Advances in Fungal Phenaloenones—Natural Metabolites with Great Promise: Biosynthesis, Bioactivities, and an In Silico Evaluation of Their Potential as Human Glucose Transporter 1 Inhibitors

Phenaloenones are structurally unique aromatic polyketides that have been reported in both microbial and plant sources. They possess a hydroxy perinaphthenone three-fused-ring system and exhibit diverse bioactivities, such as cytotoxic, antimicrobial, antioxidant, and anti-HIV properties, and tyrosinase, α-glucosidase, lipase, AchE (acetylcholinesterase), indoleamine 2,3-dioxygenase 1, angiotensin-I-converting enzyme, and tyrosine phosphatase inhibition. Moreover, they have a rich nucleophilic nucleus that has inspired many chemists and biologists to synthesize more of these related derivatives. The current review provides an overview of the reported phenalenones with a fungal origin, including their structures, sources, biosynthesis, and bioactivities. Moreover, more than 135 metabolites have been listed, and 71 references have been cited. SuperPred, an artificial intelligence (AI) webserver, was used to predict the potential targets for selected phenalenones. Among these targets, we chose human glucose transporter 1 (hGLUT1) for an extensive in silico study, as it shows high probability and model accuracy. Among them, aspergillussanones C (60) and G (60) possessed the highest negative docking scores of −15.082 and −14.829 kcal/mol, respectively, compared to the native inhibitor of 5RE (score: −11.206 kcal/mol). The MD (molecular dynamics) simulation revealed their stability in complexes with GLUT1 at 100 ns. The virtual screening study results open up a new therapeutic approach by using some phenalenones as hGLUT1 inhibitors, which might be a potential target for cancer therapy.


340
C 19 H 16 O 6 Coniothyrium cereale Enteromorpha sp. (Algae, Ulvaceae) Fehmarn, Baltic Sea, Germany [20,37]  Rousselianone A = Acetone adduct of 4,9-dihydroxy-6-methyl-7-((3methylbut-2-en-1-yl)oxy)-1Hphenalene-1,2,3-trione     Duclauxin (120) is an oligophenalenone heptacyclic dimer, consisting of dihydroisocoumarin and isocoumarin units that are joined together by a cyclopentane ring. The previously reported labeling experiment revealed that 120 originated from a heptaketide chain, which was cyclized to produce phenalenone (i) [21,61]. A triketone ii was formed by the oxidative loss of one of its carbons to yield a contracted ring, C. Its decarboxylation and regio-selective oxygen insertion, induced by enzymes or air, yielded dione iii and naphthalic anhydride iv [21]. Then, the selective reduction of iv generated a lactone (v). The dimerization of two lactone units through oxidative radical coupling between C-8 and C-9′a, which was catalyzed by oxidative enzymes, yielded a biaryl (vi) [62,63]. The latter underwent an intramolecular aldol condensation between the C-8′ and C-7 ketone group to furnish the aldol fragment (vii). The latter could experience a group of successive tailoring modifications in terms of reduction, methylation, acetylation, and dehydration, to produce 120 [56,64]. Compound 119 was assumed to be biosynthesized through ammonolysis, with the aid of one serine as a nitrogen donor and a further serine moiety′s decarboxylation, to provide 119 (Scheme 2) [56]. On the other hand, it was hypothesized that the compounds 49, 50, 52, and 54 were artifacts, resulting from the spontaneous addition of acetone or methyl ethyl ketone to the unstable triketone (Scheme 3) [48]. Meanwhile, the use of 3-pentanone as an initial extracting solvent ultimately led to the formation of 51 and 53 [48]. The enol inter-conversion of the C-4 and C-3 of IV generated an intermediate V that was converted by reduction to form VI, which was very active due to the C-8 two hydroxyl groups. Then, pyruvic acid was added to one of the hydroxyl groups to give VII. Finally, 104 and 105 were produced by the VII COOH group reduction, which was catalyzed by reductase [46]. Yu et al. proposed biosynthetic pathways for 116-118 (Scheme 5). The intermediate I was trans-aminated to yield II, which underwent oxidative cleavage to form a naphthoquinone (III) [55]. Then, a benzo[f ]chromene-1,7,10-trione derivative (IV) was generated through the cyclization between 7-OH and the 3-carbonyl group. The prenylation of IV, followed by a Claisen rearrangement, yielded VI. The coupling of communal F with VI generated 116 [66]. Furthermore, 112 and 117 were produced from the coupling between communal F and VIII [55]. Scheme 2. Biosynthetic pathway of compounds 119 and 120 [21,56,62,63].
On the other hand, it was hypothesized that the compounds 49, 50, 52, and 54 were artifacts, resulting from the spontaneous addition of acetone or methyl ethyl ketone to the unstable triketone (Scheme 3) [48]. Meanwhile, the use of 3-pentanone as an initial extracting solvent ultimately led to the formation of 51 and 53 [48]. Scheme 3. Proposed pathways for the formation of compounds 49-54 [48].

Scheme 3.
Proposed pathways for the formation of compounds 49-54 [48]. Li et al. proposed a biosynthetic pathway for compounds 21, 28, 31, and 76-81, which have a phenalenone nucleus fused to a trimethylfuran ring [39]. The trimethylfuran ring was biologically related to mevalonic acid [65]. The oxidative loss of C-6 of the heptaketide-derived phenalenone nucleus yielded P1, which could be prenylated in two various paths (A and B), leading to the formation of P2 and 28. The enzymatic epoxidation of the P2 double bond, followed by hydrolysis and then dehydration, yielded 76. The later oxidation produced 77 and 78. Similarly, the oxidation of 28 resulted in the formation of 21 and 31. Compound 28 was a precursor of 79-81 by the oxidative loss of C-7 (80) or C-5 (79 and 81) and then formed a lactone ring (80 and 81) [39]. The linkage of a prenyl side chain to 5-OH of the tricyclic intermediate II was catalyzed by prenyltransferase to yield the prenylated intermediate (III). Additionally, IV was generated from the Claisen rearrangement and cyclization of III [21] (Scheme 4). The enol inter-conversion of the C-4 and C-3 of IV generated an intermediate V that was converted by reduction to form VI, which was very active due to the C-8 two hydroxyl groups. Then, pyruvic acid was added to one of the hydroxyl groups to give VII. Finally, 104 and 105 were produced by the VII COOH group reduction, which was catalyzed by reductase [46]. Yu  to form a naphthoquinone (III) [55]. Then, a benzo[f]chromene-1,7,10-trione derivative (IV) was generated through the cyclization between 7-OH and the 3-carbonyl group. The prenylation of IV, followed by a Claisen rearrangement, yielded VI. The coupling of communal F with VI generated 116 [66]. Furthermore, 112 and 117 were produced from the coupling between communal F and VIII [55]. Scheme 5. Biosynthetic pathway of compounds 116-118 [55,66].

Bioactivities of Phenalenones
The bioactivities of some of the reported metabolites have been investigated. In this regard, 70 metabolites have been associated with some type of biological action, including cytotoxic, antimalarial, antimycobacterial, anti-inflammatory, anti-angiogenic, immunosuppressive, and antioxidant properties, as well as IDO1, α-glucosidase (AG), ACE, tyrosinase, and PTP inhibition. This information has been discussed and listed in Table 2.

Bioactivities of Phenalenones
The bioactivities of some of the reported metabolites have been investigated. In this regard, 70 metabolites have been associated with some type of biological action, including cytotoxic, antimalarial, antimycobacterial, anti-inflammatory, anti-angiogenic, immunosuppressive, and antioxidant properties, as well as IDO1, α-glucosidase (AG), ACE, tyrosinase, and PTP inhibition. This information has been discussed and listed in Table 2.
Paecilomycones A-C (1-3) were purified from Paecilomyces gunnii culture extract with the aid of a preparatory HSCCC, guided by HPLC-HRESIMS, used as a tyrosinase inhibitor. Compound 1 was similar to myeloconone A2 (4), which was formerly separated from the lichen Myeloconis erumpens [67], except that 1 has an OH group at C-8 instead of an OCH 3 group. Compound 3 was deduced as 9-amino-6,7,8-trihydroxy-3-methoxy-4-methyl-1Hphenalen-1-one; the existence of NH 2 in 3 was confirmed by a positive purple reaction with a ninhydrin reagent in the TLC plate. They were characterized by means of spectroscopic analyses. These metabolites exhibited potent tyrosinase inhibitory potential (IC 50 s 0.11, 0.17, and 0.14 mM, respectively) in the form of kojic acid (IC 50 0.10 mM), being stronger than arbutin (IC 50 0.20 mM). This influence was found to be positively related to the number of OH groups [32] (Figure 1).       an OCH3 group. Compound 3 was deduced as 9-amino-6,7,8-trihydroxy-3-methoxy-4-methyl-1H-phenalen-1-one; the existence of NH2 in 3 was confirmed by a positive purple reaction with a ninhydrin reagent in the TLC plate. They were characterized by means of spectroscopic analyses. These metabolites exhibited potent tyrosinase inhibitory potential (IC50s 0.11, 0.17, and 0.14 mM, respectively) in the form of kojic acid (IC50 0.10 mM), being stronger than arbutin (IC50 0.20 mM). This influence was found to be positively related to the number of OH groups [32] (Figure 1). Aspergillussanones A (5) and B (6) were separated from Aspergillus sp. PSU-RSPG185 broth extract. They differed from each other in the substitutions at C-8 and C-4, as well as in the C-4 configuration. The configuration of their double bonds was determined to be E, based on signal enhancement in the NOEDIFF experiment, and the 4S and 10′R in 5 and 4R and 10′R in 6 was assigned by the CD spectrum. Only compound 5 exhibited weak cytotoxic activity toward Vero cells and KB (IC50s 34.2 and 48.4 μM, respectively) in the resazurin microplate assay, compared to ellipticine (IC50 4.5 and 4.1 μM, respectively), whereas 6 was inactive against the tested cell lines. Additionally, they showed no antimalarial or antimycobacterial potential toward Plasmodium falciparum and Mycobacterium tuberculosis when using GFP (green fluorescent protein) and the microculture radioisotope technique, respectively [33] (Figure 1).
Eleven metabolites of the herqueinone subclass, including six new derivatives, entpeniciherqueinone (8), 12-hydroxynorherqueinone (11), ent-isoherqueinone (13), oxopropylisoherqueinones A (15) and B (16), and 4-hydroxysclerodin (27)  Aspergillussanones A (5) and B (6) were separated from Aspergillus sp. PSU-RSPG185 broth extract. They differed from each other in the substitutions at C-8 and C-4, as well as in the C-4 configuration. The configuration of their double bonds was determined to be E, based on signal enhancement in the NOEDIFF experiment, and the 4S and 10 R in 5 and 4R and 10 R in 6 was assigned by the CD spectrum. Only compound 5 exhibited weak cytotoxic activity toward Vero cells and KB (IC 50 s 34.2 and 48.4 µM, respectively) in the resazurin microplate assay, compared to ellipticine (IC 50 4.5 and 4.1 µM, respectively), whereas 6 was inactive against the tested cell lines. Additionally, they showed no antimalarial or antimycobacterial potential toward Plasmodium falciparum and Mycobacterium tuberculosis when using GFP (green fluorescent protein) and the micro-culture radioisotope technique, respectively [33] (Figure 1).

Artificial Intelligence (AI)-Based Target Prediction for Phenalenone Derivatives
The human glucose transporter 1 (hGLUT1) is one of 14 members of the GLUT family of integral proteins that are responsible for the facilitative transport of monosaccharides and polyols across the membrane bilayer of eukaryotic cells [68,69]. Structurally, GLUT1 consists of 12 α-helices that are folded into the C-terminal domain and the N-terminal domain, both of which consist of six transmembrane helices [70,71]. Due to its essential role in transporting glucose from the ECM (extracellular matrix) into the cells [72,73] and maintaining the viability of the cells [74], GLUT1 is ubiquitously expressed [74,75]. In many cancer types, the demand for glucose as a source of energy is increased, leading to the increased expression of glucose transporters, including GLUT1 [75,76]. Additionally, the upregulation of GLUT1 expression was found to be mediated by the stimulation of oncogenes [71], while inhibiting GLUT1 activity reduced cell proliferation and apoptosis [71,77,78]. These findings suggest that GLUT1 might be a potential target for cancer therapy [75]. Natural metabolites belonging to diverse classes have been found to possess hGLUT1 inhibition potential, such as resveratrol, phloretin, naringenin, WZB117, cytochalasin B, STF-31, pyrazolopyrimidines, (1H-pyrazol-4-yl)quinoline, and phenylalanine amides [71].
Ligand-based in silico target prediction was performed to choose a suitable target by which to investigate the potential inhibitory activity of the phenalenone derivatives [79,80]. Performing an anatomical-therapeutic chemical (ATC) code and predicting the potential targets for the investigated compounds were carried out using the SuperPred prediction web server [81,82]. From the prediction results, GLUT1 (PDB: 5EQG) was chosen as a target for the study as it had a high percentage of model accuracy and a very good probability ( Table 3). After selecting the target, the docking method was validated by redocking the co-crystalized inhibitor back into the protein crystal structure, then the docking of the listed phenalenones followed. In silico ADMET properties prediction for the listed compounds, along with molecular dynamic (MD) simulation for the two top-scoring derivatives after docking, were performed as well. * The probability of the test compound binding to a specific target, as determined by the respective target machine learning model. ** The 10-fold cross-validation score of the respective logistic regression model is presented, as the model performance varies between different targets.

In Silico ADMET Properties of Selected Ligands
All 20 phenalenones were prepared for the study by utilizing Schrodinger's Lig-Prep tool [83]. The 3D (three dimensional) structures of the compounds were generated using the OPLS3 force field setting, with an ionization state at pH 7.0 ± 0.2. After that, ADMET prediction was performed using the QikProp module on Schrodinger's suite [84]. Table 4 presented the ADMET properties that estimated the phenalenones' usefulness in terms of their biological functions, drug-likeness, physiochemical properties, and expected toxicity. The ADMET descriptors that are predicted for the derivatives are molecular weight, drug-likeness, dipole moment, total solvent accessible surface area, number of hydrogen bond donors and acceptors, predicted octanol-water partitioning, predicted aqueous solubility, estimated binding to human serum albumin, number of possible metabolites, predicted blood-brain partitioning, percentage of human oral absorption, predicted IC 50 for inhibiting HERG-K + channels, central nervous system activity, and the reactive functional group number. Most of the predicted values of ADMET descriptors fell within the recommended range.

Ligands and Protein Preparation
The compounds were prepared by converting their structures from 2D to 3D using LigPrep, and their ionization states and tautomeric forms were generated. After energyminimizing, the 3D structures of the compounds were ready for docking into the crystal structure of GLUT1 (PDB ID: 5EQG). The protein was prepared for docking using the protein preparation wizard, where its crystal structure was minimized and its H-bond network was optimized. In addition, the proper force field was specified, and the protein's formal charge was calculated after generating the amino acids' correct ionization states and the missing hydrogen addition.

Grid Box Generation and Molecular Docking
Molecular docking was performed to evaluate the binding modes of the selected compounds inside a protein binding pocket. To do that, a grid box was generated around the protein binding pocket to determine the exact site for the docking in the minimized protein crystal structure, using Maestro's Receptor-Grid-Generation tool [85]. The docking method was evaluated by re-docking the native inhibitors (PDB ID: 5RE) back into the crystal structure in which it was co-crystallized. The binding interactions of the re-docked inhibitor are shown in Figure 15. H-bonding was observed between the CO and the NH of the 4-fluorophenylalanine moiety, an adjacent water molecule, and with Glu380, respectively. The second carbonyl group seemed to have H-bonded with Gln161, as well as three nearby water molecules; the phenolic OH acted as both HBD and HBA with water molecules as well. into the crystal structure in which it was co-crystallized. The binding interactions of the re-docked inhibitor are shown in Figure 15. H-bonding was observed between the CO and the NH of the 4-fluorophenylalanine moiety, an adjacent water molecule, and with Glu380, respectively. The second carbonyl group seemed to have H-bonded with Gln161, as well as three nearby water molecules; the phenolic OH acted as both HBD and HBA with water molecules as well. After validating the docking method, the 3D structures of the minimized phenalenone derivatives were docked into GLUT1. The docking results are presented in Table 5, which shows that, except for derivative 118, all phenalenones scored higher than the native inhibitor (−11.206 kcal/mol). Derivatives 60 and 64 were on the top of the list, scoring −15.777 and −15.239 kcal/mol, respectively. After validating the docking method, the 3D structures of the minimized phenalenone derivatives were docked into GLUT1. The docking results are presented in Table 5, which shows that, except for derivative 118, all phenalenones scored higher than the native inhibitor (−11.206 kcal/mol). Derivatives 60 and 64 were on the top of the list, scoring −15.777 and −15.239 kcal/mol, respectively.   Figure 16 shows the binding interactions of compound 60 after docking. While the Phe26 side chain forms pi-pi stacking with the fused ring system of 60, several H-bonds are formed between the OH and carbonyl groups of the ring system and the amino acids Gln161, Asn411, Asn415, and Tyr292, as well as forming water bridges with the adjacent water molecules. Additional water bridges are formed with the oxygen and OH groups of the substituted tetrahydropyran at the end of the aliphatic chain.
As for compound 64, the carbonyl oxygens and the OH groups of the 3-membered ring system formed several water bridges and H-bonds with the water molecules and His160 and Gln161 side chains. The aliphatic chain and the cyclopentyl moiety interacted with Asn415 and the adjacent water molecules through the OH groups ( Figure 17). As for compound 64, the carbonyl oxygens and the OH groups of the 3-membered ring system formed several water bridges and H-bonds with the water molecules and His160 and Gln161 side chains. The aliphatic chain and the cyclopentyl moiety interacted with Asn415 and the adjacent water molecules through the OH groups ( Figure 17).

Molecular Dynamic Simulation (MD)
The MD simulation is a tool that is applied to mimic the physiological environment, to monitor any changes in the protein's 3D conformation and the binding affinity that might take place during the simulation, and then compare them to the original conformation and affinity of the crystal structure [86]. For that reason, Desmond software [87,88] was used to perform the MD study and evaluate the stability and the binding affinity of the protein-compound complexes at pH 7.0 ± 0.2, over a 100-ns period. The MD was performed only for the two phenalenone derivatives that scored the highest in the docking study, namely, compounds 60 and 64, as well as the co-crystallized inhibitor, 5RE. The root mean square deviation (RMSD) of the complexes of compounds and GLUT1 measures the mean change of atoms (of protein and of ligand) at the end of the simulation and compares it to the atoms in their original conformation at 0 ns. The RMSD graph for the GLUT1-5RE complex showed that their plots could be laid over each other, indicating high stability of the complex throughout the simulation, and their RMSD values were within the accepted 1-3 Å range ( Figure 18A). For the compound 60-GLUT1 complex, the protein was relatively stable throughout the duration of the experiment, with an RMSD value within the accepted 1-3 Å range. Compound 60 RMSD, however, was stable for about 58% of the time (~1.4-1.5 Å), then the confirmation of the ligand atoms changed drastically afterward, where it re-stabilized again between 5.9 and 7.1 Å by the end of the run. This indicated a change in the 3D conformation of the compound inside the pocket

Molecular Dynamic Simulation (MD)
The MD simulation is a tool that is applied to mimic the physiological environment, to monitor any changes in the protein's 3D conformation and the binding affinity that might take place during the simulation, and then compare them to the original conformation and affinity of the crystal structure [86]. For that reason, Desmond software [87,88] was used to perform the MD study and evaluate the stability and the binding affinity of the protein-compound complexes at pH 7.0 ± 0.2, over a 100-ns period. The MD was performed only for the two phenalenone derivatives that scored the highest in the docking study, namely, compounds 60 and 64, as well as the co-crystallized inhibitor, 5RE. The root mean square deviation (RMSD) of the complexes of compounds and GLUT1 measures the mean change of atoms (of protein and of ligand) at the end of the simulation and compares it to the atoms in their original conformation at 0 ns. The RMSD graph for the GLUT1-5RE complex showed that their plots could be laid over each other, indicating high stability of the complex throughout the simulation, and their RMSD values were within the accepted 1-3 Å range ( Figure 18A). For the compound 60-GLUT1 complex, the protein was relatively stable throughout the duration of the experiment, with an RMSD value within the accepted 1-3 Å range. Compound 60 RMSD, however, was stable for about 58% of the time (~1.4-1.5 Å), then the confirmation of the ligand atoms changed drastically afterward, where it re-stabilized again between 5.9 and 7.1 Å by the end of the run. This indicated a change in the 3D conformation of the compound inside the pocket during the run, as the compound adjusted its pose to one with lower free energy. This was most likely due to the presence of many rotatable bonds in the aliphatic side chain ( Figure 19A). The RMSD for the GLUT1-64 complex fell within the acceptable range as well ( Figure 20A).
The secondary structure of the protein was scrutinized throughout the MD analysis to ensure that the percentage secondary structure element (%SSE) is intact over the simulation time. Figure 18B demonstrated the integrity of the SSE of the protein when complexed with 5RE. The top plot showed the distribution of the SSE (α-helices and β-sheets) throughout the protein, represented by the residue index. The middle plot checked the overall %SSE, while the bottom plot assessed each SSE over the course of the simulation. Both plots indicated that the overall %SSE of the protein was maintained, and each SSE was stable over the course of the simulation. A similar result was observed for the GLUT1-60 and GLUT1-64 complexes (Figures 19B and 20B, respectively).
The interaction between the test compounds and GLUT1 was also examined by the MD study. Figure 21A illustrates the stacked-bar graph displaying the types of interactions between the pocket residues and the bound ligand. The binding interactions are colorcoded in the legend of the figure. The three most prominent observed interactions were as follows. Glu380 achieved H-bonding with the amide nitrogen, with a value of 0.75, while Phe291 formed pi-pi stacking with the 4-fluorophenyl moiety, with a value of 0.8. The carbonyl oxygen of the compound interacted by a water bridge and through H-bonding with Gln161 (normalized value of~0.83). Additional interactions included that of Gln283, which interacted with the second carbonyl indirectly through a water bridge. Figure 21B shows the 2D view of the binding interactions, depicting interactions that were maintained for at least 30% of the simulation time. Figure 21C is the timeline representation of the stacked-bar graph that presented the interaction pattern of each of the pocket residues of GLUT1 with the 5RE during the 100 ns of simulation time. The orange color means that there was an interaction, while the darker colors indicate that the residue formed more than one interaction with the ligand.
The interactions between GLUT1 and 60 included H-bonding and a water bridge with Asn411, yielding a~1.4 value. In addition, H-bonding and the water bridge contact points were formed with the residues Thr137, Gln161, and Tyr292, along with a hydrophobic interaction with Trp388 ( Figure 22A). The 2D view of 60 complexed with GLUT1 showed relatively stable interactions between the compound and the residues Asn411, Trp388, Tyr292, Gln161, and Thr137 throughout the MD run ( Figure 22B). From the timeline representation of the stacked bar plot presented in Figure 22C, the same interactions that are illustrated in the stacked bar plot were maintained over 100 ns. However, the interaction of Thr137 with the ligand began strongly and then started disappearing at~70 ns until the end of the run. This might explain the change in the RMSD plot for 60 after~58 ns ( Figure 19A).
The amino acid residues involved in binding to derivative 64 included Trp388, Gln238, His160, Gln161, Pro141, and Thr137, with values of between 0.5 and 1.4. A water bridge was observed with Pro401 and Gly138 ( Figure 23A). The interactions in the 2D view ( Figure 23B) and timeline plot ( Figure 23C) agreed with those in the stacked bar graph ( Figure 23A).

ADMET Properties Prediction
The Maestro QikProp Schrodinger module [84] was utilized for the prediction of ADMET properties and drug-likeness for the selected compounds. The properties included absorption, distribution, metabolism, excretion, toxicity, and others.

Preparation of Protein and Ligands PDB Structures
The crystal structure of the GLUT1 with the PDB ID: 5EQG was selected for the experiment because the co-crystalized ligand has a similar structure to the compounds that are to be tested. From the protein databank (PDB), the protein crystal structure 5EQG was downloaded as a PDB file [89] and was then optimized and prepared by the Protein-Preparation wizard of Schrodinger [83,90,91]. Protein preparation and optimization included identifying the bond order for the known HET groups and untemplated residues, adding hydrogen, breaking bonds to metals, adding zero-order bonds between metals and adjacent atoms, and correcting the formal charges to metals and the nearby atoms. Water molecules further than 5 Å from HET groups were removed from the structure of HET groups, and the disulfide bonds were re-generated. Ligands were prepared using the Lig-Prep tool [83], which involved the generation of metal HET states and cofactors at pH 7 ± 2.0. Additionally, the optimization of hydrogen bonds at pH 7.0 using PROPKA [92], the removal of water molecules of >3 Å from HET groups, and applying restrained minimization using the OPLS4 force field were performed.    . The panel at the top illustrates the total number of specific interactions that the protein has made with the compound over the course of the trajectory. The panel below illustrates which residues interacted with the ligand in each trajectory frame. The dark or orange color indicated that more than one specific interaction was seen between certain residues and the ligand. # Number of contacts.

Grid Generation and Docking
Glide's Receptor-Grid-Generation tool [85] was used to generate a grid box around the co-crystalized inhibitor 5RE in the binding site of the protein PDB: 5EQG. The docking of the phenalenones was performed inside this box. The non-polar atoms were set for a van der Waals (VdW) radii scaling factor of 1.0 and the cut-off of partial charge was 0.25. Schrodinger's Ligand Docking tool was used to perform the docking procedure [85,93]. The docking protocol was set as standard precision (SP), while the ligand sampling method was flexible. The default settings were used for other parameters. illustrates the total number of specific interactions that the protein has made with the compound over the course of the trajectory. The panel below illustrates which residues interacted with the ligand in each trajectory frame. The dark or orange color indicates that more than one specific interaction was made between some residues and the ligand.

MD Simulation
MD simulation experiments were performed using the Schrodinger suite [87,88]. The selected protein-compound complexes were obtained from the docking results and were tuned through the "System-Builder" tool. TIP3P was selected as the solvent mode, and the chosen box shape was the orthorhombic shape. Na ions were added to neutralize the system, and the box dimensions were 10 Å. The duration of the MD simulations was 100 ns per trajectory, and the number of atoms, temperature, and pressure were kept constant (NPT ensemble). Conversely, the temperature was set at 300.0 K, the pressure was set at 1.01325 bar, and the force field was OPLS4. The panel below illustrates which residues interacted with the ligand in each trajectory frame. The dark or orange color indicates that more than one specific interaction was made between some residues and the ligand.

Conclusions
Fungi-derived metabolites possess substantial medicinal values and a large structural diversity that can provide an untapped potential for drug candidates and medications. This review summarized 139 fungal phenalenone derivatives and their biosynthesis and biological activities, reported from 2014 until August 2021. Most of them are mainly identified from Penicillium (37 compounds), Coniothyrium (23 compounds), Aspergillus (22 compounds), and Talaromyces (22 compounds) ( Figure 24).
Fungal phenalenones were derived from polyketide precursors that underwent different cyclization and tailoring reactions, leading to extreme structural diversity and high complexity. Hence, searching for diverse biosynthetic pathways for the various phenalenone derivatives will be a future challenge and offer an interesting research field for natural product researchers. These metabolites have been isolated and purified using various chromatographic tools, such as SiO 2 , Sephadex LH-20, preparative TLC, ODS, and HPLC, as well as preparative HSCCC-guided HPLC-HRESIMS, LC-MS-guided, and UV-HPLC guided analyses. Most of the separated phenalenones possess unique and unprecedented functionality or ring systems. Their configuration was assigned using various tools and experiments, such as X-rays, CD, ECD calculation, specific rotations, and chemical modifications, as well as the NOESY, NOEDIFF, and GIAO NMR shift calculations. They were evaluated for various activities, such as cytotoxic, antimalarial, antimycobacterial, anti-inflammatory, anti-angiogenic, immunosuppressive, and antioxidant properties, as well as IDO1, α-glucosidase (AG), ACE, IDO1, tyrosinase, and PTP inhibition ( Figure 25). Some of the reported derivatives possessed powerful activities greater than in the used controls, such as anti-HIV (e.g., 71 and 74), immunosuppressive (e.g., 17), anti-tumor (e.g., 120, 125, 129, and 131), and antibacterial behavior (e. g., 24, 31, 69, 83, and 85), in addition to tyrosinase (e. g., 1-3), α-glucosidase (e. g., 29, 34, 46, and 47), pancreatic lipase (e.g., 29 and 34), PTP (e. g., 101, 120, and 127), and ACE (e.g., 104) inhibitory behavior. Further in vivo and clinical studies should be conducted to validate these bioactivities. Conversely, many of the newly separated phenalenones possessed weak or no bioactivity, which represents a common problem for the study of the natural product. This could be due to the insufficient amounts of the new compounds and the lack of effective bioactivity screening methods. Cancer is one of the most significant worldwide health concerns and there is a continuous need for developing new targets for treating this disease. GLUT1 substantially increases the uptake of glucose into the cytoplasm and is over-expressed in various tumor cells. Therefore, it is likely to be a potential target for treating cancer. Based on the in silico studies, such as molecular docking, ADMET characteristics predication, and MD, some phenalenones were found to possess remarkable capacity as GLUT1 inhibitors; therefore, they could be potential leads for cancer treatment. It is noteworthy that the described results in this work are reported for the first time for this class of fungal metabolites and undoubtedly represent a substantial contribution in terms of further investigation, as well as in vitro and in vivo evaluations. (NPT ensemble). Conversely, the temperature was set at 300.0 K, the pressure was set at 1.01325 bar, and the force field was OPLS4.

Conclusions
Fungi-derived metabolites possess substantial medicinal values and a large structural diversity that can provide an untapped potential for drug candidates and medications. This review summarized 139 fungal phenalenone derivatives and their biosynthesis and biological activities, reported from 2014 until August 2021. Most of them are mainly identified from Penicillium (37 compounds), Coniothyrium (23 compounds), Aspergillus (22 compounds), and Talaromyces (22 compounds) ( Figure 24). Fungal phenalenones were derived from polyketide precursors that underwent different cyclization and tailoring reactions, leading to extreme structural diversity and high complexity. Hence, searching for diverse biosynthetic pathways for the various phenalenone derivatives will be a future challenge and offer an interesting research field for natural product researchers. These metabolites have been isolated and purified using various chromatographic tools, such as SiO2, Sephadex LH-20, preparative TLC, ODS, and HPLC, as well as preparative HSCCC-guided HPLC-HRESIMS, LC-MS-guided, and UV-HPLC guided analyses. Most of the separated phenalenones possess unique and unprecedented functionality or ring systems. Their configuration was assigned using various tools and experiments, such as X-rays, CD, ECD calculation, specific rotations, and chemical modifications, as well as the NOESY, NOEDIFF, and GIAO NMR shift calculations. They were evaluated for various activities, such as cytotoxic, antimalarial, antimycobacterial, anti-inflammatory, anti-angiogenic, immunosuppressive, and antioxidant properties, as well as IDO1, α-glucosidase (AG), ACE, IDO1, tyrosinase, and PTP inhibition (Figure 25). Some of the reported derivatives possessed powerful activities greater than in the used controls, such as anti-HIV (e.g., 71 and 74), immunosuppressive (e.g., 17), anti-tumor (e. g., 120, 125, 129, and 131), and antibacterial behavior (e. g., 24, 31,  69, 83, and 85), in addition to tyrosinase (e. g., 1-3), α-glucosidase (e. g., 29, 34, 46, and 47), Discovering bioactive phenalenones for drug use can be accelerated by applying and developing new technology, such as the methods of biosynthetic gene cluster (BGCs) activation for mining hidden new compounds [94], the use of metabolomic and genomic approaches [95], and modern machine deep learning techniques to discover the structurally distinct bioactive molecules [96,97]. Finally, we believe that the therapeutic potential and chemical diversity of the fungal phenalenones after more in-depth research will provide medicinal chemists and biologists with a more promising sustainable treasure trove for drug discovery. potential target for treating cancer. Based on the in silico studies, such as molecular docking, ADMET characteristics predication, and MD, some phenalenones were found to possess remarkable capacity as GLUT1 inhibitors; therefore, they could be potential leads for cancer treatment. It is noteworthy that the described results in this work are reported for the first time for this class of fungal metabolites and undoubtedly represent a substantial contribution in terms of further investigation, as well as in vitro and in vivo evaluations. Discovering bioactive phenalenones for drug use can be accelerated by applying and developing new technology, such as the methods of biosynthetic gene cluster (BGCs) activation for mining hidden new compounds [94], the use of metabolomic and genomic approaches [95], and modern machine deep learning techniques to discover the structurally distinct bioactive molecules [96,97]. Finally, we believe that the therapeutic potential and chemical diversity of the fungal phenalenones after more in-depth research will provide medicinal chemists and biologists with a more promising sustainable treasure trove for drug discovery.