Review of Oxepine-Pyrimidinone-Ketopiperazine Type Nonribosomal Peptides

Recently, a rare class of nonribosomal peptides (NRPs) bearing a unique Oxepine-Pyrimidinone-Ketopiperazine (OPK) scaffold has been exclusively isolated from fungal sources. Based on the number of rings and conjugation systems on the backbone, it can be further categorized into three types A, B, and C. These compounds have been applied to various bioassays, and some have exhibited promising bioactivities like antifungal activity against phytopathogenic fungi and transcriptional activation on liver X receptor α. This review summarizes all the research related to natural OPK NRPs, including their biological sources, chemical structures, bioassays, as well as proposed biosynthetic mechanisms from 1988 to March 2020. The taxonomy of the fungal sources and chirality-related issues of these products are also discussed.


Plant Growth Regulation
Cinereain (1), the first OPK peptide, could significantly inhibit the growth of etiolated wheat coleoptiles (p < 0.01) at 10 −4 and 10 −3 M and cause mild necrosis and chlorosis in corn, but it did not have any effect on intact greenhouse-grown bean and tobacco plants [6].

Anti-Inflammatory Activity
In a topical resiniferatoxin (RTX)-induced mouse ear edema assay, oxepinamide A (4) showed good topical anti-inflammatory activity with 82% inhibition of edema at the standard testing dose of 50 µg per ear [20].

Biosynthesis
The biosynthesis of OPK NRPs remains unsolved despite the fact that some biosynthetic pathway studies have been performed on similar quinazolinone alkaloids [38][39][40][41]. Possible biosynthetic pathways of several OPK compounds have, however, been proposed. Janoxepin (7) was suggested to be derived from the condensation of anthranilic acid with a diketopiperazine ring formed between two leucine residues, followed by oxidation of the benzoyl derivative to give the oxepine derivative [21]. Similarly, oxepinamide D (12) was proposed to be biosynthesized by the condensation of a diketopiperazine with an anthranilic acid and subsequent oxidation of the benzene ring to form an arene oxide, which was opened through a thermal 6π electrocyclic ring-opening process. Oxepinamides E-G (13)(14)(15) were formed by dehydration on the 2,5-diketo ring, followed by the addition of water between C-6 and C-12 [24]. Circumdatins A (2) and B (8) were proposed to be biosynthesized by oxidation of circumdatins H and J to form a benzene oxide, where a retro-pericyclic reaction (benzene oxide-oxepine tautomerism) took place to produce the final products [23,42]. Similar to janoxepin (7), the backbone of varioxepine A (20) has also been proposed to be from the condensation of anthranilic acid with a diketopiperazine, followed by epoxidation of the benzene ring to form the oxepine derivative. A series of reactions were proposed, including a second epoxidation, ring arrangement, epoxy opening, prenylation, dihydroxylation, and/or cyclization to yield the end product [28]. Protuboxepin D (27) was proposed to be formed by condensation of D-phenylalanine, L-isoleucine, and anthranilic acid, followed by oxidation of the benzene ring to form the oxepine derivative through an epoxy precursor and sequent oxidation at C-3 to form the hydroxyl group. Protuboxepin C (26) was a methylation product of protuboxepin D (27) [31]. A recent report proposed that additional opening and oxidation could happen on the oxepine ring, which then may undergo addition of water, cyclization, and methylation to yield unique (di/tetra)-hydropyran-pyrimidinone-ketopiperazine heterotricyclic products [34].

Discussion
In total, thirty-five OPK compounds have currently been characterized from natural sources. The speed of novel OPK product discovery has been increasing in recent years, as over half of the currently described products were isolated during the past eight years (Figure 2A). It is quite noteworthy that all these compounds were isolated from five fungal genera. Specifically, 70% of OPK NRPs, including the rediscovered cases, were obtained from the genus Aspergillus, followed by genus Penicillium accounting for 14%, Acremonium 7%, Paecilomyces 7%, and Botrytis 2% ( Figure 2B). Interestingly, all type C producers are from Aspergillus section Circumdati, including A. ochraceus, A. ostianus, and A. westerdijkiae, and a large proportion of type A and B compounds were obtained from different isolates of in particular the two species A. versicolor and A. protuberus, both belonging to A. versicolor clade in section Nidulantes [43][44][45]. In general, OPK compounds have been reported from species in the closely related fungal families Aspergillaceae (Aspergillus, Penicillium) and Trichocomaceae (Paecilomyces). Botrytis cinerea and Acremonium species are distantly related to Aspergillaceae and Trichocomaceae, but they were also reported to produce OPK compounds. Unfortunately, several of the reported strains have not been deposited in any culture collections affiliated to the World Federation for Culture Collections (WFCC), which is possibly why their identity has not been validated. It is also notable that even though some species reported bear the initials of a collection center, their strain number cannot be traced in the corresponding collection system. For example, Aspergillus ochraceus DSM 7428 cannot be found in DSMZ collection, and Aspergillus versicolor (AS 3.4186) cannot be traced in CGMCC collection. While the identification of Botrytis cinerea (ATCC 64157) can be verified, the identification of Acremonium (strain unavailable) was based on fatty acid methyl ester (FAME) profiles, a method which has not been authenticated for identification purposes in filamentous fungi. Genome mining of Botrytis and Acremonium species will show whether OPK compounds are taxonomically widespread or restricted to Aspergillaceae and Tricocomaceae.
Based on the number of rings and conjugation systems on the backbone, OPK NRPs were categorized into three types: A, B, and C. Type A dominating the OPK NRPs with 25 compounds shares the same 7/6/6 backbone, whereas type B OPK's contains a larger conjugation system. Type C 7/6/7/6 backbone has one more ring than types A and B due to incorporation of a second anthranilic acid moiety, and some products even display a complex 7/6/7/6/5 ring system with an additional pyrrolidine-ring from proline. In nature, a lot of other OPK similar products have been isolated, such as the quinazolinones [7,13]. Due to their possible related biosynthetic pathways, mistakes might happen during structure elucidation [22,23]. One common issue with OPK compounds is the absolute configuration (AC) determination of α carbons and R groups of the amino acids. In many reports, NOESY, Marfey´s reaction, X-ray crystallography, and ECD methods were applied. However, one might observe a mixture of D-and L-products after the acid hydrolysis and derivation process when using Marfey´s reaction method. Hydrolysis conditions thus may need to be optimized. In the case of a chiral center at a flexible position, it can be very challenging to solve the correct configuration. Success has recently been achieved by comparing the experimental VCD spectrum with calculated data [33]. The chiral centers within the R group of the isoleucine residue in eight OPK compounds (4, 5,  9, 10, 11, 16, 23, and 24) remain uncharacterized. The chirality also makes it confusing when referring to a structure in a publication. For example, the drawings of oxepinamide E and F (13)(14) showed a 17R configuration (wrong) but was described as 17S (correct configuration by X-ray Crystallography) in the same paper [24]. Additionally, the chiral center of janoxepin (7) was determined as R configuration by Marfey´s method, but the drawing mistakenly exhibited S configuration [21]. Such errors also happened when the structures were drawn in different publications, like the chirality of the two α carbons of both versicoloids A and B (23-24) drawn in a recent paper [33], which displayed opposite configurations from the original structures [29]. Care should be taken to avoid making such erroneous configurational drawings. Moreover, it is also notable that both D-and L-amino acids can participate in building the OPK products based on all the characterized structures. Therefore, proposing the chirality of α carbon from a biogenetic prospect can be challenging. Based on the number of rings and conjugation systems on the backbone, OPK NRPs were categorized into three types: A, B, and C. Type A dominating the OPK NRPs with 25 compounds shares the same 7/6/6 backbone, whereas type B OPK's contains a larger conjugation system. Type C 7/6/7/6 backbone has one more ring than types A and B due to incorporation of a second anthranilic acid moiety, and some products even display a complex 7/6/7/6/5 ring system with an additional pyrrolidine-ring from proline. In nature, a lot of other OPK similar products have been isolated, such as the quinazolinones [7,13]. Due to their possible related biosynthetic pathways, mistakes might happen during structure elucidation [22,23]. One common issue with OPK compounds is the absolute configuration (AC) determination of α carbons and R groups of the amino acids. In many reports, NOESY, Marfey´s reaction, X-ray crystallography, and ECD methods were applied. However, one might observe a mixture of D-and L-products after the acid hydrolysis and derivation process when using Marfey´s reaction method. Hydrolysis conditions thus may need to be optimized. In the case of a chiral center at a flexible position, it can be very challenging to solve the correct configuration. Success has recently been achieved by comparing the experimental VCD spectrum with calculated data [33]. The chiral centers within the R group of the isoleucine residue in eight OPK compounds (4, 5, 9, 10, 11, 16, 23, and 24) remain uncharacterized. The chirality also makes it confusing when referring to a structure in a publication. For example, the drawings of oxepinamide E and F (13)(14) showed a 17R configuration (wrong) but was described as 17S (correct configuration by X-ray Crystallography) in the same paper [24]. Additionally, the chiral center of janoxepin (7) was determined as R configuration by Marfey´s method, but the drawing mistakenly exhibited S configuration [21]. Such errors also happened when the structures were drawn in different publications, like the chirality of the two α carbons of both versicoloids A and B (23-24) drawn in a recent paper [33], which displayed opposite configurations from the original structures [29]. Care should be taken to avoid making such erroneous configurational drawings. Moreover, it is also notable that both D-and L-amino acids can participate in building the OPK products based on all A wide range of bioassays have been applied to assess the potential bioactivity of the OPK type of compounds. Though they in general seem to be inactive against human pathogenic strains of Candia albicans, some showed potential in treating plant-pathogenic fungi such as Fusarium graminearum and Colletotrichum acutatum. Notably, protuboxepin A (16) has the potential to become a new and effective anticancer drug as it displayed antiproliferative activity by disrupting microtubule dynamics through the tubulin polymerizing in tumor cells [37] despite several other reports showing that OPK compounds did not seem to be active against cancer cells. Antibacterial tests have shown that varioloids A and B (21)(22) exhibited promising activities against several species, while the rest of antibacterial tests did not display antibacterial activity effects. Interestingly, oxepinamides D-G and H−K (12-15, 32-35) all selectively showed transactivation effects on LXRα, which implied their potential use as novel LXR agonists in the treatment of atherosclerosis, diabetes, and Alzheimer's disease.
Overall, this class of compounds seem to share similar biosynthetic steps to form the OPK backbone, which is likely biosynthesized by the condensation of three amino acids, including one or two anthranilic acids, to form the tricyclic core. Subsequent epoxidation on the benzene ring of the first anthranilic acid residue followed by a ring rearrangement then produces the oxepine moiety. Several successive tailoring reactions can happen before the full construction of the final product(s) (Figure 3). Based on the knowledge of the biosynthesis of fumiquinazolines, and their well documentated proposed biosynthetic pathways, we anticipate that the OPK NRPs biosynthetic gene cluster contains at least a tri-modular NRPS gene with three adenylation domains, including one or two anthranilate-activating domains, and a gene responsible for oxidizing the phenyl moiety of the anthranilic residue to form the oxepine unit [39,40]. Additionally, an epimerization domain as part of NRPS is needed to convert L-amino acids to D-amino acids in the structures with a D-amino acid residue. Other tailoring genes are also required to encode for OPK related enzymes such as anthranilate synthase, oxidoreductases, and transporters. two anthranilic acids, to form the tricyclic core. Subsequent epoxidation on the benzene ring of the first anthranilic acid residue followed by a ring rearrangement then produces the oxepine moiety. Several successive tailoring reactions can happen before the full construction of the final product(s) (Figure 3). Based on the knowledge of the biosynthesis of fumiquinazolines, and their well documentated proposed biosynthetic pathways, we anticipate that the OPK NRPs biosynthetic gene cluster contains at least a tri-modular NRPS gene with three adenylation domains, including one or two anthranilate-activating domains, and a gene responsible for oxidizing the phenyl moiety of the anthranilic residue to form the oxepine unit [39,40]. Additionally, an epimerization domain as part of NRPS is needed to convert L-amino acids to D-amino acids in the structures with a D-amino acid residue. Other tailoring genes are also required to encode for OPK related enzymes such as anthranilate synthase, oxidoreductases, and transporters.

Conclusions
All the OPK NRPs described here were isolated from fungal sources with most compounds reported from species within the families Aspergillaceae (Aspergillus, Penicillium) and Trichocomaceae (Paecilomyces). Type A and B compounds share the same 7/6/6 backbone, with the former dominating OPK NRPs with twenty-five reported compounds, while Type C OPKs have a larger 7/6/7/6 backbone with four products reported. In general, these compounds showed promising activities against various phytopathogenic fungi and exhibited transactivation effects on LXRα. In addition, the skeleton of OPK NRPs is likely derived from the condensation of three amino acids, including one or two anthranilic acid(s), and the oxepine moiety is formed by the epoxidation of the benzene ring followed by ring arrangement. However, experimental investigation is needed to support this hypothesis. With the advance of separation skills and spectroscopic techniques, more oxepinecontaining compounds are likely to be discovered. Considering that many of these compounds were reported from Aspergilli, ongoing whole genome sequencing of all species in genus Aspergillus will possibly set the scene for genomic driven approaches towards new OPK NRPs [46,47].