Fusarium Cyclodepsipeptide Mycotoxins: Chemistry, Biosynthesis, and Occurrence

Most of the fungi from the Fusarium genus are pathogenic to cereals, vegetables, and fruits and the products of their secondary metabolism mycotoxins may accumulate in foods and feeds. Non-ribosomal cyclodepsipeptides are one of the main mycotoxin groups and include beauvericins (BEAs), enniatins (ENNs), and beauvenniatins (BEAEs). When ingested, even small amounts of these metabolites significantly affect human and animal health. On the other hand, in view of their antimicrobial activities and cytotoxicity, they may be used as components in drug discovery and processing and are considered as suitable candidates for anti-cancer drugs. Therefore, it is crucial to expand the existing knowledge about cyclodepsipeptides and to search for new analogues of these compounds. The present manuscript aimed to highlight the extensive variability of cyclodepsipeptides by describing chemistry, biosynthesis, and occurrence of BEAs, ENNs, and BEAEs in foods and feeds. Moreover, the co-occurrence of Fusarium species was compared to the amounts of toxins in crops, vegetables, and fruits from different regions of the world.


Introduction
Fungi belonging to the Fusarium genus produce a wide range of secondary metabolites, including the non-ribosomal depsipeptide mycotoxins, such as beauvericins (BEAs), beauvenniatins (BEAEs), enniatins (ENNs), and their analogues [1][2][3][4]. BEAs, BEAEs, and ENNs were included in the cyclodepsipeptide group of compounds, often found in high concentrations in grains, crops, vegetables, fruits, and even eggs, as a result of fungal infection [5][6][7][8][9]. They are involved in plant-pathogen interaction and may lead to many plants diseases, which can be very dangerous for animals health, including humans [10][11][12][13][14]. For example, ENNs produced by Fusarium species may act synergistically as a phytotoxin complex, which causes wilt and necrosis of plant tissue [15]. Moreover, ENN B affects mouse embryo development by inducing the dosage-related apoptosis or necrosis in mouse blastocytes [16]. On the other hand, BEA demonstrated neurotoxic properties in mice. In higher concentrations (7.5 and 10 µM), it affected the skeletal muscle fibers [17].
Additionally, BEA has a harmful influence on the reproductive system. The progesterone synthesis in cumulus cells was decreased when exposed to BEA [18]. Moreover, BEA inhibited estradiol and progesterone synthesis in bovine granulosa cells [19]. Also, ENN B reduced progesterone, testosterone, containing one, two, or three groups of D-2-hydroxy-3-methylpentanoic acid (D-Hmp), respectively [44]. Some of the reported ENNs are isomers, with the same amino acid composition but in different positions, e.g., ENN J 1 , J 2 , J 3 or ENN A and F [39,43,45]. On the other hand, even though the ENNs are not isomers, they share the same molecular weight. Therefore, the MS/MS technique with acid hydrolysis or NMR is sometimes necessary during the detection of cyclodepsipeptides for their correct identification.
BEAEs possess hybrid structures between the aliphatic (enniatin-type) and aromatic (beauvericin-type) cyclodepsipeptides [2,3,26,30]. Moieties of N-methyl-phenylalanine, N-methyl-leucine, and/or N-methyl-valine are the parts of BEAEs' structures. BEAE A contains one N-methyl-valine, whereas BEAE B, G 1 , G 2 , and G 3 have two. BEAE L has one N-methyl-leucine in its structure. Apart from the D-2-hydroxyisovaleric acid (D-Hiv) group, three of the BEAE isomers, namely BEAE G 1 , G 2 , and G 3 , contain two D-2-hydroxy-3-methylpentanoic acid (D-Hmp) groups in different combinations. At first, all BEAEs were described as cyclodepsipeptides from Acremonium sp., however further research revealed that Fusarium species are also able to produce these compounds [2,3,26,30].  Toxins 2020, 12, x FOR PEER REVIEW 5 of 21   lodepsipeptides are biosynthesized by a multi-domain non-ribosomal peptide synthase mposed of enzymatic modules used to elongate the proteinogenic and non-proteinogeni well as carboxyl and hydroxy acids [48,49]. The modules respond to the order and numb rs incorporated into the chain. Separate NRPS modules are required to assemble the prod al module consists of the three core domains: adenylation (A) domain, thiolation or peptidy T or PCP) domain, and condensation (C) domain. Moreover, each module and each ac is used only once for the recognition and activation of the precursors through adenylati adenylation domain), covalent thioester tethering (T: thiolation or PCP: peptidyl carrier , which tethers the activated precursor to a 4′-phosphopantetheine (PP) cofactor through a t d transport substrates to the active sites of the domains, and condensation (C domain rs via catalyzing the peptide bond (C-N) formation between the elongated chain and the a cid. The main domains may be supported by additional domains of the NRPS, such ation (E) domain, which catalyzes the transformation of an L-amino acid into a D-amino ac merization (E/C) domains, which catalyze the epimerization and condensation. NRPSs co al reductase (R) domain, which is responsible for reducing the final peptide, the methylati which catalyzes N-methylation of the amino acid substrate, the cyclization (Cy) dom the formation of oxazoline or thiazoline rings by internal cyclization of cysteine, se e residues, and the oxidation (Ox) domain, which catalyzes the formation of an aromatic oxidation of a thiazoline ring. The last domains (TE-thioesterase domains), mostly locate PS module, are responsible for releasing the full-length NRPS product from the enzyme

Biosynthesis
Cyclodepsipeptides are biosynthesized by a multi-domain non-ribosomal peptide synthase (NRPS) that is composed of enzymatic modules used to elongate the proteinogenic and non-proteinogenic amino acids, as well as carboxyl and hydroxy acids [48,49]. The modules respond to the order and number of the precursors incorporated into the chain. Separate NRPS modules are required to assemble the product and a minimal module consists of the three core domains: adenylation (A) domain, thiolation or peptidyl-carrier protein (T or PCP) domain, and condensation (C) domain. Moreover, each module and each active site domain is used only once for the recognition and activation of the precursors through adenylation with ATP (A: adenylation domain), covalent thioester tethering (T: thiolation or PCP: peptidyl carrier protein domain), which tethers the activated precursor to a 4 -phosphopantetheine (PP) cofactor through a thioester bond and transport substrates to the active sites of the domains, and condensation (C domain) of the precursors via catalyzing the peptide bond (C-N) formation between the elongated chain and the activated amino acid. The main domains may be supported by additional domains of the NRPS, such as the epimerization (E) domain, which catalyzes the transformation of an L-amino acid into a D-amino acid or the dual/epimerization (E/C) domains, which catalyze the epimerization and condensation. NRPSs contain an additional reductase (R) domain, which is responsible for reducing the final peptide, the methylation (MT) domain, which catalyzes N-methylation of the amino acid substrate, the cyclization (Cy) domain that catalyzes the formation of oxazoline or thiazoline rings by internal cyclization of cysteine, serine, or threonine residues, and the oxidation (Ox) domain, which catalyzes the formation of an aromatic thiazol through oxidation of a thiazoline ring. The last domains (TE-thioesterase domains), mostly located at the final NRPS module, are responsible for releasing the full-length NRPS product from the enzyme through cyclization or hydrolysis [48][49][50][51][52].
Enniatin biosynthesis is catalyzed by the 347 kDa multienzyme enniatin synthase (ESYN1) purified for the first time from Fusarium oxysporum and further characterized by Zocher and coworkers [53]. Extensive molecular research revealed the basis of cyclic oligopeptide biosynthesis and allowed us to identify esyn1, a gene encoding enniatin synthase, as the essential enzyme of the metabolic pathway [39,[54][55][56][57]. The biochemical characterization revealed that the enzyme possesses two substrate activation modules EA and EB, composed of approximately 420 amino acid residues. The EA module activates and participates in binding the α-D-hydroxy acids, while the EB module activates the amino acids. These two modules consist of a conserved 4-phosphopantetheine binding site at the C-terminus, with a highly conserved serine residue. An additional 4-phosphopantetheine group and N-methyltransferase domain M are present in the EB module. Also, a putative condensation (C) domain exists between the EA and EB modules. The M domain is highly conserved among N-methyl peptide synthases of prokaryotic and eukaryotic origin, thus it represents only local sequence similarities to the structural elements of other AdoMet-dependent methyltransferases. A dipeptidol unit is formed due to the interaction between the EA and EB modules and later, it is transferred and condensed into a thiol group. Three such successive condensations of the enzyme-bound dipeptidols are followed by the ring s closure into the enniatin (ENN) molecule [4,[58][59][60][61] (Figure 3A,B).  The primary precursors of the ENNs are valine, leucine or isoleucine, D-2-hydroxyisovaleric acid, and S-adenosylmethionine and their synthesis is entirely dependent on the cyclization reaction of linea hexadepsipeptide. The amino acid specificity of ESYN1 contributes to the chemical diversity of ENNs and this is why different types of ENNs are produced by Fusarium scirpi, F. lateritium, and F. sambucinum. Th Esyn domains activating L-valine in F. scirpi and preferably activating L-isoleucine in F. sambucinum ar nearly identical, with an exception of the three regions showing significant differences in their structures This difference in the activation can be accredited to the mutations that eventually occurred in the amin acid recognition sites of various enniatin synthases. In spite of the variability in amino acid units, certai structures. This difference in the activation can be accredited to the mutations that eventually occurred in the amino acid recognition sites of various enniatin synthases. In spite of the variability in amino acid units, certain ENNs can only be isolated from specific Fusarium strains, in which the enniatin synthase prefers some amino acids over others during biosynthesis [4,53,[62][63][64][65].
BEAs are also formed as cyclic trimers assembled from three D-Hiv-N-methyl-L-amino acid dipeptidol monomers ( Figure 4A) [50,51]. Similarly, they are also produced by a thiol template mechanism and synthesized by beauvericin synthase (BEAS) enzyme, which consists of a single polypeptide chain of about 351 kD [41,50]. For the first time, the 250 kDa BEAS enzyme was characterized by Peeters et al. [66] from the entomopathogenic fungus Beauveria bassiana, although Xu et al. [50], who conducted a more in-depth analysis, described a 33,475 bp beauvericin gene cluster including a 9570 bp bbBeas gene. Five years later, Zhang and coworkers [51] cloned and characterized 9413 bp beauvericin synthase gene (fpBeas) from Fusarium proliferatum.
The C 1 , A 1 , and T 1 domains within the first module of FpBEAS and ESYN (EA module) synthases have the same role in cyclodepsipeptide formation [51]. Nevertheless, the two depsipeptide synthases differ in A 2 domain substrate specificity within module 2 (ESYN EB module), i.e., apart from that of enniatin synthase, beauvericin synthase preferably accepts N-methyl-L-phenylalanine and some other aliphatic hydrophobic amino acids (e.g., leucine or isoleucine) [50]. Furthermore, their incorporation efficiency reduces with the length of side chains, where ortho-, meta-, and para-fluoro-substituted phenylalanine derivatives and N-methyl-L-leucine, N-methyl-L-norleucine, and N-methyl-L-isoleucine residues could replace N-methyl-L-phenylalanine. Domains C 2 , T 2a;b , M 2 , and C 3 within module 2 of BEAS and ESYN play the same role in both synthases ( Figure 4B) [50,66].
The depsipeptides, including BEAs, have a common 2-hydroxycarboxylic acid ingredient-D-2 -hydroxyisovalerate (D-Hiv) that is formed from 2-ketoisovalerate (2-Kiv) by a highly specific chiral reduction reaction catalyzed by 2-ketoisovalerate reductase (KIVR) enzyme [50,52,[67][68][69][70]. KIVR has a significant role in the biosynthesis of BEAs as was clearly understood when BEA production was inhibited in a KIVR knock-out B. bassiana mutant [67]. Kiv is formed from pyruvate during the biosynthesis of valine and it is the key intermediate in several metabolic pathways, including pantothenate biosynthesis in fungi, bacteria, and plants. It is also involved in producing phosphopantetheinyl prosthetic groups of acyl or peptidyl carrier proteins and co-enzyme A ( Figure 5) [50,52,67,69,70]. synthase, beauvericin synthase preferably accepts N-methyl-L-phenylalanine and some other aliphatic hydrophobic amino acids (e.g., leucine or isoleucine) [50]. Furthermore, their incorporation efficiency reduces with the length of side chains, where ortho-, meta-, and para-fluoro-substituted phenylalanine derivatives and N-methyl-L-leucine, N-methyl-L-norleucine, and N-methyl-L-isoleucine residues could replace N-methyl-L-phenylalanine. Domains C2, T2a;b, M2, and C3 within module 2 of BEAS and ESYN play the same role in both synthases ( Figure 4B) [50,66].  R knock-out B. bassiana mutant [67]. Kiv is formed from pyruvate during the biosynthesis the key intermediate in several metabolic pathways, including pantothenate biosynthesis and plants. It is also involved in producing phosphopantetheinyl prosthetic groups o carrier proteins and co-enzyme A ( Figure 5) [50,52,67,69,70]. ificant sequence homologies were identified for certain Fusarium enzymes, which genetic background for the synthesis of both depsipeptide compounds. Zhang et al. [51] analysis that FpBEAS (GenBank acc. no. JF826561.1) has 64% identity to ESYN (GenBank 45) as it was proven that some Fusarium species, like F. poae, F. proliferatum, or F. oxyspor produce ENNs and BEA simultaneously. This is justified by the fact that both toxins ic pathway [1,44,71,72]. Reports suggest that there is a high probability that the single PC nd/or BEAS-specific marker can detect potential BEAs and ENNs-producing fun nated soil and plant material [39,55,73].

um Species and Cyclodepsipeptide Mycotoxins in Food and Feed
nt crops are critical mainly in terms of yield and diverse use for foods and feeds. They suf of fungal diseases and Fusarium species are among the most damaging pathogens, produc ry metabolites, such as cyclodepsipeptides. Cyclodepsipeptides biosynthesis has been obse ium species (Table 2) and F. acuminatum, F. concentricum, F. proliferatum, F. verticilli m, and F. tricinctum produce a broad spectrum of ENN, BEA, and BEAE analogues. The re species formed only individual mycotoxin groups, such as BEA, ENNs, or a mixture r, in a few research papers, it was not specified which Fusarium species produced ENNs of mycotoxins was described as a "mix of ENNs" ( Table 2). arium species can cause many plant diseases and one of them is Fusarium head blight (FHB tating for cereal species, particularly as it is a major problem regarding wheat production s. Usually, one or more Fusarium species (F. graminearum, F. culmorum, F. avenaceum, F. poa Significant sequence homologies were identified for certain Fusarium enzymes, which shows a common genetic background for the synthesis of both depsipeptide compounds. Zhang et al. [51] revealed in their analysis that FpBEAS (GenBank acc. no. JF826561.1) has 64% identity to ESYN (GenBank acc. no. CAA79245) as it was proven that some Fusarium species, like F. poae, F. proliferatum, or F. oxysporum were found to produce ENNs and BEA simultaneously. This is justified by the fact that both toxins share a metabolic pathway [1,44,71,72]. Reports suggest that there is a high probability that the single PCR based esyn1and/or BEASspecific marker can detect potential BEAs and ENNs-producing fungi from contaminated soil and plant material [39,55,73].

Fusarium Species and Cyclodepsipeptide Mycotoxins in Food and Feed
Plant crops are critical mainly in terms of yield and diverse use for foods and feeds. They suffer from a range of fungal diseases and Fusarium species are among the most damaging pathogens, producing toxic secondary metabolites, such as cyclodepsipeptides. Cyclodepsipeptides biosynthesis has been observed for 44 Fusarium species (Table 2) and F. acuminatum, F. concentricum, F. proliferatum, F. verticillioides, F. oxysporum, and F. tricinctum produce a broad spectrum of ENN, BEA, and BEAE analogues. The remaining Fusarium species formed only individual mycotoxin groups, such as BEA, ENNs, or a mixture of these. However, in a few research papers, it was not specified which Fusarium species produced ENNs and the presence of mycotoxins was described as a "mix of ENNs" ( Table 2).
Fusarium species can cause many plant diseases and one of them is Fusarium head blight (FHB), which is devastating for cereal species, particularly as it is a major problem regarding wheat production in many countries. Usually, one or more Fusarium species (F. graminearum, F. culmorum, F. avenaceum, F. poae, and F. sporotrichioides) are involved as causal agents [74]. The occurrence of many Fusarium species may increase the accumulation of mycotoxins in grains or plants and introduce them into the food chain [71,75,76]. Humidity and temperature determine the disease severity, but geographical conditions, plant genotype, and local pathogen populations also play essential roles [54,77].  "ENN"-enniatin; "BEA"-beauvericin; "ALLOBEA"-allobeauvericin; "BEAE"-beauvenniatin.
Available literature data relate both to identifying Fusarium fungi isolated from various hosts and analyzing their mycotoxin biosynthesis capacity (Table 3). Efforts are also being made to assess contamination levels with these toxins of raw plant materials and food and feed products (Table 4). Mainly, the content of BEA and four ENNs (ENN A, ENN A 1 , ENN B, ENN B 1 ) has been investigated [8,25]. BEA and ENNs are common contaminants and were detected in plant crops and grains throughout the world. The occurrence of BEA, ENN A, ENN A 1 , ENN B, and ENN B 1 in naturally contaminated crops has been studied much more extensively than the occurrence of other cyclodepsipeptides [1,39]. Table 3 summarizes the most effective producers of depsipeptides among Fusarium fungi isolated from different crops and geographical areas. F. avenaceum, F. equiseti, F. proliferatum, and F. sporotrichioides were the most common species isolated from plants. The best producer of BEA was F. proliferatum (FPG61_CM), isolated from garlic in Spain, with the concentration reaching 671.80 µg/g [6]. The highest yielding producers of ENNs were F. avenaceum (KF1330), isolated from wheat in Poland, and F. tricinctum (3405), isolated from wheat in Finland [5,39]. Both strains produced in the highest amounts ENN B (895.46 µg/g, 690 µg/g) and ENN B 1 (452.46 µg/g, 1200 µg/g) [5,39]. Table 4 presents the maximum amounts of BEA and ENNs in naturally contaminated plant crops described in the literature. The highest contamination level of BEA was found to be 1731.55 µg/g in Polish maize [95]. When compared to other cyclodepsipeptides, it was also the highest concentration of mycotoxin in crops. In Tunisian sorghum, maximum concentrations of ENN A (95.6 µg/g) and ENN B 1 (120.1 µg/g) were detected [96]. The highest amount of ENN A 1 was 813.01 µg/g and 814.42 µg/g in Spanish maize and rice, respectively [97]. ENN B was found with a maximum level of 180.6 µg/g in Tunisian wheat [96]. The data show very high variability of investigated cyclodepsipeptides and it can be concluded that each strain of Fusarium species possesses a unique ability to biosynthesize these compounds. In addition to crops, cyclodepsipeptides are also found in food and feed [98][99][100][101][102][103]. Cyclodepsipeptides were identified mainly in cereal food, with very high levels of ENN A 1 and B 1 in breakfast cereals from Morocco (668 and 795 µg/g, respectively) [99]. In feed samples, ENNs and BEA levels were very low and did not exceed 0.48 µg/g for BEA (poultry feed) and 2.19 µg/g for ENNs (poultry feed) [101].  "ND"-not detected; "NA"-not analyzed.  "ND"-not detected; "NA"-not analyzed.

Conclusions
Fungi from the Fusarium genus produce a unique set of cyclodepsipeptide analogues of different amounts. The described mycotoxins are involved in plant-pathogen interaction, thus they were detected in a range of foodstuffs or feeds originating from many countries. They may be very dangerous for human health because of their biological activities. On the other hand, cyclodepsipeptides possess antimicrobial, insecticidal, antifungal, and antibiotic activities, which may help develop new drugs. In addition, because of their cytotoxicity, cyclodepsipeptides may have applications in anti-cancer therapy. Moreover, new BEAs, ENNs, or BEAEs with different amino/hydroxy acid compositions are detected each year inside in vitro fungal cultures. It was proven that not only fungi from Fusarium genus naturally produce cyclodepsipeptides, but also other fungi belonging to Beauveria, Acremonium, and Paecilomyces genera. Therefore, it is essential to continually improve the knowledge regarding these compounds, their structure, diversity, and toxicity to screen products of fungal secondary metabolism and monitor the dispersion of phytopathogenic fungi, which are potent producers of threatening mycotoxins. Moreover, it would be beneficial to bettering the understanding of cyclodepsipeptide biosynthesis to investigate the diversity and evolution history of the BEAS/ESYN synthase gene cluster from various fungi.