achievements in this respect in the short term.


**Table 2.** Myrtucommulone-related compounds obtained synthetically.

#### **4. Other Biological Sources**

Following recent discoveries concerning a number of valuable plant-derived drugs that have been also detected as fermentation products of endophytic fungal strains [72], myrtucommulones A and D (**1**, **11**) were extracted from the culture filtrates of a strain of *Neofusicoccum australe* endophytically associated with myrtle [73]. This finding represents the start point for new search terms addressed to a comparative elucidation of the genetic base of myrtucommulone biosynthesis, and possible applicative opportunities for more economic production to be exploited in view of drug development. Actually, while our laboratory investigations are in progress, we can anticipate that this extraordinary aptitude is shared with more endophytic strains isolated from myrtle in several locations (Figure 9). A few of these strains have been taxonomically identified and were found to belong to infrequent species, such as *Neosetophoma italica, Neocucurbitaria cava, Colletotrichum karsti,* and *Helminthosporium asterinum*. Following this preliminary assessment concerning myrtle, more evidences in this respect may be expected if metabolomic investigations are extended to endophytic microorganisms from other species in the Myrtaceae.

**Figure 9.** Detection of myrtucommulones A and B through HPLC-DAD analysis in culture extracts of endophytic fungi isolated from *M. communis*. (**a**) A1304B (*N. australe*); (**b**) A1306A (*H. asterinum*); and, (**c**) M15M2B (*N. italica*). Methods for culturing, extraction and chromatography have been previously reported [73].

Finally, quite interesting is the finding of myrtucommulone I (**17**), together with syncarpic acid and some more identified and unidentified alkylated phloroglucinols in propolis of the Australian stingless bee *Tetragonula carbonaria*, which has been taken into account to explain the antibacterial properties of this bee product [74].

#### **5. Biological Activities**

#### *5.1. Antibacterial Activity*

Myrtucommulones A and B (1–2), the founder compounds in this review, were preliminarily characterized for their antibiotic activity in agar plate diffusion assays against Gram-positive bacteria, namely *Staphylococcus aureus, Staphylococcus epidermidis*, *Bacillus subtilis, Bacillus pumilus, Enterococcus faecalis, Corynebacterium diphtheriae,* and *Corynebacterium xerosis* [75]. Similar assays provided concurrent indications in this respect for a few more products, such as the bullataketals (8–9) [31], myrtucommulones C–E (10–12) [33], and eucalyptone G (13) [29]. Afterwards, effectiveness against Gram-positive bacteria has been repeatedly reported for other related compounds, and circumstantiated with details concerning their minimum inhibitory (MICs) (Table 3). Conversely, assays against Gram-negative bacteria have been generally unfruitful, with a few questionable exceptions reporting inhibitory effects against *Escherichia coli* for rhodomyrtone A (7) [28,46], eucalyptone G [29], isomyrtucommulone B (5) [25], and callistenone H (38) [48]. Effects against Gram-negative species other than *E. coli*, such as *Pseudomonas aeruginosa, Salmonella typhi,* and *Shigella flexneri*, were reported for myrtucommulones C–E (10–12) [33].





15

 pneumoniae;

 Spy = Streptococcus

Spn = Streptococcus

 pyogenes; Ss = Streptococcus

 salivarius.

Particularly interesting is the activity against multi-resistant bacterial strains, especially methicillin-resistant *S. aureus* (MR*Sa*) and vancomycin-resistant *E. faecalis*, exhibited by products, such as callistrilone A (**40**) [45], and rhodomyrtone A (**7**), which did not induce resistance even after 45 passages in vitro [87]. Moreover, the latter product has recently displayed notable effects against both cell division and spore formation in *Clostridium difficile* [83].

Mechanisms of antibacterial activity have been quite thoroughly investigated in the case of **7**. Gene assays and proteomic profiling experiments in *B. subtilis* indicate that the cytoplasmic membrane is the main target of this compound. In *S. aureus,* it was reported to decrease the membrane potential at low doses, and to cause the release of ATP and cytoplasmic proteins. Local membrane damage was confirmed through lipid staining, and the protective effect displayed by saturated fatty acids was explained in terms of counteractive mending. It can be speculated that resistance to **7** by Gram-negative bacteria is due to the reduced penetration of the product through the outer membrane, and its neutralization by lipopolysaccharides. Moreover, interferences in the proteome and metabolome of *Streptococcus pneumoniae* were documented after exposure to **7**, consisting in a reduction of the levels of two enzymes (glycosyltransferase and glucose-1-phosphate uridylyltransferase) and of the uridine diphosphate derivatives of glucose, glucuronic acid, and *N*-acetyl-D-galactosamine, which participate in the synthesis of the capsule; as a matter of fact, a reduction in capsule size was confirmed through a colorimetric assay and electron microscopy [89].

Exposure of MR*Sa* to subinhibitory concentrations of **7** revealed a significant modulation of gene expression. Prominent changes involved genes encoding essential proteins for metabolic pathways and processes, such as membrane function, ATP-binding cassette transportation, and metabolism of amino acids, lipoproteins and nucleotides. Although the amino acid content of peptidoglycan in rhodomyrtone-treated MR*Sa* did not differ significantly from the control, data gathered on genes involved in the biosynthesis of amino acids and the diaminopimelate pathway indicate that peptidoglycan represents a target for bioactivity of this compound [91]. Moreover, proteome analyses in MR*Sa* revealed that exposure to subinhibitory concentrations of **7** affects the expression of several major functional classes of whole cell proteins, which act as surface antigens and virulence factors, or are involved in cell wall biosynthesis, cell division, oxidative stress, and various metabolic pathways. Transmission electron micrographs confirmed that **7** causes morphological and ultrastructural alterations in the bacterial cells, affecting the cell wall with abnormal septum formation and ensuing cell lysis [79].

The protein secretome was also investigated in a representative clinical MR*Sa* isolate, where the immunodominant antigen A, the staphylococcal secretory antigen, and other antigenic proteins involved in cell wall hydrolysis were downregulated after treatment with a subinhibitory concentration of **7**. Ribosomal and cytoplasmic proteins, such as glycerol phosphate lipoteichoic acid synthase and the stage V sporulation protein G (SpoVG), were found in the treated sample, while glycerophosphoryl diester phosphodiesterase, and another lipase precursor were absent. Finally, the finding of several cytoplasmic proteins in the supernatant of the treated cultures indicated impairment in the cell wall synthesis [80].

Again, the proteomic approach was followed in assays carried out with *Streptococcus pyogenes.* Various enzymes associated with important metabolic pathways, including alcohol dehydrogenase, glyceraldehyde-3-phosphate dehydrogenase, Xaa-His dipeptidase, ornithine carbamoyltransferase, putative *O*-acetylserine lyase, enolase (2-phosphoglycerate dehydratase), fructose-bisphosphate aldolase, and cysteine synthase, were strongly affected in a clinical isolate treated with **7** at half the MIC. Moreover, a series of known virulence factors, such as glyceraldehyde-3-phosphate dehydrogenase, CAMP factor, and exotoxin C, were downregulated [84]. It was also found that **7** reduces the synthesis of staphyloxanthin, a pigment promoting resistance to reactive oxygen species (ROS) whose shortage increases the bacterial susceptibility to H2O2 and singlet oxygen killing [92]. Furthermore, it could reduce biofilm formation by *S. aureus* and *S. epidermidis*, ensuing a reduction in the transcription signals of biofilm-related genes, with a different autolysin profile detectable in the treated cells [77,78]. Besides

formation, effects on the disorganization of established biofilms have been recently documented in assays carried out on *Propionibacterium acnes* [86].

Another documented molecular effect of **7** consists in the competitive binding of the tubulin homologue protein FtsZ. In fact, conformational changes in this main bacterial cell division protein were observed in both the (S)- and (R)- binding states of **7**. The compound reduced FtsZ polymerization by 36% and inhibited guanosine triphosphatase activity by up to 45%. However, at inhibitory concentrations, the compound had no effect on FtsZ localization in *B. subtilis*, and cells did not elongate after treatment. Higher concentrations of **7** affected the localization of FtsZ and of its membrane anchor proteins FtsA and SepF, showing that it did not specifically inhibit FtsZ but rather impaired multiple divisome proteins. Cell morphology was sometimes modified to a bean-like shape, possibly implying that the compound may also target cell wall synthesis, or maintenance [93].

At a more applicative level of investigation, it has been observed that subinhibitory concentrations of **7** affect the pathogenicity of oral bacteria (*S. aureus, Streptococcus mutans*) by impairing their adherence to both buccal epithelial cells and a polystyrene support in in vitro assays [84]. Moreover, **7** in a liposomal encapsulated preparation has been proposed for the treatment of bovine mastitis based on observations that were carried out in a bovine udder epidermal tissue model demonstrating remarkable effects against adhesion and invasion of the bacterial agents (*S. aureus* and *S. epidermidis*) [88].

Within a general framework of consistent antibacterial properties, data gathered in Table 3 emphasize that activity may appreciably change among the structural analogues. As an example, the higher efficacy of callistenone A (**27**) in comparison with its B isomer (**28**) is reported to depend on the point of attachment of the isovaleryl side chain [39]. This is confirmed by specific studies based on synthetic analogues, providing indication that the acyl tail of myrtucommulones and related compounds is a prerequisite for the antibacterial properties, and that its affinity for lipids is critical for activity more than its spatial dimension [94,95]. Additional clues in this respect also derive from assays concerning cytotoxicity, which in the case of tomentodiones is potentiated by an isobutyryl chain [50].

#### *5.2. Bioactivities against Other Microorganisms and Viruses*

The bioactivity of myrtucommulones and related acylphloroglucinols has been also investigated on fungi, with general negative results deriving from assays on species such as *Candida albicans, Cryptococcus neoformans, Microsporum gypseum,* and *Saccharomyces cerevisiae* [27,39].

Conversely, there are positive indications concerning antimalarial properties. In fact, antiplasmodial activity in the nanomolar range was reported for myrtucommulone A (**1**), and at a lower extent for semimyrtucommulone (**6**) [5,96]. Watsonianone B (**32**) displayed a potent activity against strain 3D7 of *Plasmodium falciparum* (IC50 0.289 μM), particularly against the young ring stages, coupled with selectivity towards a human embryonic kidney cell line (HEK 293) [40]. Strong effects against the chloroquine resistant strain Dd2 (IC50 0.10 μM) and low toxicity towards HEK 293 also characterize rhodomyrtosone F (**35**) [42], while a more moderate activity was assessed for rhodomyrtone A (**7**) against both 3D7 and Dd2 (IC50 1.84 μM and 4.00 μM, respectively) [30]. Moreover, tomentosone A (**23**) inhibited the growth of both chloroquine resistant and sensitive strains (IC50 1.49 μM and 1.0 μM, respectively), while its B analogue (**24**) was significantly less active [37]. Structural comparisons indicated that the syncarpic acid moiety is essential for antiplasmodial activity [97]. Additional data concerning antiplasmodial properties of the above products extracted from the flowers of *Angophora woodsiana* are reported in another paper by the same research group [30].

Antiviral effects were displayed in vitro by the mixture of compounds **3**–**4**, so far only extracted from two *Kunzea* species, based on the inhibition of the cytopathic effects of *Herpes simplex* type 1 (HSV-1) and Polio type 1 viruses [26]. Moderate effects in similar assays against HSV-1 have been also reported for callistrilones H–I (**38**–**39**) [49].

#### *5.3. Antioxidant and Anti-Inflammatory Activities*

Besides the possible applications in the treatment of infective diseases based on the above-reported effects, multiple observations concerning consistent antioxidant and anti-inflammatory properties have resulted from investigations that were carried out in several laboratories, representing an indication of a potential for a therapeutic use in the treatment of a series of disorders, ranging from allergopathies to cardiovascular diseases. With reference to the latter, myrtucommulone A (**1**) and semimyrtucommulone (**6**) were reported to exert powerful antioxidant properties during the degradation of cholesterol, preserving the LDL form from oxidative damage induced by copper ions, and inhibiting the increase of oxidative products deriving from polyunsaturated fatty acids [98,99]. At micromolar concentrations both compounds suppressed eicosanoid biosynthesis in vitro and in vivo by directly inhibiting cyclooxygenase (COX)-1 and 5-lipoxygenase. Moreover, they were successful in preventing the mobilization of Ca2+ in polymorphonuclear leukocytes, mediated by G protein signalling pathways, with the first compound acting at lower concentrations, and suppressed the formation of ROS and the release of elastase [100].

Binding affinity to the thyrotropin-releasing hormone (TRH) receptor-2, which is known to play a role in the phosphoinositide metabolism and is regarded as a potential therapeutic target to treat pain, was also reported for myrtucommulones A, D, and F–I (**1**, **11**, **14**–**17**) at micromolar concentrations [23]. After experiments carried out on both a human lung adenocarcinoma cell line (A549) and in a cell-free assay based on microsomal preparations of A549 cells stimulated with interleukin (IL)-1β, myrtucommulone resulted to be the first natural product to inhibit microsomal prostaglandin synthase-1 that efficiently suppresses prostaglandin formation without significant inhibition of cyclooxigenases, hence without displaying the typical side effects of non-steroidal anti-inflammatory drugs [101]. Furthermore, **1** exerted anti-inflammatory effects in the pleurisy model. In particular, a reduction was observed in the exudate volume, leukocyte numbers, lung injury, and neutrophil infiltration, and in a series of more specific effects mediated by enzymes and cytokines [102].

The interest for a potential therapeutic use of rhodomyrtone A (**7**) is also based on consistent properties that may prevent or delay the progression of inflammation in skin diseases, such as psoriasis. After stimulating human skin organ cultures with TNF-α and IL-17A to mimic skin inflammation, **7** significantly decreased inflammatory gene expression and the secretion of inflammatory proteins. Particularly, it inhibited TNF-induced extracellular signal–regulated kinases (ERK), c-Jun *N*-terminal kinases (JNK), the mitogen-activated protein (MAP) kinase p38, and phosphorylation of the NF-κB transcription factor p65, suggesting that it acts by modulating MAP kinase and NF-κB signalling pathways. Moreover, it reversed imiquimod-induced skin hyperplasia and epidermal thickening in mice [103]. The potential of **7** as an anti-psoriasis agent is further increased by its property to inhibit proliferation and to induce growth arrest and apoptosis in HaCaT keratinocytes [104]. The expression of pro-inflammatory molecules, including IL-1β, IL-6, TNF-α, and inducible nitric oxide synthase (iNOS) was enhanced in THP-1 monocytes that were stimulated with a high dose of heat-killed MR*Sa*. In contrast, monocytes stimulated with lower doses did not express these cytokines. However, in monocytes stimulated with heat-killed MR*Sa* at low doses, **7** significantly increased the expression of pro-inflammatory mediators, IL-6 and iNOS, and displayed some anti-inflammatory activity by reducing TNF-α expression. Treatment with **7** also significantly upregulated the expression of key pattern recognition receptor proteins (TLR2 and CD14). The ability of **7** to eliminate the resistant bacteria was observed within 24 h after treatment, following enhancement of the expression in monocytes of MR*Sa* recognition receptors, which possibly improved MR*Sa* clearance by modulating pro- and anti-inflammatory cytokine responses [105].

The possible application of the anti-inflammatory and immunomodulatory properties of **7** has been suggested in aquaculture as a result of observations carried out in vitro on head kidney macrophages of the rainbow trout (*Oncorhynchus mykiss*). In fact, exposure to **7** (1 μg mL−1) induced changes in the expression of genes involved in innate immune and inflammatory responses, particularly with reference to pro-inflammatory cytokines (IL-1β, IL-8, TNF-α), anti-inflammatory cytokines (IL-10, TGF-β), the antioxidant enzyme glutathione peroxidase 1, and other inducible enzymes (iNOS, COX-2, arginase). Co-exposure of **7** with lipopolysaccharides led to a downregulation of genes encoding for some of the above inflammation-related products and a reduction in ROS levels [106].

Myrtucommuacetalone (**26**) exhibited a significant inhibitory effect against production of nitric oxide, a ROS generated by NADPH oxidases in human peripheral blood phagocytes whose excess is associated with the pathogenesis of various diseases, such as colitis, diabetes, septic shock, and ischemic neuronal damage. Moreover, it inhibited the proliferation of T-cells, which is a relevant effect for the prevention or treatment of autoimmune disorders, such as Parkinson's disease, rheumatoid arthritis, and diabetes [38]. Additionally, inhibitory effects on nitric oxide production after lipopolysaccharide stimulation in murine macrophage cells (RAW 264.7) have been documented for baefrutones A–D (**64**–**67**) [52], and for tomentodione T (**61**) and rhodomyrtosones B, G, and I (**19**, **43**, **22**) [50].

#### *5.4. Cytotoxic and Antiproliferative Activities*

Apart from the diverse implications arising from the above properties, the observation of a pro-apoptotic effect induced on human cancer cell lines more directly introduces a possible relevance of myrtucommulones and associated acylphloroglucinols as antitumor drugs. Tretiakova et al. [107] first showed that, at micromolar concentrations, myrtucommulone A (**1**) induces apoptosis in several cancer cell lines, such as PC-3 (androgen-independent prostate carcinoma), LNCaP (androgen-dependent prostate carcinoma), KFR (rhabdomyosarcoma), HL-60 (acute promyelocytic leukemia), MM6 (acute monocytic leukemia), H9 (cutaneous T-cell lymphoma), DLD-1 (colorectal adenocarcinoma), and Jurkat (acute T-cell leukemia). Cell death occurred through the mitochondrial pathway involving the activation of caspase-3, -8, and -9, cleavage of poly(ADP-ribose)polymerase (PARP), release of nucleosomes into the cytosol, and DNA fragmentation. A lower cytotoxic effect was displayed on non-transformed human peripheral blood mononuclear cells and foreskin fibroblasts. Apoptosis appeared to be mediated by the intrinsic pathway, with the loss of the mitochondrial membrane potential in MM6 cells and the release of cytochrome c from mitochondria. Interestingly, Jurkat cells deficient in caspase-9 were resistant to apoptosis, and no processing of PARP or caspase-8 was evident. Conversely, in cell lines that were deficient in either CD95 signalling or caspase-8, myrtucommulone was still able to induce cell death and PARP cleavage.

A more direct indication that **1** induces apoptosis by triggering the intrinsic pathway and directly disrupting the mitochondrial functions resulted in assays carried out on HL-60 cells. In these cells, the compound caused the loss of the mitochondrial membrane potential and suppressed mitochondrial ATP synthesis, consequently inducing the adenosine monophosphate-activated protein kinase (AMPK), an energy sensor involved in apoptosis of cancer cells. More in detail, **1** acts as a protonophore that primarily dissipates the mitochondrial membrane potential through a direct structural interaction, and suppresses the proton motive force that impairs mitochondrial viability and activates AMPK due to lowered ATP levels [108]. The chaperonin heat-shock protein 60 (HSP60) also represents a molecular target of myrtucommulone A, which binds the protein and modulates its mitochondrial functions. Particularly, in a protein refolding assay the compound was found to prevent HSP60-mediated reactivation of denatured malate-dehydrogenase [109].

Myrtucommulone A (**1**) also induced apoptosis in several chronic myelogenous leukemia cell lines (K-562, MEG-01, KBM-5) in consequence of downregulation of anti-apoptotic proteins, as evidenced by nuclear fragmentation and PARP cleavage. Interestingly, the compound displayed differential toxicity, since peripheral blood mononuclear cells from healthy donors that were used as control were unaffected [110].

In further studies carried out on murine breast cancer cells (4T1), **1** was found to trigger apoptosis at micromolar concentrations through both the intrinsic and extrinsic apoptotic pathways. The compound mediated an increased expression of several apoptotic genes, such as Fas, FasL, Gadd45a, Tnf, Tnfsf12, Trp53, and caspase-4. Moreover, the results of a wound healing experiment

showed that it is also able to inhibit cancer cell migration [111]. All these effects were enhanced when treatment was operated in combination with epirubicin or cisplatin, evidencing a synergistic effect that could be exploited for setting more effective therapeutic schemes [112].

Another relevant effect of **1** that may contrast tumor development consists in a reduction of the expression of endoglin, a membrane glycoprotein that has a crucial role in angiogenesis. Treatment with this product reduced the chondrogenic potential in human mesenchymal stem cell (hMSC) lines, possibly as a consequence of the NF-κB p65 activation, while the adipogenic or osteogenic differentiation was not dramatically affected. The exploitation of these properties could be useful in targeted differentiation studies [113].

It is known that hMSCs can be observed in tissues surrounding tumors, where they could play a role in regulating cancer cell behaviour through paracrine signalling. Therefore, the modulation of their secretome is highly significant in view of attempts to control the disease. **1** was effective in modulating cytokine expression in hMSCs with a decrease of TNF-α, IL-6, IL-8, the vascular endothelial growth factor (VEGF), and the basic fibroblast growth factor (FGF-2), and in reducing the proliferation, migration, and clonogenicity of human bladder (HTB-9) and 4T1 cancer cells [114]. Moreover, when considering the prominent role that is played by the epithelial-mesenchymal transition in cancer progression and metastasis, the ability of **1** to interfere in this process again by modulating signalling pathways and inhibiting phosphorylation of multiple proteins represents a notable therapeutic property [115]. Besides reducing the viability and proliferation in HTB-9 cells, in cancer stem cells the compound downregulated the expression of markers associated to pluripotency and multipotency (i.e., NANOG, OCT-4, SOX-2, SSEA-4, TRA-1-60, CD90, CD73, and CD44), and decreased sphere-forming ability [116].

As an integration to data concerning efficacy, perspectives for a possible pharmaceutical use of myrtucommulones are corroborated by the absence of substantial cytotoxicity for non-malignant cells [107], and evidence of molecular stability in human and rat plasma [117].

Additional indications concerning the inhibitory effects on tumor cells have resulted in several laboratories from assays that were carried out with a number of compounds of the myrtucommulone series. Particularly, antiproliferative activity has been reported for bullataketals A and B (**8**–**9**) on murine leukemia cells (P388) [31], for myrtucommulones A and J (**1, 21**) on another prostate cancer cell line (DU145), HEpG2 (human liver carcinoma) and MT-4 (lymphocytic leukemia) [35], on PC3 and DU145 treated with a mixture of **1** and myrtucommulone D (**11**) [73], on HCT116 (human colorectal carcinoma) treated with **11** and isomyrtucommulone B (**5**) [25], on HeLa (human cervix uteri carcinoma) for tomentodiones S and T (**60**–**61**), rhodomyrtone A (**7**), and rhodomyrtosones A, B, G, and I (**18**, **19**, **43**, **22**) [50].

Rhodomyrtone A (**7**) is cytotoxic for several types of eukaryotic cells, and eryptosis induced in human erythrocytes progresses along with cell shrinkage, membrane blebbing, and phosphatidylserine translocation to the cell surface. The distinctive interactions with the cytoplasmic membrane assimilate **7** to amphipathic products, introducing it as a useful tool in studies on membrane physiology [118]. While confirming that it is responsible for membrane invaginations that form intracellular vesicles trapping a broad range of membrane proteins, Saeloh et al. [119] observe that **7** does not behave as a typical membrane-inserting molecule; in fact, it transiently binds to phospholipid head groups and causes a distortion of lipid packing, which explains membrane fluidization and curvature. However, both the transient binding mode and the ability to form protein-trapping vesicles are unique properties, which are possibly indicative of a peculiar mechanism of action.

More detailed investigations provide further insight in the antitumor properties of **7**, which has been recently characterized as an antimetastatic agent for the treatment of skin cancer cells after studies that were carried out on an epidermoid carcinoma cell line (A431). In fact, at subcytotoxic concentrations the compound reduced migration of tumor cells, as well as their adhesive and invasive ability. At the molecular level it was able to inhibit the focal adhesion kinase (FAK) and phosphorylation of protein kinase B (AKT), c-Raf, ERK1/2, and p38 involved in the downregulation of enzyme activities and the expression of matrix metalloproteinase (MMP)-2 and -9. Moreover, the compound increased the expression of TIMP-1 and TIMP-2, which are inhibitors of MMP-9 and MMP-2, respectively, and inhibited the expression and phosphorylation of NF-κB in a dose-dependent manner [120].

#### *5.5. Other Pharmacological Perspectives*

The α-glucosidase-inhibitory activity displayed by myrtucommulones C–E (**10**–**12**) could introduce therapeutic application of these compounds in another widespread disease, such as diabetes [33]. Moreover, rhodomyrtosone E (**33**) showed weak effects on the translocation to the plasma membrane of the insulin-responsive glucose transporter 4 (GLUT-4) protein, representing another attractive target for anti-diabetic drug development [41]. Finally, the strong inhibitory activity towards soluble epoxyde hydrolases that is exhibited by myrtucommulone B (**2**) and callistenone B (**28**) could be exploited for the treatment of a variety of pathological conditions [24].

Besides data resulting from specific assays carried out with the purified compounds, the widespread use in ethnomedicine of extracts of plant species in the Myrtaceae represents a reliable guide for the possible pharmaceutical applications of these products [121–123]. Based on its consistent antibacterial properties, the use of myrtle extracts has been proposed for the treatment of mild bacterial disorders such as vaginosis [124], and an ethanolic myrtle extract (Myrtacine®) has been registered for the treatment of acne lesions whose active principles are claimed to be myrtucommulones A and B [125,126]. For the latter dermatological use, rhodomyrtone A has been proposed in an innovative liposome encapsulation in order to overcome problems deriving from the poor water solubility [127]. The same kind of preparation has also been tested for zootechnical use [88].

#### **6. Conclusions**

Structures and properties of 69 myrtucommulone-related products that were extracted from plant species belonging to the Myrtaceae during the past 45 years have been reviewed in this paper. Considering that about half of these compounds have been discovered in the last two years, and that attention of the scientific community to the exploitation of natural resources of bioactive products is increasing more and more, it is predictable that their number is going to increase quite quickly in the future. Furthermore, in the aim to study more in detail aspects concerning the relationships between structure and bioactivity of these products, synthetic studies have been recently started in a few laboratories from which quite a high number of analogues have been obtained, indicating that these structural models may be used for the synthesis of novel variants with improved effects [94,95,128,129].

Advances in the exploitation of these valuable products require a more refined capacity to detect their occurrence in plants [130–132], and to perform a careful extraction and purification of the effective analogues. Alternative opportunities to obtain these products through controlled fermentation also represent a basic investigational line that is likely to attract the attention of leading actors in the field of drug discovery and development in the near future.

**Author Contributions:** Conceptualization, R.N. and A.A.; experimental data, P.F.; literature search, M.M.S.; writing—review and editing, R.N., M.M.S., A.A.

**Funding:** Financial support was provided by Ministero Italiano dell'Istruzione, dell'Università e della Ricerca (MIUR) through Finanziamento delle Attività Base della Ricerca (FFABR) 2017.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2018 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 (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Chlamyphilone, a Novel** *Pochonia chlamydosporia* **Metabolite with Insecticidal Activity**

**Federica Lacatena 1,†, Roberta Marra 1,†, Pierluigi Mazzei 2,3, Alessandro Piccolo 1,3, Maria Cristina Digilio 1, Massimo Giorgini 4, Sheridan L. Woo 4,5, Pierpaolo Cavallo 6,7, Matteo Lorito 1,4 and Francesco Vinale 1,4,\***


Academic Editor: Josphat Matasyoh

Received: 14 January 2019; Accepted: 14 February 2019; Published: 19 February 2019

**Abstract:** Metabolites from a collection of selected fungal isolates have been screened for insecticidal activity against the aphid *Acyrthosiphon pisum*. Crude organic extracts of culture filtrates from six fungal isolates (*Paecilomyces lilacinus*, *Pochonia chlamydosporia*, *Penicillium griseofulvum*, *Beauveria bassiana*, *Metarhizium anisopliae* and *Talaromyces pinophilus*) caused mortality of aphids within 72 h after treatment. In this work, bioassay-guided fractionation has been used to characterize the main bioactive metabolites accumulated in fungal extracts. Leucinostatins A, B and D represent the bioactive compounds produced by *P. lilacinus.* From *P. griseofulvum* and *B. bassiana* extracts, griseofulvin and beauvericin have been isolated, respectively; 3-*O*-Methylfunicone and a mixture of destruxins have been found in the active fractions of *T. pinophilum* and *M. anisopliae*, respectively. A novel azaphilone compound, we named chlamyphilone, with significant insecticidal activity, has been isolated from the culture filtrate of *P. chlamydosporia*. Its structure has been determined using extensive spectroscopic methods and chemical derivatization.

**Keywords:** secondary metabolites; beneficial microbes; pea aphid; azaphilones

#### **1. Introduction**

Alarm over the impact of pesticides on the environment and human health is increasing year after year, and rigorous pesticide registration procedures have been introduced. Through Directive 2009/128/EC, the European Community has severely restricted the use of synthetic pesticides in plant protection, and the new regulations have reduced the number of chemicals available in agriculture [1]. Novel pesticides, including natural product-based formulations, have been developed to counteract the evolution of resistance among plant pathogen and pest populations [2].

Microorganisms biosynthesize thousands of compounds displaying different biological activities, e.g., acting as antibiotics, therapeutic agents, toxins, hormones, etc. [3]. Thus, the exploitation of the microbial metabolome has become an important area of research to isolate novel natural products potentially useful for agricultural applications [3]. The improvement of screening technologies can be used to extend the search of new microbial active metabolites, i.e., through (i) upgrading of fermentation techniques, (ii) the development of new methods for detection, and (iii) genetic manipulation to create mutants able to produce qualitatively or quantitatively different metabolites [4].

Microorganisms have been extensively screened for antibiotic production; the main producers are fungi and bacteria, particularly the actinomycetes *Streptomyces* spp. [5]. A large number of manuscripts deal with microbial metabolites, which have demonstrated efficacy as crop protection agents, but only a few compounds have been marketed so far [2]. Examples of commercially available microbial metabolites with insecticide activity are: (1) abamectin and anthelmintic produced by the soil-dwelling actinomycete *S. avermitilis*; (2) milbemycin (also known as milbemectin), an insecticide and acaricide from *S. hygroscopicus* subsp. *aureolacrimosus*; (3) polynactins, secondary metabolites from the actinomycete *S. aureus*, isolated and applied as a mixture of tetranactin, trinactin and dynactin; (4) spinosad, a secondary metabolite from the soil actinomycete *Saccharopolyspora spinosa* [2].

The interaction between some fungal strains and the plant establish a molecular cross-talk in which fungal metabolites can act as elicitors that activate the expression of genes involved in plant defence response, and promote the growth of the plant [6]. Identification of new bioactive compounds may be obtained with a good microbial collection or isolating metabolites not expressed under standard laboratory conditions [6–8].

In this work, we show the results of a screening aimed to isolate microbial metabolites with insecticidal activity. Various genera of fungi, including some entomopathogenic species, were grown in liquid culture and their major secondary metabolites were investigated. Bioassay-guided fractionation used to isolate the bioactive metabolites allowed the identification of seven known compounds and a new metabolite named chlamyphilone. This compound was fully characterized by spectrometric analysis and chemical derivatization. The fungal metabolites have been screened for insecticidal activity against *Acyrthosiphon pisum* (Hemiptera: Aphididae), the pea aphid, whose high rate of increase makes it a useful model for screening [9].

#### **2. Results**

The microbes used in the present study includes various genera of fungi identified according to morphological features and molecular analyses (rDNA-ITS and β-tubulin gene sequencing). In particular, 5 isolates belong to the *Trichoderma* genus (*T. tomentosum* F19, *T. asperellum* CINO1, *T. harzianum* M10, *T. harzianum* F53, *T. velutinum* F28) and 4 to *Penicillium* (*P. chrysogenum* F5, *P. decumbens* F29, *P. griseofulvum* F11, *P. restrictum* F55). *Beauveria bassiana* BB1, *Chloridium virescens* F57, *Metarhizium anisopliae* MA3, *Paecilomyces lilacinus* 100379, *Pochonia chlamydosporia* B and *Talaromyces pinophilus* F36CF are present as single species.

All fungi have been grown in static conditions and the culture filtrates were extracted exhaustively with ethyl acetate. The crude extracts were fractionated by column chromatography to provide a total of 79 organic fractions. The extracts and all the fractions were tested at different concentrations on *Acyrthosiphon pisum*. Table 1 reports the percentage of mortality at 72h of the active extracts/fractions obtained from different fungal isolates. Only extracts obtained from *P. lilacinus*, *P. griseofulvum*, *P. chlamydosporia*, *B. bassiana*, *M. anisopliae* and *T. pinophilus* showed significant insecticidal activity (data obtained from the other fungal strains are not showed). Moreover, the bioassay-guided fractionation indicated that only one fraction per strain was responsible of a significant aphid mortality after 72 h exposure (Table 1).

Chemical characterization of the active fractions was obtained using NMR and/or LC-MS qTOF analyses. Fraction No. 3 of *P. lilacinus* (showing 30% aphid mortality; Table 1) was constituted by a mixture of leucinostatins A, B and D (**1**, **2** and **3,** respectively; Figure 1), with the molecular weights (MW) 1218.634 g/mol, 1204.607 g/mol and 1104.4799 g/mol, respectively (Figure S1).

**Table 1.** Insecticidal activity of fungal organic extracts and active fractions as mortality (%) of the pea aphid *Acyrthosiphon pisum* at 72 h after exposure. Different lower letters refer to significant differences (*p* < 0.05) among treatments at the same incubation time. Extracts from *T. tomentosum* F19, *T. asperellum* CINO1, *T. harzianum* M10, *T. harzianum* F53, *T. velutinum* F28, *P. chrysogenum* F5, *P. decumbens* F29, *P. restrictum* F55 and *C. virescens* F57 did not show significant insecticidal activity.


**Figure 1.** Chemical structures of leucinostatin A, **1**; leucinostatin B, **2**; leucinostatin D, **3**; griseofulvin, **4**; chlamyphilone, **5**; Ac-chlamyphilone, **6**; beauvericin, **7**; dextrusin B2, **8**; 3-*O*-Methylfunicone, **9**.

The mycotoxin griseofulvin (**4** in Figure 1) was determined as the main metabolite of *P. griseofulvum* and was isolated in the active fraction number 7 (showing 73,3% aphid mortality; Table 1). Figure S3 reports the mass spectrum of the isolated compound.

The residue recovered after the organic extraction (350 mg) of *P. chlamydosporia* culture filtrate was subjected to flash column chromatography, eluting with CH2Cl2/MeOH (90:10 *v*/*v*). Fractions showing similar thin-layer chromatography (TLC) profiles were combined and further purified by using preparative TLC separation (Si gel; CH2Cl2/MeOH 90:10 *v*/*v*). Both fractions and pure compounds were tested for insecticidal activity. Twelve milligrams of a novel compound, named chlamyphilone (**5** in Figure 1; 5 mg/L), were obtained as a yellow amorphous solid in pure form after TLC (Rf in CH2Cl2/MeOH 90:10 *v*/*v* = 0.8) and has: [α] <sup>25</sup> <sup>D</sup> −21.9◦ (c 1; CH2Cl2); UV (CH2Cl2) <sup>λ</sup> max (log <sup>ε</sup>) 219 (3.57), 332 (4.23). 1H and 13C NMR spectral data of chlamyphilone (in CDCl3) are presented in Table 2. LC-MS qTOF analysis detected the precursor ions at *m*/*z* 221.0814 (pseudomolecular ion [M + H]+); 463.1360 [M2 + Na]+; 243.0916 [M + Na]+; 221.0814 [M + H]+ (calcd. 221.0814); 203.1313 [M + H − H2O]+ (Figure S4). The results obtained by 13C NMR (Table <sup>2</sup> and Figure S6) and LC-MS qTOF analyses (Figure S4) were consistent with a compound having a molecular weight of 220.0814, corresponding to the molecular formula C12H12O4 with seven unsaturations.

**Table 2.** 1H and 13C NMR spectral data of chlamyphilone (in CDCl3).


\* All correlations represent 2 or 3 bond couplings. Abbreviation, s: singlet, d: doublet, t: triplet, dd: doublet of doublets.

The UV spectrum showed absorption peaks characteristic of the azaphilone family, which are natural products containing a 6*H*-isochromene-6,8(7*H*)-dione. Six carbon signals at 102.9, 114.4, 115.3, 145.2, 151.7, and 155.1 ppm revealed three double bonds, whereas signals at 196.3 and 196.5 ppm indicated the presence of two ketones. The other four signals in the 13C spectrum were all shifted upfield in the 12.7–83.11 ppm range. The DEPT data demonstrated that one of the protons in the molecule is bound to oxygen, and that in the molecule there are three CH3, two CH, and seven fully substituted C atoms. The 1H-1H COSY, HSQC analysis, and the chemical shift evaluation (Table 2) allowed the identification of the structural fragment. In Table 2 all the compound signals are reported.

The connectivity of the spin systems was deduced by a long-range 1H-13C heterocorrelated experiment that was obtained with the HMBC (Table 2). In particular, the most relevant HMBC correlations are reported in Figure 2, which implied that the structure of the metabolite **5** is 7-hydroxy-3,4,7-trimethyl-isochromene-6,8-dione. The configuration of chlamyphilone was evident from NOESY experiment. The MM-2 energy calculation (16.0 Kcal/mol) was run to find the most stable conformational model (Figure S12).

**Figure 2.** HMBC correlation of chlamyphilone.

To confirm the proposed structure, a sample of chlamyphilone was acetylated using acetic anhydride/pyridine (**6** in Figure 1). The 1H-NMR spectrum of the product (acetic acid 3,4,7-trimethyl-6,8-dioxo-7,8-dihydro-6H-isochromen-7-yl ester) contained one acetate resonance, confirming the presence of the hydroxyl group bound to the carbon at 83.1. The 13C-NMR spectrum contained an additional signal for the acetyl carbonyl resonance at 170.8 and for the acetate methyl carbon at 18.4 ppm.

Chlamyphilone (**5** in Figure 1) showed the highest insecticidal activity against *Acyrthosiphon pisum* with a median Lethal Dose (LD50) of 175 μg/mL and a minimal inhibitory concentration (MIC) of 150 μg/mL (Table 3).


**Table 3.** Insecticidal activity of chlamyphilone **5** at different concentrations. Treatments were evaluated as mortality (%) of the pea aphid *Acyrthosiphon pisum* at 72 h after exposure to **5**.

The insecticidal activity of the *B. bassiana* strain used in the present work was mainly dependent on the presence of beauvericin (**7** in Figure 1). This metabolite is a cyclohexadepsipeptide mycotoxin with the molecular formula C45H57N3O9 (Figure S13).

From *M. anisopliae* extract, destruxin B2 has been characterized (**8** in Figure 1). Destruxins are cyclic hexadepsipeptides composed of an α-hydroxy acid and five amino acid residues. Individual destruxins differ in terms of the hydroxy acid, *N*-methylation, and R group of the amino acid residues. Destruxin B2 has the molecular formula C29H49N5O7, as revealed by the MS spectrum (Figure S14).

Finally, the major active metabolite produced by *T. pinophilus* strain F36CF was 3-*O*-methylfunicone (OMF—**9** in Figure 1), a well-known γ-pyrone derivative, previously isolated from *Talaromyces* spp.

#### **3. Discussion**

In this work, a functional screening of fungal strains, preliminarily selected because of their ability to synthetize active metabolites, has been carried out. Bioassay-guided fractionation has been used to isolate microbial metabolites with insecticidal activity against the pea aphid *A. pisum* (Hemiptera: Aphididae), long used as a model to study plant-insect interactions [9]. This aphid is an important pest of several plants, is a phloem-feeding insect determining direct negative effects on plants in terms of

nutritive subtraction and injection of toxic saliva. Moreover, it has been involved in the transmission of diverse plant viruses [9].

The active fraction obtained from *P. lilacinus* was constituted by a mixture of leucinostatins A, B, and D. Leucinostatins are peptides containing an unsaturated fatty acid, nine amino acid residues, and a basic component joined together by amide linkages [10]. These antibiotic metabolites also showed antitumor and nematocidal activities and an uncoupling effect on rat liver mitochondrial function [10,11]. To the best of our knowledge, our work is the first report on the insecticidal activity of leucinostatins against *A. pisum*.

Griseofulvin was determined to be the main metabolite of *P. griseofulvum*. The insecticidal activity of this metabolite has been previously demonstrated in the corn earworm, *Helicoverpa zea* Boddie, and the fall armyworm, *Spodoptera frugiperda* (J. E. Smith), by oral administration in an artificial diet (250 ppm) [12–14]. In the case of soft-tegument aphids like *A. pisum*, the activity is probably topical as oral administration may be excluded because of their peculiar feeding pattern. In fact, aphids use their piercing, sucking mouthparts to feed on plant fluid [9].

The extract of *P. chlamydosporia* culture filtrate yielded 12 mg of a novel compound named chlamyphilone. This metabolite belongs to the class of azaphilones, which contain a 6*H*-isochromene-6,8(7*H*)-dione or an isoquinoline-6,8(2*H*,7*H*)-dione skeleton (and its substituted derivatives thereof). This is the first report on the isolation of an azaphilone metabolite from *P. chlamydosporia*. A similar azaphilone, named myxostiol, with plant growth regulating activity, has been previously isolated from *Myxotrichum stipitatum* [15]. Over 170 different azaphilone compounds, classified into 10 structural groups, have been isolated from fungi belonging to 23 genera [16]. Azaphilones display a wide range of biological activities, such as antimicrobial, antifungal, antiviral, antioxidant, cytotoxic, nematocidal, and anti-inflammatory [16]. Many of these properties may be explained by the reactions of azaphilones with amino groups, such as those found in amino acids, proteins, and nucleic acids, resulting in the formation of vinylogous c-pyridones [16–18]. To the best of our knowledge, this is the first evidence of the insecticidal activity of a metabolite belonging to the class of azaphilones. Recently, the involvement of small molecules (e.g., aurovertins) in the interaction between nematodes and *P. chlamydosporia* has been demonstrated [19].

An active beauvericin has been extracted from culture filtrate of *B. bassiana*. This metabolite is a mycotoxin whose insecticidal properties have been previously reported (e.g., against the wheat aphid *Schizaphis graminum* at 0.5 mg/mL) [20,21]. This molecule contains three residues, each with D-2-hydroxyisovaleric acid (Hiv) and L-*N*-methylphenylalanine linked alternately [21].

Destruxin B2 is the active molecule isolated from *M. anisopliae* extract. Destruxins exhibited a wide variety of biological activities, but are best known for their insecticidal and phytotoxic properties [22,23]. However, the effect of destruxin B2 against *Acyrthosiphon pisum* is reported here for the first time.

*T. pinophilus* strain F36CF produced 3-*O*-methylfunicone. This compound was downregulated in the presence of *T. harzianum* M10 [8], exhibited notable antibiotic and antitumor properties, and recently exhibited an insecticidal effect, thus expanding the biological activities of this compound [24]. Recently, talarodiolide, a new 12-membered macrodiolide, was isolated and characterized from the culture filtrate of a *T. pinophilus* strain [25]. This metabolite did not show insecticidal activity, highlighting that the strain collection is important to select the best producers of bioactive compounds.

#### **4. Materials and Methods**

**Fungal strains.** The microbes used in the present study were present in the fungal collection at Department of Agricultural Sciences, University of Naples Federico II (UNINA Collection) or isolated from different sources (Table 4).


**Table 4.** Fungal strains used in the present study for the isolation of metabolites with insecticidal activity against the pea aphid *A. pisum*.

\* UNINA Collection: fungal collection available at the Department of Agricultural Sciences/University of Naples Federico II, Naples, Italy.

The fungi were identified according to morphological features and molecular analyses. Briefly, an amount of 106 spores/mL was inoculated in a 250-mL Erlenmeyer Flasks containing 100 mL of sterile Potato Dextrose Broth (PDB, HIMEDIA, Mumbai, India). Each flask was incubated at 25 ◦C in an orbital shaker (150 r.p.m.) and fungi were left to grow for seven days. Mycelium was recovered, ground to a powder in liquid nitrogen, and used to perform the DNA extraction. Genetic analysis was carried out through PCR and sequencing of the rDNA Internal Transcribed Spacer (ITS) and β-tubulin gene, using the most common primers to identify fungal strains: ITS1 and ITS4 [26], as well as tub2 and BenA [27]. The PCR products were subjected to gel electrophoresis, excised, and sequenced [26,27]. Analysis of the ITS gave 99% of identity with GenBank sequences of the fungi reported in Table 4, confirming the identity of the microbes.

**Secondary metabolites production.** The fungi were maintained on potato dextrose agar (PDA, HIMEDIA) at room temperature and sub-cultured bimonthly. Liquid cultures were prepared in 5000 L-Erlenmeyer flasks containing 1 L of PDB and inoculated with five mycelial plugs (7 mm2) from a fresh PDA culture of each fungal strain. After 30 days of incubation at 25 ◦C in static conditions, the cultures were filtered through filter paper (Whatman No. 4) and the filtrate was analyzed by LC-MS qTOF [28].

**Extraction and isolation of fungal secondary metabolites.** The culture filtrates were acidified to pH 4 with 5 M HCl and extracted exhaustively with ethyl acetate (EtOAc). The combined organic extracts were dried (Na2SO4) and solvent eliminated under reduced pressure at 35 ◦C (Rotavapor RV 10 IKA® - Werke GmbH & Co. KG, Staufen, Germany). The residues recovered were fractionated by column chromatography (silica gel; 200 g) eluted with different eluents (Table 5) and the homogeneous fractions were collected as reported in Table 5. For thin layer chromatography (TLC) the following solvents were used: dichloromethane (CH2Cl2)/methanol (MeOH) 90:10; 80:20 (*v*/*v*), chloroform (CHCl3)/MeOH 90:10; 80:20 (*v*/*v*); EtOAc/petroleum ether 90:10; 80:20 (*v*/*v*) [28]. All fractions were tested for insecticidal activity against *A. pisum*, and, where necessary, further purified by preparative TLC (Table 5).


**Table 5.** List of solvents used for chromatographic separations (column chromatography or preparative TLC) and total number of homogeneous collected fractions.

\* EtOAc: ethyl acetate; CH2Cl2: Dichloromethane; MeOH: Methanol.

**General experimental procedures.** The isolated molecules were solubilised in 700 μL of deuterated chloroform (99.8% CDCl3 – Sigma-Aldrich, Darmstadt, Germany) and transferred into a stoppered NMR tube (5 mm, 7, 507-HP-7, NORELL, Morganton, NC, USA) where remaining void volume was gently degassed by a N2 flux. Proton and carbon solvent signals were used as reference to calibrate both 1H and 13C frequency axes. A 400 MHz Bruker Avance spectrometer (Bruker Co., Billerica, MA, USA), equipped with a 5 mm Bruker Broad Band Inverse probe (BBI), working at the 1H and 13C frequencies of 400.13 and 100.61 MHz, respectively, was used for the NMR measurements (at 25 ±1 ◦C).

Monodimensional 1H and 13C acquisitions were conducted as follows: proton spectra were acquired with 2 s of thermal equilibrium delay (number of scans = 64), a 90◦ pulse length 7.7 μs, 50 transients and 16 ppm (6410.2 Hz) as spectral widths, whereas proton-decoupled carbon acquisitions were executed by both inverse-gated and DEPT 135◦ pulse sequences, adopting 7 and 5 s of equilibrium delay, 12,500 and 2400 transients, respectively, and a spectral width of 250 ppm (25.152 KHz). A time domain of 32,768 points was adopted for all cited mono-dimensional experiments. Homo-nuclear 1H-1H COSY (COrrelation SpectroscopY), TOCSY (TOtal Correlation SpectroscopY), NOESY (Nuclear Overhauser Enhancement SpectroscopY), and hetero-nuclear 1H–13C HSQC (Hetero-nuclear Single-Quantum Correlation) and HMBC (Hetero-nuclear Multiple Bond Coherence) experiments (2D) were used for structural identification of metabolites. 2D homo- and heteronuclear spectra experiments were acquired with 48 and 80 scans, respectively, 16 dummy scans, a time domain of 2k points (F2) and 256 experiments (F1). TOCSY and NOESY experiments were conducted with a mixing time of 80 and 1000 ms, respectively, while HSQC and HMBC experiments were optimized for 145 Hz short and 6.5 Hz long range JCH couplings, respectively. All executed 2D experiments were gradient enhanced, except for the TOCSY acquisition. A Qsine weighting function associated to a magnitude mode was used to process NOESY spectrum with the purpose to emphasize the weak cross-peaks and minimize the noise artefacts. The free induction decay (FID) of mono-dimensional spectra was multiplied by an exponential factor corresponding to 0.1 Hz, for 1H and 13C acquisitions, and to 1 Hz for DEPT 135◦ experiment. All above mentioned spectra were baseline corrected and processed by using Bruker Topspin Software (v.4.0.2).

LC-MS/MS Q-TOF analysis were done on an Agilent HP 1260 Infinity Series liquid chromatograph equipped with a DAD system (Agilent Technologies, Santa Clara, CA, USA) coupled to a Q-TOF mass spectrometer model G6540B (Agilent Technologies). Separations were performed on a Zorbax Eclips Plus C18 column, 4.6 × 100 mm, with 3.5 μm particles (Agilent Technologies). The analyses were done at a constant temperature of 37 ◦C and using a linear gradient system composed of A: 0.1% (*v*/*v*) formic acid in water, and B: 0.1% (*v*/*v*) formic acid in acetonitrile. The flow was 0.6 mL/min, 95% A graduating to 100% B in 12 min, 100% B 12-15 min, 95% A 15-17 and equilibrating 95% A 17–20 min. The UV spectra were collected by DAD every 0.4 s from 190 to 750 nm with a resolution of

2 nm. The MS system was equipped with a Dual Electrospray Ionization (ESI) source and operated with Agilent MassHunter Data Acquisition Software, rev. B.05.01 in the positive or negative mode. Mass spectra were recorded in the range *m*/*z* 100–1600 as centroid spectra, with 3 scans per second. Two reference mass compounds were used to perform the real-time lock mass correction, purine (C5H4N4 at *m*/*z* 121.050873, 10 μmol/L) and hexakis (1H,1H, 3H-tetrafluoropentoxy)-phosphazene (C18H18O6N3P3F24 at *m*/*z* 922.009798, 2 μmol/L). The capillary was maintained at 4000 V, fragmentor voltage at 180 V, cone 1 (skimmer 1) at 45 V, Oct RFV at 750 V. Gas temperature was 350 ◦C during the run at 11 L/min, and the nebulizer was set at 45 psig. The injected sample volume was 5 μL.

MS/MS spectra were simultaneously recorded for confirmation purposes of new compounds, using the operating parameters described above, unless otherwise stated. The instrument was operated in the range *m*/*z* 100-1000, recording two spectra per second in targeted acquisition mode (targeted mass: 244.1197, Z = 1, RT 5.88 ±0.5 min). The sample collision energy was set to 20 V.

LC-MS data were evaluated using MassHunter Qualitative Analysis Software B.06.00 and compared to known compounds included in an in-house database. The database contains information of about 4000 known secondary metabolites isolated from more than 80 different fungal genera, and recorded according to their name, molecular formula, monoisotopic mass and producing organism. Positive identifications of fungal metabolites were reported if the compound was detected with a mass error below 10 ppm and with a sufficient score. Standards were used to confirm the chemical identifications.

UV spectra were recorded with a V-730 UV-Visible Spectrophotometer JASCO (Mary's Court, Easton, MD, USA). Column chromatography was performed using silica gel (Merck silica gel 60 GF254; Merck, Darmstadt, Germany), and TLC with glass pre-coated silica gel GF254 plates (Merck Kieselgel 60 GF254, 0.25 mm). The compounds were detected on TLC plates using UV light (254 or 366 nm) and/or by dipping the plates in a 5% (*v*/*v*) H2SO4 solution in ethanol followed by heating at 110 ◦C for 10 min [28,29].

**Acetylation of chlamyphilone.** Acetic anhydride (40 μL) was added to chlamyphilone (**1**, 2.5 mg), dissolved in dry pyridine (80 μL 2.5 mg), and the residue was purified by preparative TLC on silica gel (petroleum ether/aceton, 20:80, *v*/*v*) to yield the acetyl derivative **6** (1.4 mg, 49%) [29]. The reaction was monitored by LC-MS analysis (Figure S1).

**In vivo insecticidal assay.** Insecticidal activity of organic extracts, each column fraction and pure metabolites were tested against the pea aphid *Acyrthosiphon pisum*. Aphids were reared on *Vicia faba* var. *aguadulce* in a growth chamber (20 ± 1 ◦C, 70% RH, 18 h light/6 h dark photoperiod). Small populations were synchronized to obtain newborn nymphs every 24 h. Third-instar nymphs were dipped in the assay solution for 10 s and put on a paper towel to dry. Then, each insect was carefully transferred by a soft paintbrush on a leaf plug [30,31]. Twenty treated (or untreated—solvent control) 3rd-instar nymphs were placed on two circular (35 mm diameter) leaf plugs on 2% water agar layer in a Petri dish, in order to keep the leaf turgid. Plates were incubated at the climatic conditions described above and the number of dead aphids was assessed 24, 48, and 72 h after treatments. Extracts, fractions, and pure compounds were tested at 500, 400, 300, 200, 100, and 50 μg/mL. Each metabolite solution was also tested for phytotoxicity on *V. faba*. The aphid mortality has been calculated in each treatment as the mean value ± standard deviation, in three replicates for each time point (24, 48, and 72 h). The experiments were repeated at least twice.

**Statistical analysis.** Data analysis was performed with SPSS 11.0 software (Statistics for Windows Version 24.0, IBM Corp., Armonk, NY, USA), and statistical analysis was done using one-way analysis of variance (ANOVA). The Least Significant Difference (LSD) post hoc test with *p* < 0.05 was used to analyse the multiple comparisons. The mean values between different treatments at the same time point and the mean values between the same treatments at different time points were compared.

#### **5. Conclusions**

In this work six known metabolites with activity against the pea aphid *A. pisum* have been isolated from culture filtrates of selected fungal strains; for some of them this is the first report of insecticidal activity (Table S1). Moreover, a novel anti-aphid natural product, named chlamyphilone, has been isolated form the culture filtrate of *P. chlamydosporia* and fully characterized. The isolation of natural products with beneficial effects for plants may help to formulate novel secondary metabolites–based biopesticides, which represent promising alternatives to synthetic chemicals in agriculture. Further investigations should optimize extraction protocols and define environmental parameters that can affect the commercial formulations based on these natural compounds.

**Supplementary Materials:** The following are available online, Figures S1–S11 and S13-S14: NMR and MS spectra of compounds. Figure S12: Conformational model of the new compound named chlamyphilone. Table S1: Biological activity of the isolated compounds.

**Author Contributions:** F.V. and F.L. conceived the experiments and analyzed all the results. R.M. analyzed all the results and characterized the fungal strains. M.C.D. and M.G. conducted with F.L. the experiments on insects. M.L., P.C., F.L., and S.L.W. conducted the experiments with fungal strains. P.M., A.P. and F.V. characterized all the isolated metabolites. All authors reviewed the manuscript.

**Funding:** This work was supported by the following projects: MIUR–PON [grant number Linfa 03PE\_00026\_1], [grant number Marea 03PE\_00106].

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


**Sample Availability:** Samples of the compounds **1**-**9** are available from the authors.

© 2019 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 (http://creativecommons.org/licenses/by/4.0/).

## *Article* **The Issue of Misidentification of Kojic Acid with Flufuran in** *Aspergillus flavus*

**Marina DellaGreca 1, Gaetano De Tommaso 1, Maria Michela Salvatore 1, Rosario Nicoletti 2,3, Andrea Becchimanzi 3, Mauro Iuliano <sup>1</sup> and Anna Andolfi 1,\***


Academic Editors: Francesco Vinale and Maria Luisa Balestrieri Received: 11 April 2019; Accepted: 30 April 2019; Published: 2 May 2019

**Abstract:** In the course of investigations on the complex phenomenon of bee decline, *Aspergillus flavus* was isolated from the haemocoel of worker bees. Observations on the metabolomic profile of this strain showed kojic acid to be the dominant product in cultures on Czapek-Dox broth. However, an accurate review of papers documenting secondary metabolite production in *A. flavus* also showed that an isomer of kojic acid, identified as 5-(hydroxymethyl)-furan-3-carboxylic acid and named flufuran is reported from this species. The spectroscopic data of kojic acid were almost identical to those reported in the literature for flufuran. This motivated a comparative study of commercial kojic acid and 5-(hydroxymethyl)-furan-3-carboxylic acid, highlighting some differences, for example in the 13C-NMR and UV spectra for the two compounds, indicating that misidentification of the kojic acid as 5-(hydroxymethyl)-furan-3-carboxylic acid has occurred in the past.

**Keywords:** kojic acid; flufuran; 5-(hydroxymethyl)-furan-3-carboxylic acid; *Aspergillus flavus*; bees; immunomodulation

#### **1. Introduction**

Fungi have evolved the capability to produce a great number of secondary metabolites involved in the improvement of their ecological fitness, and many of them play important biological roles as virulence factors, chemical defense agents, developmental regulators, insect attractants, and chemical signals for communication with other organisms. On these properties is founded the pharmacological exploitation of many products as antibiotic, antiviral, antitumor, antihypercholesterolemic, and immunosuppressant agents [1–5]. In this respect, fungi are prime targets of a vigorous investigational activity, based on the employment of the most advanced analytical and structure elucidation techniques [6,7].

A wide array of fungal secondary metabolites has been ascribed to the ascomycete genus *Aspergillus*, well-known for its ubiquity and cosmopolitan distribution [7,8]. The species *Aspergillus flavus* is well-known as a foodstuff contaminant and a mycotoxin producer, and in this respect its metabolomic profile has been quite well characterized [9,10]. However, this fungus has been also reported in association with plants and animals in many different environments [11,12]. Particularly, it has been directly isolated from bee (*Apis mellifera*) individuals in different developmental stages and health conditions [13]. Although reported as the agent of stonebrood disease, further elucidation is required if

*A. flavus* basically behaves like an entomopathogen, or if its relationships with bees are more complex and eventually involve modulation of the immunitary response to other pathogens and parasites of these insects. Generally, the fungus is considered an opportunistic pathogen of immunocompromised individuals, gaining access to the host through ingestion, or taking advantage from the interaction of bees with other pathogens and parasites which negatively affect host immunocompetence and cuticle integrity [13–16].

During investigations on the honeybee colony collapse, a multifactorial syndrome mainly related to the compresence of immunosuppressive viruses and *Varroa destructor*, a parasitic mite which is also known as a possible vector [16–18], strains of *A. flavus* were repeatedly isolated from the haemocoel of worker bees.

The present paper reports the identification of kojic acid (KA) as the main metabolite obtained from this source. Furthermore, revision of the structure of flufuran, previously characterized as 5-(hydroxymethyl)furan-3-carboxylic acid from *A. flavus* and other fungal species is proposed, based on a comparison of spectroscopic data of commercially available compounds and some derivatives.

#### **2. Results**

#### *2.1. Comparison of NMR Data Obtained from KA* (**1**) *and 5-(Hydroxymethyl)furan-3-carboxylic Acid* (**2**)

A white solid was obtained from the extraction with ethyl acetate (EtOAc) of the culture filtrate of strain AB1EET of *A. flavus*, which consisted in a main metabolite, as deduced by its 1H and 13C-NMR spectra. NMR data collected for the extracted compound showed significant similarities with proton and carbon chemical shifts reported in literature for both KA (**1**) [19] and 5-(hydroxymethyl)furan-3-carboxylic acid (**2**) [20], a compound characterized from *A. flavus* and other fungal species (Table 1). A full analysis of the previous reports revealed discrepancy between the carbon chemical shifts of furan compounds [21] and those assigned for 5-(hydroxymethyl)furan-3-carboxylic acid. In order to clarify this issue, commercially available KA and 5-(hydroxymethyl)furan-3-carboxylic acid were submitted to spectroscopic and potentiometric investigations.


**Table 1.** Previous reports dealing with flufuran identification.

<sup>1</sup> This paper is in Chinese and could not be examined in detail. <sup>2</sup> 7-*O*-Acetyl derivative was also isolated. <sup>3</sup> Methyl ester was also isolated.

The comparison of 1H-NMR spectra, recorded in CD3OD, showed no significant differences between 5-(hydroxymethyl)furan-3-carboxylic acid and KA (Figures 1 and 2). Notwithstanding, the 1H-NMR spectra recorded in DMSO*d*<sup>6</sup> for **2** revealed the coupling between the proton of the hydroxyl group in C-7 and protons of the methylene group which resonate as triplet and doublet, respectively, at δ 5.31 and 4.40 (*J* = 5.8 Hz) and the presence of the signal of proton of the carboxylic group at δ 12.60 (Figure 3).

**Figure 1.** 1H-NMR spectra of KA (**1**, red) and 5-(hydroxymetyhyl)-furan 3-carboxylic acid (**2**, green) recorded at 400 MHz in CD3OD.

**Figure 2.** Proton and carbon chemical shifts of KA (**1**) and 5-(hydroxymethyl)furan-3-carboxylic acid (**2**). Data reported in red were erroneously assigned to flufuran. The spectra were recorded in CD3OD.

**Figure 3.** 1H-NMR spectra of KA (**1**, red) and 5-(hydroxymetyhyl)-furan 3-carboxylic acid (**2**, green) recorded at 400 MHz in DMSO*d6*.

The main differences between **1** and **2** were observed, as expected, in 13C-NMR spectra. The C-3, C-5 and C-6 carbons of **2** resonate at δ 120.1, 156.4 e 165.1 (Figure 2 and Figure S1), respectively, assigned by 2D-NMR spectra (Figures S2 and S3). However, the 13C-NMR spectrum of flufuran showed chemical shift values identical to KA (Figure 2), indicating a misinterpretation of the flufuran structure.

#### *2.2. Comparison of Potentiometric, UV, and MS Spectrophotometric Measurements*

The acid-base behavior of KA and 5-(hydroxymethyl)furan-3-carboxylic acid was determined by potentiometric and spectrophotometric methods. The calculated protonation constants cologarithm values are 7.68 ± 0.05 and 4.03 ± 0.05, for compound **1** and **2** respectively. Potentiometric data are reported in Figure 4.

**Figure 4.** *Z*H(pH) function in 0.1 M NaClO4 for KA (squares) and 5-(hydroxymethyl)furan-3-carboxylic acid (circles). The solid lines are calculated with protonation constants cologarithm value 7.68 (on squared) and 4.03 (on circles).

In particular KA constant is very similar to that in 0.1 M KCl (7.7) [33]. While no value on the protonation of 5-(hydroxymethyl)furan-3-carboxylic acid had been previously reported.

UV spectra of the pure compounds, recorded in water in the interval 200–400 nm, show λmax at 216 and 270 nm for KA, and at 240 nm for **2**. Similar values were observed when the spectra were recorded in methanol.

In agreement with acid-base behavior of both compounds, measurements conducted in pH function confirmed that KA spectra are highly influenced by pH [34]. In fact, absorbance value at 315 nm increases from 3 to 9 pH range, instead the peak at 270 nm decreases in the same interval. Conversely, UV spectra of compound **2** present minimal variation in function of pH (Figure 5). Finally with pH increasing, KA shows a isosbestic point at 285 nm, while 5-(hydroxymethyl)furan-3-carboxylic acid presents one at 240 nm.

**Figure 5.** (**A**) UV spectra of 2.0 <sup>×</sup> 10−<sup>3</sup> M in 0.1 M NaClO4 for KA. (**B**) UV spectra of 2.0 <sup>×</sup> 10−<sup>3</sup> M in 0.1 M NaClO4 for 5-(hydroxymethyl)furan-3-carboxylic acid (optical path 0.2 cm).

The fragmentation pattern observed in MALDI TOF/MS spectra evidenced relevant analogies between the two compounds, as to be expected from considering the nature of the respective functional groups. In fact, from peaks corresponding to the protonated ions [M+H]<sup>+</sup> at *m*/*z* 143 of **1** and **2**, fragments deriving from the loss of OH, CHO, and COOH groups were observed in both cases.

#### *2.3. Comparison of KA and 5-(Hydroxymethyl)furan-3-carboxylic Acid Derivatives*

Considering reports concerning the presumptive isolation of the 7-*O*-acetyl derivative of flufuran [25] and of its methyl ester [29], in this work the differences between derivatives of **1** and **2** obtained from common acetylation and methylation reaction were also evaluated.

The proton spectrum of 5-(acetoxymethyl)furan 3-carboxylic acid (**4**, Figure S4), produced after acetylation of **2**, showed the down shift of Δδ 0.55 of the signal assigned to the methylene group C-7 which resonates as singlet at δ 5.08, and the presence of a further singlet at δ 2.07 assigned to the CH3 of the acetyl group (Figure 6 and Figure S4). As discussed in relation to **1** and **2** the differences between the 1H-NMR spectra of **4** and 7-*O*-acetylflufuran [25] are not so marked compared to the ones of 13C-NMR spectra (Figure 6 and Figure S5).

**Figure 6.** Proton and carbon chemical shifts of 7-*O*-acetylKA, 5,7-*O,O'*-diacetylKA, 5-*O*-methylKA, 5-*O*-methyl-7-*O'*-acetylKA (**3**, **5**, **6**, and **8**); 5-(acetoxymethyl)furan-3-carboxylic acid, methyl 5-(hydroxymethyl)furan-3-carboxylate, methyl 5-(acetoxymethyl)furan-3-carboxylate (**4**, **7**, and **9**).

From the data reported on the diacetylated flufuran [22], it is possible to exclude its formation from **2**; most likely the described compound is 5,7-*O,O*- -diacetylKA (**5**, Figure 6), derived from **1**.

The treatment of **1** and **2** with diazomethane led to the production of 5-*O*-methylKA (**6**) and methyl 5-(hydroxymethyl)furan-3-carboxylate (**7**, Figure 6) as deduced from 1H and 13C-NMR spectra. In fact, NMR spectra of **7** (Figures S6 and S7) showed further singlets, respectively, at δ 3.83 and at δ 50.6 attributed to methoxy group. As observed for the other compounds under examination, the comparison of 13C-NMR spectra of methyl derivatives of **1** and **2** revealed significant differences (**6** and **7**, Figure 6).

Finally, the preparation of 5-*O*-methyl-7-*O'*-acetylKA (**8**) and methyl 5-(acetoxymethyl)furan-3-carboxylate (**9**) and the interpretation of their one-dimensional NMR spectra (Figures S8 and S9) allowed to confirm the derivative structures, and to establish that the derivative i.e., methyl 5-(acetoxymethyl)furan-3-carboxylate) reported by Evidente et al. [22] is undoubtedly **8**.

#### **3. Discussion**

The furan compound 5-(hydroxymethyl)furan-3-carboxylic acid was identified for the first time as a natural product in cultures of *Polyporus ciliatus*[20], and afterwards reported from other fungi (Table 1). Although furan derivatives are not an extensive class of fungal compounds [35], over the past few years the literature concerning these natural products has been enriched by new reports [36–40]. According to their biosynthetic origin, the structure of natural furan derivatives shows the presence of furan rings substituted in position 2 and 5 [35]. Flufuran represented an exception to such a general model.

In the present paper, we produce evidence that in most of the previous reports the product identified as flufuran was instead KA. In addition to being known as a typical secondary metabolite of both *A. flavus* and *A. oryzae*, KA has been reported from several *Penicillium* spp. [41]. Other fungal species are claimed as producers of some derivatives [42], indicating that biosynthetic ability for this compound might be more widespread. To the best of our knowledge, our amendments to previous findings represents the first evidence of its production in important fungal genera, such as *Fusarium* and *Pestalotiopsis*, which deserve further confirmation also with reference to correct taxonomic identification of producing strains. In view of the present revision, the bioactivities erroneously assigned to flufuran should be instead referred to KA, integrating the known profile of biological properties of this γ-pyrone compound [43]. Actually, these data show that KA and 5-(hydroxymethyl)furan-3-carboxylic acid are susceptible of misidentification. Particularly, some instrumental techniques commonly used for the identification of secondary metabolites in fungal extracts (i.e., 1H-NMR and ESI MS data) are not able to distinguish between these two compounds. However, huge differences can be observed between **1** and **2** in the UV spectra obtained with different solvents and pH. The acid-base properties are also valid to differentiate between **1** and **2**. Finally, observations concerning acetyl and methyl derivatives also confirmed the misidentification of KA with flufuran.

#### **4. Materials and Methods**

#### *4.1. General Experimental Procedures*

1H and 13C-NMR spectra were recorded at 400 and at 100 MHz, respectively, in CD3OD unless otherwise noted, on a Bruker spectrometer (AscendTM400) (Bremen, Germany); the same solvent was used as internal standard. Potentiometric titrations were performed in an air-bath thermostat kept at (25.00 ± 0.05) ◦C. A programmable computer-controlled Data Acquisition Unit 34970A, (Agilent Tecnologies Inc., Santa Clara, CA, USA) was used to perform the potentiometric measurements. The glass electrodes were Metrohm (Herisau, Switzerland) of 60102-100 type, and Ag/AgCl electrode was utilized as reference. The EMF values were measured with a precision of ± 0.01 mV using a Keithley 642 type Digital Electrometer (Tektronix Inc., Beaveron, OR, USA). UV–VIS spectra were recorded by model Cary 5000 Spectrophotometer by Varian (Palo Alto, CA, USA) from 200 to 600 nm (optical path 0.2 cm) at 25.0 ◦C, under a constant flow of nitrogen.

MALDI-TOF/MS spectra were acquired by a 4700 Proteomics Analyzer (Applied Biosystems, Framingham, MA, USA). Compounds were detected in reflector mode using α-cyano-4-hydroxycinnamic acid (CHCA) as matrix.

Analytical and preparative TLC were performed on silica gel plates (Kieselgel 60, F254, 0.25 mm) (Merck, Darmstadt, Germany). The spots were visualized by exposure to UV radiation (253 nm), or by spraying with 10% H2SO4 in MeOH followed by heating at 110 ◦C for 10 min. 5-(hydroxymethyl)furan-3-carboxylic acid and KA were purchased from Enamine Ltd. (Kyiv, Ukraine) and Alfa Aesar (Karlsruhe, Germany), respectively.

#### *4.2. Isolation of* A. flavus *from Bees*

Worker bee individuals were collected from experimental apiaries at the Department of Agriculture, University of Naples Federico II (Portici, Italy) in October–November 2018. Freshly emerged bees (<48 h) were anaesthetized by chilling on ice for 20 min, and surface-disinfected by dipping in 70% ethanol for 30 s. The intersegmental membrane between head and thorax was cut using a sterile scalpel. Ten microliters of the liquid matter exuding from haemocoel were aseptically collected using a micropipette, and spotted on potato-dextrose agar (PDA, Oxoid) amended with 50 mg/L chloramphenicol. Plates were incubated at 32 ◦C in the dark. Hyphal tips from the emerging fungal colonies were transferred to fresh PDA plates to obtain pure cultures for morphological identification and storage. The strains obtained were readily identified as belonging to the *Aspergillus* section *Flavi*, based on the formation of colonies producing a dense felt of yellow-green rough conidiophores with radiate heads, and subglobose echinulate conidia. An orange reverse could be observed in cultures on a specific medium (AFPA) [44]. Unequivocal ascription to the species *A. flavus* resulted after rDNA-ITS and calmodulin gene sequencing. To this purpose, total genomic DNA was extracted from fresh mycelium taken from pure culture of strain AB1EET using CTAB protocol [45]. According to reported methodology, primers ITS1F and ITS4 were used to amplify rDNA-ITS, while primers CF1M and CF4 were used to amplify the calmodulin gene [45,46]. The original DNA sequences obtained in this study have been deposited in GenBank under the codes MK611561 (ITS) and MK611938 (calmodulin). The calmodulin gene sequence proved to be more respondent for unequivocal species identification, yielding 100% homology with sequences from 60 strains of *A. flavus* available in GenBank.

#### *4.3. Production and Extraction of KA*

Strain AB1EET was cultured in Czapek-Dox broth (Oxoid) following a previously reported procedure [47]. The freeze-dried culture filtrates (750 mL) were dissolved in 100 mL, acidified with HCl 2 N at pH 3, and extracted three times with an equal volume of EtOAc. Organic phases were combined, dried with Na2SO4, and evaporated under reduced pressure originating a white solid residue (350.7 mg) identified as KA.

#### *4.4. Determination of Protonation Constants*

The evaluation of the protolysis constants was conducted through spectrophotometric and potentiometric titration, at 25 ◦C in 0.1 M NaClO4 as ionic medium [48]. By experimental data the average number (*Z*H) of protons released per molecule was assessed through the equation:

$$Z\_{\rm HI} = ([\rm H^{+}] - C\_{\rm H} - K\_{\rm w}/[\rm H^{+}])/C\_{\rm L} \tag{1}$$

where [H+] is hydrogenionic concentration, *C*<sup>H</sup> is the total acid concentration, *C*<sup>L</sup> is the compound concentration and Kw is ionic product (10<sup>−</sup>13.7, in 0.1 M NaClO4). Experimental data were processed with Hyperquad softare [49].

#### *4.5. UV and MS Data of KA and 5-(Hydroxymethyl)furan-3-carboxylic Acid*

Kojic acid (**1**). UV λmax nm (log ε): (H2O) 216 (4.12), 270 (3.93); (MeOH) 225 (3.99), 270 (3.99); (pH 3.0) 215 (3.47), 270 (3.30); (pH 5.0) 216 (3.51), 268 (3.17), 318 (2.63); (pH 8.0) 227 (3.66), 315 (3.5); (pH 9.0) 227 (3.72), 315 (3.15). MALDI TOF/MS: *m*/*z* 143 [M+H]+, 125 [M-OH]+, 113 [M-CHO]+, 97 [M-COOH]<sup>+</sup>, 69 [M-CH2OH-COO]<sup>+</sup>.

5-(Hydroxymethyl)furan-3-carboxylic acid (**2**). UV λmax nm (log ε): (H2O) 240 (3.39); (MeOH) 240 (3.38); (pH 2.5) 243 (2.97); (pH 3.0) 243 (3.07); (pH 4.0) 243 (3.07); (pH 5.0) 243 (3.12). MALDI TOF/MS: *m*/*z* 143 [M+H]+, 125 [M-OH]+, 113 [M-CHO]+, 97 [M-COOH]+.

#### *4.6. Sample Methylation*

Fifteen milligrams of samples [KA, 5-(hydroxymethyl)furan-3-carboxylic acid, 5-(acetoxymethyl)furan-3-carboxylic acid (**1**, **2** and **4**)] were dissolved in MeOH (1.5 mL); an ethereal solution of CH2N2 was slowly added until a yellow color became persistent. The reaction mixtures were stirred at room temperature for 4 h. The solvent was evaporated under a N2 stream at room temperature. Residues of each reaction were analyzed by TLC on silica gel; **6**, was evidenced at Rf 0.37 by eluting with EtOAc-MeOH (9:1), while Rf 0.54 and 0.82 corresponded to **7** and **9** respectively, as eluted with CHCl3-i-PrOH (92:8).

#### *4.7. Sample Acetylation*

Ten mg of samples [KA, 5-(hydroxymethyl)furan-3-carboxylic acid, 5-*O*-methylkojic acid (**1**, **2**, **6**)], dissolved in pyridine (30 μL), were converted into the corresponding acetyl derivatives (**5, 4, 7**) by acetylation with Ac2O (30 μL) at room temperature overnight. The reaction was stopped by addition of MeOH, and the azeotrope formed by addition of benzene was evaporated in a N2 stream at 40 ◦C. Residues of each reaction were analyzed by TLC on silica gel; **5** was evidenced at Rf 0.44 by eluting with CHCl3-i-PrOH (95:5), while Rf 0.54 and 0.82 corresponded to **7** and **9** respectively, as eluted with CHCl3-i-PrOH (92:8).

**Supplementary Materials:** The following are available online, Figures S1–S3. NMR spectra of KA (**1**) and 5-(hydroxymethyl)furan-3-carboxylic acid (**2**); Figures S4–S9. 1H and 13C-NMR spectra of acetyl and methyl derivatives of **2**.

**Author Contributions:** M.D., R.N., and A.A. conceived and organized the manuscript, and wrote the text; R.N. and A.B. isolated and cultivated the fungal strain; M.M.S. and A.A. extracted the culture filtrate and prepared sample derivatives; M.D. and A.A. performed the NMR analysis; G.D.T. and M.I. performed potentiometric and spectrophotometric measurements; M.D., M.M.S, R.N., and A.A edited and reviewed the manuscript.

**Funding:** This research was funded by Finanziamento delle Attività Base della Ricerca (FFABR) 2017 of Ministero dell'Istruzione, dell'Università e della Ricerca (MIUR, Italy).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Sample Availability:** Samples of the compounds **1**-**9** are available from the authors.

© 2019 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 (http://creativecommons.org/licenses/by/4.0/).

#### *Article*
