verA Gene is Involved in the Step to Make the Xanthone Structure of Demethylsterigmatocystin in Aflatoxin Biosynthesis

In the biosynthesis of aflatoxin, verA, ver-1, ordB, and hypA genes of the aflatoxin gene cluster are involved in the pathway from versicolorin A (VA) to demethylsterigmatocystin (DMST). We herein isolated each disruptant of these four genes to determine their functions in more detail. Disruptants of ver-1, ordB, and hypA genes commonly accumulated VA in their mycelia. In contrast, the verA gene disruptant accumulated a novel yellow fluorescent substance (which we named HAMA) in the mycelia as well as culture medium. Feeding HAMA to the other disruptants commonly caused the production of aflatoxins B1 (AFB1) and G1 (AFG1). These results indicate that HAMA pigment is a novel aflatoxin precursor which is involved at a certain step after those of ver-1, ordB, and hypA genes between VA and DMST. HAMA was found to be an unstable substance to easily convert to DMST and sterigmatin. A liquid chromatography-mass spectrometry (LC-MS) analysis showed that the molecular mass of HAMA was 374, and HAMA gave two close major peaks in the LC chromatogram in some LC conditions. We suggest that these peaks correspond to the two conformers of HAMA; one of them would be selectively bound on the substrate binding site of VerA enzyme and then converted to DMST. VerA enzyme may work as a key enzyme in the creation of the xanthone structure of DMST from HAMA.


Introduction
Aflatoxins (AFs) are highly toxic, carcinogenic, teratogenic, and mutagenic secondary metabolites produced primarily by certain strains of Aspergillus flavus and Aspergillus parasiticus [1]. Other strains (A. nomius, A. pseudotamarii, A. bombycis and more) have also been reported to produce AFs, and aflatoxigenic fungi were shown to be widely distributed mainly in tropical and subtropical regions [2][3][4][5][6]. The AF contamination of crops such as corn, cotton, and nuts has serious effects on the health of animals and humans, and it also causes economic problems due to crop losses [7][8][9].
Aflatoxigenic fungi can potentially produce eight types of AF: the four major AFs, i.e., aflatoxins B 1 (AFB 1 ), B 2 (AFB 2 ), G 1 (AFG 1 ), and G 2 (AFG 2 ) and the four minor AFs, aflatoxins M 1 (AFM 1 ), Figure 1. Outline of the biosynthetic pathway of aflatoxins (AFs) and the AF gene cluster. (A) The formations of 1-group AFs (AFB1 and AFG1) from VA and 2-group AFs (AFB2 and AFG2) from VB. In the biosynthesis of AF, there are pathways for the formation of 1-group AFs (AFB1 and AFG1) from VA and 2-group AFs (AFB2 and AFG2) from VB. VB is converted to VA, and VA and VB respectively serve as precursors of 1-group and 2-group AFs. ver-1 (aflM), ordB (aflX), hypA (aflY), and verA (aflN) genes are suspected to be involved in the step from VA to DMST as well as the step from VB to DHDMST. dmtA (aflO) gene, which is a homolog of A. nidulans stcP gene, is involved in the conversion from DMST to ST as well as the conversion from DHDMST to DHST. The formations of 1-group AFs (AFB 1 and AFG 1 ) from VA and 2-group AFs (AFB 2 and AFG 2 ) from VB. In the biosynthesis of AF, there are pathways for the formation of 1-group AFs (AFB 1 and AFG 1 ) from VA and 2-group AFs (AFB 2 and AFG 2 ) from VB. VB is converted to VA, and VA and VB respectively serve as precursors of 1-group and 2-group AFs. ver-1 (aflM), ordB (aflX), hypA (aflY), and verA (aflN) genes are suspected to be involved in the step from VA to DMST as well as the step from VB to DHDMST. dmtA (aflO) gene, which is a homolog of A. nidulans stcP gene, is involved in the conversion from DMST to ST as well as the conversion from DHDMST to DHST. A. nidulans, A. versicolor and various fungal species are known to produce sterigmatocystin (ST) as a final product [21][22][23][24][25][26]. ST is one of the latter intermediates in the biosynthesis of AF. In fact, the enzymes, the intermediates, and the reactions for the biosynthesis of ST in A. nidulans are the same as those for AF biosynthesis [10,16,25].
At least 25 genes coding for the enzymes and regulatory factor(s) needed for AF biosynthesis are clustered in a 70-kb DNA region ( Figure 1C) [10,16,17]. Although functions of most of these genes have been clarified, some remain to be identified. The pathway from VA to DMST is one of these undetermined steps. Enzymes encoded by hypA (aflY) [27], ordB (aflX) [28], and ver-1 (aflM) [29] are known to be involved in the pathway, as the deletion of each of them caused an accumulation of VA with a remarkable decrease or loss of AF production ( Figure 1A). The verA (aflN) gene of A. parasiticus is suspected to be involved in the conversion of VA to DMST, as the deletion of stcS gene of A. nidulans caused an accumulation of VA in its mycelia with a decrease in the production of ST [30], and the A. nidulans stcS gene was suggested to be a homolog of verA gene based on the relatively high similarity (75%) of their amino acid sequences [16].
Hamasaki et al. [31] first isolated sterigmatin (STM) as a metabolite of A. versicolor (Vuillemin) Tiraboschi, and they suggested that STM is an isomer of DMST because STM has a linear [3,2-b]fusion structure of the xanthone and dihydrobisfuran moieties ( Figure 1B), whereas DMST has a bending [2,3-c]fusion structure of the same moieties [32]. Fukuyama et al. also suggested that STM is produced from VA, and that a putative precursor might work to make either DMST or STM, which depends on the manner of ether linkage formation to build a xanthone ring [33]. Austocystins A-I, which are secondary metabolites produced by Aspergillus ustus, have linear structures that are similar to that of STM, and they have been proposed to be derived from VA [34].
In the biosynthesis of AFs, the xanthone formation step corresponds to the production of carcinogenicity; ST derivatives and AFs are known to have carcinogenicity and acute toxicity [22,35]. In contrast, anthraquinone derivatives such as VA and VB have low or not-yet-identified toxicity [36]. To clarify the important step of the appearance of toxicities, we herein investigated the pathway from VA to DMST in detail. In this work, we isolated each disruptant of ver-1, hypA, ordB, and verA genes ( Figure 1C) and then characterized the accumulating substances in the culture medium or mycelia. The results demonstrated that: (1) verA disruptant accumulated mostly a novel intermediate, which we named HAMA after the late Professor Takashi Hamasaki, and (2) the verA gene is involved in the pathway from HAMA to DMST through a regiospecific xanthone ring formation. We also investigated accumulating substances in the deletion mutant of A. nidulans stcP gene [37], which is a homologue of the dmtA (aflO) gene of A. parasiticus [38]. These genes are involved in the reaction from DMST to ST. We evaluated these substances to determine whether HAMA is produced before the formation of DMST in AF biosynthesis as well as ST biosynthesis. We propose possible reaction steps in the pathway from VA to DMST.

Preparation of verA Disruptant
verA gene of A. parasiticus SYS4 was disrupted via the double-crossover strategy by using the linear DNA fragments of pVERA-DD, which was constructed with the flanking sequences of verA and disruption vector (Figure 2A). The verA disruptant was then isolated through the transformation of SYS4 with the linear DNA fragments of pVERA-DD. We obtained six mutants, three of which were then confirmed to be the right disruptants by PCRs using primer pairs P4/P5, P1/P2, and P3/P6 ( Figure 2B) (Table 1) and a Southern analysis ( Figure 2C). In the latter analysis, the verA gene coding region was presented on an approx. 3.20 kb EcoRI fragment of the wildtype SYS4 genome, whereas verA disruptant DNA gave no hybridization band ( Figure 2C, Left). When ptrA gene was used as a probe, beside the band that was suspected to correspond to the putative homologous nmt-1 gene in the A. parasiticus genome [39], specific ptrA hybridization signals of 2.06 kb for EcoRI digestion were detected only in the genomic DNA of verA disruptant ( Figure 2C, Right). These results indicate that the verA gene was replaced with the ptrA-selectable marker in the genomes of the mutants. the A. parasiticus genome [39], specific ptrA hybridization signals of 2.06 kb for EcoRI digestion were detected only in the genomic DNA of verA disruptant ( Figure 2C, Right). These results indicate that the verA gene was replaced with the ptrA-selectable marker in the genomes of the mutants.  Colonies of verA disruptant did not show any clear difference in phenotype such as mycelia color or size from those of the wildtype. In contrast, the colonies on the GY agar plate showed white images on UV photos, indicating that the resulting mutants had lost or decreased AF productivity. verA gene was thus suggested to be involved in the biosynthesis of AFs.

Characterizations of the Substances Accumulating in the verA Disruptant
The verA disruptant and wildtype SYS4 were cultured by the tip culture method. The TLC analysis (Figure 3) of the culture medium showed that the verA disruptant accumulated a novel yellow fluorescent substance in both the culture medium and mycelia ( Figure 3A). Interestingly, we observed that the yellow color of the substance on the TLC plate changed gradually to a dark red color after the plate was left overnight, suggesting that this substance may be an unstable one. The TLC analysis of the culture medium as well as that of the mycelial extract also showed that the verA disruptant still produced small amounts of AFs ( Figure 3A), and when the extract of the culture medium was analyzed by HPLC, a negligible amount of AFB 1 and a small amount of AFG 1 were detected ( Table 2).
To determine the step(s) involving the verA gene in AF biosynthesis, we performed feeding experiments using several precursors of AFs. When the verA disruptant was incubated with DMST, ST, or OMST, these substances were converted to AFB 1 and AFG 1 , whereas when either VHA or VA was used, AFs were not produced ( Table 3), indicating that verA gene is involved in a certain step between VA and DMST in the biosynthesis of AF.
For the characterization of the yellow pigment, we isolated the pigment from the TLC plates after the culture media were analyzed by TLC using solution A. When strain NIAH-26 was incubated with the pigment, significant amounts of AFB 1 and AFG 1 were produced ( Figure 3B), indicating that the pigment is a precursor of these AFs.
We further analyzed the yellow pigment by TLC. Since the pigment spot appeared to be broad in the TLC analysis ( Figure 3A), we first divided the spot into three parts ( Figure 3D, Left panel), and the pigment was extracted from each part. Each extract was then analyzed by TLC using another solvent. Four dark red substances were newly detected, and the Rf values of the upper two bands were the same as DMST and STM, suggesting that the yellow pigment could be a precursor of DMST as well as STM. We herein named this pigment "HAMA" after Professor. Takashi Hamasaki. Two other substances (indicated by R1 and R2) were also detected ( Figure 3D, Right panel). However, we observed that significant amounts of AFs were not produced by either R1 or R2 in feeding experiments using strain NIAH-26. We speculate that these are dead-end products produced by HAMA, and they remain to be studied.  The drying of TLC-purified HAMA under N 2 gas followed by solubilization was repeated five times, followed by an HPLC analysis as described in panel C. Standard samples of DMST, STM, and DHDMST were also analyzed. (F) The production of DMST and STM from HAMA by the repetition of drying and solubilization. The same procedures as those described in panel E were done from zero to five times, followed by an HPLC analysis.   To confirm the productions of DMST and STM from the HAMA, we used HPLC to analyze the samples directly extracted from the culture media of the verA disruptant ( Figure 3C). A peak around 2 min was detected, whereas peaks of DMST and STM were not detected. In contrast, the TLC-purified samples (after the TLC and the scraping off of the HAMA spot, then extraction followed by drying and solubilization with a small amount of methanol) showed productions of DMST and STM ( Figure 3F, 0-time drying). When the same substances underwent repeated drying with N 2 gas followed by solubilization with methanol, the amounts of DMST and STM similarly increased with the number of repetitions ( Figure 3F). The HPLC analysis of the five-repetitions samples showed that large amounts of DMST and STM were produced. Interestingly, a small amount of DHDMST and another substance were also detected, and the latter substance was suggested to be dihydrosterigmatin (DHSTM) based on the similarity of the elution patterns of DHDMST and DMST ( Figure 3E). These results suggest that the HAMA sample extracted from the TLC plate were contaminated with a small amount of dihydroHAMA (DHHAMA).
We also investigated the heat stability of the HAMA pigment. After culture medium of the verA disruptant was autoclaved at 121 • C for 15 min, the resulting solution was fed to strain [pks-fas-1] − . When the treated medium was used, the amounts of AFs were decreased to approx. 10% of those obtained when we used the intact medium (Table 4, Exp. 1). When the same medium was treated using either acid condition (pH 1.5) or alkaline condition (pH 11.0), the amounts of AFB 1 and AFG 1 decreased to 50%-70% of those when we used the intact verA culture medium (Table 4, Exp. 2). These results suggested that HAMA is unstable against heat treatment and partially sensitive to either acid or alkaline treatment. Although small amounts of AFB 2 and AFG 2 were also produced from the HAMA pigment, it was suggested that HAMA pigment sample was contaminated with small amount of dihydroHAMA.

Relationships among the verA, ver-1, hypA and ordB Disruptants
Culture media as well as mycelial extracts of the ver-1, hypA, and ordB disruptants were analyzed by TLC after tip culture. They predominantly accumulated VA as a major intermediate in their mycelia ( Figure 4A). Each of the disruptants also produced smaller amounts of some substances, which appeared to be different depending on the disruptant. When each substance was extracted from a TLC spot and then analyzed by HPLC, we observed that the hypA disruptant accumulated a small amount of 6-deoxyversicolorin A (deoxyVA) together with VA. The ver-1 and hypA disruptants completely lost AF productivity, whereas the ordB disruptant produced small amounts of AFs ( Figure 4A). To investigate relationship among ver-1, hypA, ordB, and verA genes in AF biosynthesis, we co-cultured each combination of two of them and analyzed AFs contained in the resulting culture medium ( Figure 4B,C). All combination enhanced AF production, indicating that all genes have different functions each other although all accumulated mainly VA.  Since the culture medium of the verA disruptant contained a significant amount of HAMA pigment ( Figure 3A), we investigated the AF levels in the ver-1, hypA and ordB disruptants fed with filter-sterilized culture medium of verA disruptant ( Figure 4D). The addition of the medium of the  , 4), hypA (5, 6), or ver-1 disruptant (7, 8) were analyzed by TLC using solution B. S1: AF mixture standard, S2: VA authentic sample. (B) TLC analysis of the AF production by each of the ordB, hypA and ver-1 disruptants or their co-culture. The culture media (30 µL) of each of the ordB (1), hypA (2) and ver-1 (3) disruptants, or the co-culture of ordB and hypA disruptants (4), the co-culture of ordB and ver-1 disruptants (5), the co-culture of the ver-1 and hypA disruptants (6) were analyzed by TLC using solution (A). (C) TLC analysis of AF production in coculture of verA disruptant and each of ordB deleted mutant (lower). The culture media of each of ver-1-(1), hypA (2) or ordB (3) disruptant, or the co-culture of verA disruptant and each of ver-1 (4), hypA (5) or ordB (6) disruptants and SYS4 strain (7) together with AF standard sample (AFs) were also analyzed. (D) The feeding of disruptants with HAMA in the culture medium of verA disruptant. Each of NIAH-26 (1), ver-1 (2), hypA (3), and ordB (4) disruptants was cultured in YES medium (upper panel) or YES medium supplemented with the same volume of the culture medium of verA disruptant. The resulting culture medium was then analyzed by TLC.
Since the culture medium of the verA disruptant contained a significant amount of HAMA pigment ( Figure 3A), we investigated the AF levels in the ver-1, hypA and ordB disruptants fed with filter-sterilized culture medium of verA disruptant ( Figure 4D). The addition of the medium of the verA disruptant significantly enhanced the productions of AFB 1 and AFG 1 , indicating that the HAMA substance is a precursor at a certain step after any steps involving the ver-1, hypA and ordB genes. The productions of AFB 1 and AFG 1 by strain NIAH-26 were also confirmed ( Figure 4D).

The LC-MS Analysis of HAMA Pigment
The LC-MS analysis using condition [A] revealed that HAMA purified by TLC had a peak whose retention time was at 6.11 min in the LC chromatogram ( Figure 5A). HAMA's molecular related ions in electrospray ionization-mass spectrometry (ESI-MS) spectra were observed at m/z 374. The retention times of other smaller peaks (22.7 min and 23.8 min) on LC chromatograms were coincident to those of authentic samples of DMST and STM, respectively, and their molecular masses were commonly 310 in APPI-MS spectra ( Figure 5B-c,d). These LC-MS data demonstrated that the substances produced through the repetition of drying followed by solubilization were confirmed to be DMST and STM. We also analyzed the extract of the verA disruptant culture medium to examine the fresh sample of HAMA in more detail using condition [B]. We obtained the same results as those of the repeated dried sample ( Figure 5B, LC) except that the peaks of DMST and STM were remarkably small.
To confirm the reproducibility of HAMA's twin peaks, we further analyzed the same fresh extract of the verA disruptant culture medium by LC-MS using condition [C]. Two main peaks were observed at 27.2 min and 32.0 min on LC chromatograms ( Figure 5C). The APPI-MS spectra obtained for either peak commonly gave ions of m/z 375 [M + H] + (100%) ( Figure 5C-e,f). These results indicated that HAMA is composed of two isomers or conformers, and their molecular masses are both 374.

Accumulation of Small Amounts of STM Together with DMST in A. nidulans stcP-Disruptant
Strain TAHK64.42 is a stcP disruptant of A. nidulans FGSC26 [37], and stcP gene is a homolog of dmtA gene of A. parasiticus [38]. After these strains were incubated in oat flake medium, metabolites produced by either strain were analyzed by TLC ( Figure 6A). Strain TAHK64.42 newly produced a large amount of DMST and a small amount of STM. The production of STM in this strain was also confirmed by an HPLC analysis ( Figure 6B). The detection of STM suggested that HAMA accumulated together with DMST in the stcP disruptant and would be non-enzymatically changed to STM during the extraction procedure in this work. We speculate that HAMA was formed before a formation step of DMST in the biosynthesis of ST.

verA Gene and HAMA Intermediate
This work demonstrated that verA gene is involved in building the xanthone moiety of DMST or DHDMST in the biosynthesis of AF. The postulated pathway of the conversion of VA to DMST is shown in Figure 7. The verA disruptant accumulated a novel intermediate, HAMA, which is a precursor of AFB1/AFG1 ( Figure 3B). The molecular mass of HAMA was 374, and HAMA changed non-enzymatically to DMST and STM at the ratio of 1:1 ( Figure 3E,F). Since TLC-purified HAMA also contained much smaller amounts of DHDMST and a metabolite presumed to be DHSTM ( Figure 3E), DHHAMA seemed to be contaminated in the HAMA sample. We thus suspect that verA gene is involved in the reaction from HAMA to DMST as well as the reaction from DHHAMA to DHDMST in AF biosynthesis. STM and DHSTM have not been isolated from aflatoxigenic fungi because the HAMA and DHHAMA that are transiently formed as intermediates of AFs are immediately changed to DMST and DHDMST, and to AFs. In contrast, STM was isolated from A. nidulans stcP disruptant because the conversion step from DMST to ST was completely blocked by the gene disruption ( Figure  6A). Kelker et al. [37] showed that a significant amount of the substance thought to be STM together with DMST was produced in the A. nidulans stcP disruptant. Like A. nidulans, A. versicolor species produce ST as an end-product, and Hamasaki et al. first isolated STM from A. versicolor (Vuillemin)

verA Gene and HAMA Intermediate
This work demonstrated that verA gene is involved in building the xanthone moiety of DMST or DHDMST in the biosynthesis of AF. The postulated pathway of the conversion of VA to DMST is shown in Figure 7. The verA disruptant accumulated a novel intermediate, HAMA, which is a precursor of AFB 1 /AFG 1 ( Figure 3B). The molecular mass of HAMA was 374, and HAMA changed non-enzymatically to DMST and STM at the ratio of 1:1 ( Figure 3E,F). Since TLC-purified HAMA also contained much smaller amounts of DHDMST and a metabolite presumed to be DHSTM ( Figure 3E), DHHAMA seemed to be contaminated in the HAMA sample. We thus suspect that verA gene is involved in the reaction from HAMA to DMST as well as the reaction from DHHAMA to DHDMST in AF biosynthesis. STM and DHSTM have not been isolated from aflatoxigenic fungi because the HAMA and DHHAMA that are transiently formed as intermediates of AFs are immediately changed to DMST and DHDMST, and to AFs. In contrast, STM was isolated from A. nidulans stcP disruptant because the conversion step from DMST to ST was completely blocked by the gene disruption ( Figure 6A). Kelker et al. [37] showed that a significant amount of the substance thought to be STM together with DMST was produced in the A.  Aspergillus ustus produces austocystins A-I as its secondary metabolites. Like STM, these substances have a linear fusion structure of the xanthone and dihydrobisfuran moieties. Since A. ustus does not produce DMST, another enzyme for regiospecific ring closure to make the linear structure of austocystins A-I would work in this species. A comparison of this possible enzyme to VerA enzyme would be useful for gaining a better understanding of the stereospecific reactions of the intermediates.
We also observed that the heat treatment of the culture medium of verA disruptant with autoclaving greatly decreased (7-13%) the production of AF when the treated culture medium was used in the feeding experiment (Table 4, Exp. 1), suggesting that heat treatment caused a disruption of HAMA's structure. In contrast, the acid or alkaline treatment of the culture medium caused a partial decrease (89-43%) of AF production in the feeding experiments (Table 4, Exp. 2). We suspect that this partial inhibition is due to the productions of DMST and STM from HAMA by the pH changes. The structural changes of HAMA caused by these treatments remain to be studied.
The HPLC analysis revealed that the verA disruptant produced a slight amount of AFG1 (Table  2). However, the TLC analysis of the culture medium and the mycelial extract of the verA disruptant indicated that they produced small but significant amounts of AFB1 and AFG1, suggesting that the disruptant might be a leaky mutant ( Figure 3A, lanes 1 and 2). However, these results are not easily understood because verA gene encodes a cytochrome P450 monooxygenase, and the reaction for Aspergillus ustus produces austocystins A-I as its secondary metabolites. Like STM, these substances have a linear fusion structure of the xanthone and dihydrobisfuran moieties. Since A. ustus does not produce DMST, another enzyme for regiospecific ring closure to make the linear structure of austocystins A-I would work in this species. A comparison of this possible enzyme to VerA enzyme would be useful for gaining a better understanding of the stereospecific reactions of the intermediates.
We also observed that the heat treatment of the culture medium of verA disruptant with autoclaving greatly decreased (7-13%) the production of AF when the treated culture medium was used in the feeding experiment (Table 4, Exp. 1), suggesting that heat treatment caused a disruption of HAMA's structure. In contrast, the acid or alkaline treatment of the culture medium caused a partial decrease (89-43%) of AF production in the feeding experiments (Table 4, Exp. 2). We suspect that this partial inhibition is due to the productions of DMST and STM from HAMA by the pH changes. The structural changes of HAMA caused by these treatments remain to be studied.
The HPLC analysis revealed that the verA disruptant produced a slight amount of AFG 1 (Table 2). However, the TLC analysis of the culture medium and the mycelial extract of the verA disruptant indicated that they produced small but significant amounts of AFB 1 and AFG 1 , suggesting that the disruptant might be a leaky mutant ( Figure 3A, lanes 1 and 2). However, these results are not easily understood because verA gene encodes a cytochrome P450 monooxygenase, and the reaction for making the xanthone structure seems to be too unique for another enzyme to be substituted for VerA enzyme. It is possible that a small amount of HAMA accumulates in cells of the verA disruptant and then non-enzymatically change to DMST in the cells, and the resulting DMST is fed by the verA disruptant to produce AFB 1 and AFG 1 . A future study should determine whether the verA disruptant is leaky.

The Postulated Pathway from VA to HAMA
A. parasiticus verA (aflN) gene was originally suspected to be a homolog of A. nidulans stcS gene based on the homology (75%) of their deduced amino acid sequences [16]. Our present findings confirmed that the verA gene is not a homolog of A. nidulans stcS gene. The pathway from VA to DMST contains a step for the formation of the xanthone structure from the anthraquinone moiety in the biosynthesis of AF. We propose that the ver-1, ordB, and hypA disruptants are involved in the conversion of VA to HAMA pigment, and that the verA gene is involved in the reaction from HAMA to DMST (Figure 7).
Although the ver-1 and hypA disruptants did not produce any AFs, the ordB disruptant produced small amounts of AFs ( Figure 4A), which was previously reported [27]. The co-culture of two of verA, ver-1, ordB and hypA disruptants remarkably enhanced the production of AFs ( Figure 4B,C), indicating that these genes have differing functions in AF biosynthesis. Ehrlich et al. [27] suggested that the common accumulation of VA by the ver-1, ordB and hypA disruptants may be due to their work as an enzyme complex in AF biosynthesis. In the present study, we searched for any related enzymatic activities in the pathway from VA to DMST by using cell-free systems of each disruptant, and we did not detect any significant enzyme activities. The detailed enzyme reactions remain to be investigated.
The pathway from VA to DMST has been speculated to be composed of the epoxidation of the anthraquinone moiety of VA, reduction, Baeyer-Villiger oxidation, and decarboxylation [40]. We herein hypothesize the existence of a reaction scheme in which intermediates 1-8 are involved in the pathway from VA to DMST (Figure 7). Since ordB is thought to encode a metal-oxidase [28], we suggest that OrdB enzyme may catalyze the epoxidation of VA and the subsequent epoxide ring-opening step of intermediate 1. In contrast, ver-1gene had been suggested to encode an NADPH-dependent reductase [29]. Since the reactions from intermediates 2 to 4 are composed of two successive reduction reactions, we suspect that the Ver-1 enzyme catalyzes these reactions.
We observed that these disruptants also accumulated minor products together with VA and that the minor products were different depending on each disruptant. Among them, we determined that hypA disruptant accumulated a small amount of 6-deoxyversicolorin A, and the same result was obtained by Erlich et al. [27]. In contrast, ver-1 and ordB disruptant accumulated small amounts of other substances, which were not identified herein. Although HypA enzyme has no typical homologous domain to other proteins, Erlich et al. [27] expected that the amino acid sequence of HypA enzyme would have partial similarity to those of some enzymes related to a Baeyer-Villiger reaction. We speculated that HypA enzyme may catalyze the reaction from intermediate 5 to intermediate 6 and the next hydrolytic ring opening of intermediate 6 to afford HAMA 7. The involvement of HypA enzyme in this step may explain why the hypA disruptant produced a small amount of 6-deoxyVA together with the major product, VA. The 6-deoxyVA is suspected to have been produced from intermediate 5 through a dehydration reaction. Since strain NIAH-26 fed with 6-deoxyVA did not produce any AFs, we speculate that 6-deoxyVA is a dead-end product derived from an intermediate for AF production.

The Postulated Pathway from HAMA to DMST
VerA enzyme catalyzes the reactions from HAMA 7 to DMST (Figure 7). The enzyme catalyzes two steps: the concerted decarboxylation and dehydration of HAMA 7 to afford intermediate 8, and the regiospecific ring-forming dehydrogenation of 8 to give DMST. In the LC-MS analyses, HAMA gave two peaks ( Figure 5), suggesting some possibilities regarding its structural features; one is that HAMA gives two conformers (7a and 7b) because of a restricted rotation around the bond with steric hindrance as well as a hydrogen bond between the ketone carbonyl and phenolic hydroxyl as shown in Figure 7. The other possibility is that HAMA has two chiral centers on the cyclohexenone moiety and gives two diastereomers. The reaction steps from VA to HAMA are enzyme-catalyzed steps and are suspected to be stereospecific. Therefore, the former has higher possibility.
Both conformers 7a and 7b could be changed to the same intermediate 8.
We also speculate that one of the two conformers 7a and 7b would be selectively bound on the substrate binding site of VerA enzyme, and then converted to intermediate 8 and DMST. The remaining conformer would be changed to the other conformer that is suitable for the VerA enzyme reaction through a rotation around the bond at C-9 and C-9a of the HAMA molecule.
Our present results demonstrated that HAMA can non-enzymatically change to DMST and STM at the ratio of~1:1 in their quantities ( Figure 3D-F and Figure 5B). The yellow color of the spots of HAMA on the TLC plates changed to dark red, which is the same color as those of ST derivatives such as DMST and STM under UV light. The repetitions of the drying and solubilization of the HAMA increased the amounts of DMST and STM. These results indicated that decarboxylation, dehydration, and dehydrogenation can occur spontaneously. Since STM did not serve as a precursor of AFs in the feeding experiment with NIAH-26 mutant, we speculate that STM is an artificial dead-end product from intermediate 8.
Therefore, VerA enzyme is the key enzyme for the formation of the xanthone structure of various ST derivatives. Since acute toxicity and carcinogenicities have been reported after xanthone structure formation, VerA enzyme is also involved in the appearance of these toxicities. We also speculate that because aflatoxigenic fungi produce 2-group AFs (AFB 2 and AFG 2 ) together with 1-group AFs (AFB 1 and AFG 1 ), VerA enzyme is involved in the reaction from VB to DHDMST through the formation of DHHAMA.

Fungal Strains
A. parasiticus SYS4 (NRRL-2999), the wild-type AF-producing strain, was used as a recipient strain for gene disruption. A. parasiticus NIAH-26, a UV-irradiated mutant from strain SYS4, produces all enzymes in the AF biosynthesis pathway from norsolorinic acid to AFs, although it produces neither AFs nor any precursors [41,42]. We used the NIAH-26 mutant for the feeding experiments. We isolated disruptants of verA, ver-1, hypA and ordB genes herein and used them for the characterization of function of verA gene. A. parasiticus [pks-fas-1] − , which has the deletion of the 18-kb region from the pks gene to fas-1 in the AF gene cluster of strain NR-1 [21], was used for the examination of the stability of the HAMA substance. Strains A. nidulans FGSC26 and TAHK64.42 (stcP deletion mutant of FGSC26) were a kind gift from Dr. Nancy Keller [37].
The concentrations of the metabolites in methanol were determined from UV absorption spectra by using the following molar absorption coefficients (λ max ) [22,36]

Media and Culture Conditions
For the production of AFs and metabolites produced by fungi, YES medium (2% yeast extract, 20% sucrose) or GY agar medium (2% glucose, 0.5% yeast extract, and 2.0% agar) was used. For the culture of A. nidulans strains, oat flake medium (3 g of oat flakes and 0.5 mL of water) (Keller medium) was used.
For the standard culture followed by a thin-layer chromatography (TLC) or high-performance liquid chromatography (HPLC) analysis, the tip culture method was used [41]. Spores (approx. 1 × 10 −4 ) of each SYS4 and various gene disruptants were inoculated into 250 µL of YES medium in a 1-the ml pipetman tip. After culture at 28 • C for 4 days, culture media and the mycelia were separated by centrifugation and then used for further analyses. For the preparation of larger amounts of intermediates for the physico-chemical analyses of HAMA, the verA disruptant was cultured in 100 mL of YES medium at 28 • C for 4 days [49]. Aliquots of the medium were used for the TLC and the LC-MS.

Preparation of verA Disruptant
To construct the verA disruption vector, three steps were taken: the 2-kb PstI/KpnI polymerase chain reaction (PCR) fragment (pcr1) of the selectable marker gene ptrA [50] was amplified with primer pair P19/P20 (Table 1) and inserted into the corresponding sites in the pSP72 vector ( Figure 2B). Then, a 1.2-kb fragment of the upstream of verA and the 3 coding region of ver-1 (pcr2) and a 1.45-kb fragment of the downstream of verA and the 3 coding region of avnA (pcr3) were amplified using A. parasiticus SYS4 genomic DNA as the template. The two primer pairs P7/P8 and P9/P10 with their restriction endonuclease sites are underlined and the enzymes are indicated in parentheses in Table 1. The XhoI-PstI fragment from pcr2 and the KpnI-BglII fragment from pcr3 were cloned into the corresponding sites in the pSP72 bearing ptrA to give the disruption vector, pVERA-DD. The sequence of pVERA-DD was checked by a restriction analysis. For transformation, pVERA-DD was linearized with Cfr9I to release the 4.5-kb insertion part from the pSP72 vector.
For fungal transformation, the preparation of the protoplasts from A. parasiticus SYS4 and the transformation with DNA were performed as described by Wen et al. [49] with some modifications. For the transformation of Cfr9I-digested pVERA-DD, 4-8 µg of DNA was added to 0.1 mL of protoplasts. The pyrithiamine (PT)-resistant transformants were screened on CDA regeneration medium with 0.1 µg mL −1 pyrithiamine [50]. Approximately 90-120 PT-resistant transformants per 6 µg of DNA was obtained. The transformants were then transferred to a GY agar plate and cultured.
The desired verA disruptant was then selected by assessing the impairment of AF production by UV photography [51]. On a UV picture taken under 365-nm UV light, the AF production-impaired disruptants appeared as white colonies, and the AF-producers were identified as gray or black colonies. Confirmation of the deletion was performed by PCR analysis. DNA of the white mutant was prepared using FastPrep FP100A (Qbiogene, Carlsbad, CA, USA) as described [49].
To investigate the replacement of the verA gene with the ptrA marker gene by PCR, we used three PCR primer pairs (Table 1): P3/P6 and P4/P5 for testing the insertion of ptrA into SYS4 genomic DNA, and P1/P2 for testing the deletion of verA. After we compared the PCR results for transformants with those for the recipient strain, the selected verA disruptant was purified three times by single colony isolation on GY plates. The mutation in these selected verA − /SYS4 transformants was re-confirmed by PCR. The deletion of the verA gene in the genome was also confirmed by a Southern analysis.
Total genomic DNA of verA mutant and strain SYS4 was subjected to restriction enzyme digestion with EcoRI and separated by agarose gel electrophoresis, followed by blotting to a Hybond-N+ membrane (GE Healthcare, Buckinghamshire, UK). The filters were hybridized with the verA and ptrA probes. The 0.83-kb verA probe was amplified from genomic DNA using primer pairs verA-F442 [5 -GTTTCGACTCCCTCGGC-3 ] and verA-R2922 [5 -TGCGGCCCTGAGCTTCT-3 ]. The 2.0-kb ptrA probe was amplified from pPTRI plasmid (Takara Bio, Shiga, Japan) using the primers as described [52]. Hybridization and detection were performed using the Alkphos Direct Labelling and Detection System (GE Healthcare, Chicago, IL, USA) according to the supplier's manuals.

Preparation of ver-1, ordB, and hypA Deletion Mutants
To delete the ordB gene in the genome of A. parasiticus SYS4, we constructed the ordB gene disruption plasmid, pORDB-L/R, in a two-step procedure. A 1.4-kb fragment (ordB-R) of the 3 -flanking region of ordB gene including the 3 non-coding sequence of the gene and the coding sequence of moxY gene and a 1.4-kb fragment (ordB-L) of the 5 -flanking region of the ordB gene including the 3 -non-coding region and the coding sequence of hypA gene were amplified by PCR using two sets of primer pairs with endonuclease restriction sites underlined to facilitate sub-cloning. The resulting PCR products were successively cut with KpnI for the 3 -flanking region and HindIII/SalI for the 5 -flanking region, and then inserted into the pSP72-ptrAKpnI/EcoRV and HindIII/XhoI sites to give a pORDB-L/R, which was then linearized by BglII and Eam1105I. The final linear replacement construct, which contained the 2-kb selectable marker ptrA flanked by a 1.4-kb fragment of the 3 -flanking region and a 1.35-kb fragment of the 5 -flanking region was used for transforming the SYS4 strain. Fungal protoplast preparation and transformation with DNA were performed as described [49]. The PT-resistant transformants were screened for CDA regeneration medium with 0.1 mg L −1 PT as described above or as described by Kubodera et al. [50] and then transferred to a GY agar plate to detect decreases in AF production and accumulations of the precursors. The genomic DNAs of the resulting RT-resistant transformants were prepared for a PCR analysis to confirm the detection of the ordB gene in the transformants.
To delete the hypA gene in the genome of A. parasiticus SYS4, we constructed the hypA gene disruption plasmid, pHYPA-L/R, in a two-step procedure. The resulting PCR products were successively cut with SalI/HindIII for the 3 -flanking region and KpnI for the 5 -flanking region, and then inserted into the pSP72-ptrA XhoI/HindIII and KpnI/EcoRV sites to give a pHYPA-L/R, which was then linearized by BglII. The final linear replacement construct, which contained the 2-kb selectable marker ptrA flanked by a fragment of the 3 -flanking region and a fragment of the 5 -flanking region was used for transforming the SYS4 strain. The transformation, screening, and characterization of the PT-resistant transformants were performed as described above.
To delete the ver-1 gene in the genome of A. parasiticus SYS4, we made the ver-1 gene disruption construct in a two-step procedure. A 1.1-kb fragment of the 5 -flanking region of ver-1 gene including the 5 non-coding upstream sequence of the gene and the coding sequence of norA gene and a 1.1-kb fragment of the 3 -flanking region of the ver-1 gene including the 3 -non-coding downstream region of ver-1 and the coding sequence of verA gene were amplified by PCR using two sets of primer pairs with endonuclease restriction sites underlined to facilitate sub-cloning. Pair 1: norA-XhoI-F [5 -ATGCATTTGCTCGAGCCAACGGACTTACCC-3 ] and ver-1-up-PstI-R [5 -AAAACTGCAGCTCTGCCTCTATCCAAAGCC-3 ]. Pair 2: ver-1-down-KpnI-F [5 -GGGGTACCCGCTATATACTCGTGGGTGA-3 ] and verA-BglII-R [5 -GCCGGACGAGAATTGCTG GGAGATCTTCAG-3 ]. A 2-kb selectable marker ptrA gene was amplified by PCR using the primer pairs pPTRI-KpnI-F [5 -GGGGTACCGGGCAATTGATTACGGGATCCCA-3 ] and pPTRI-PstI-R [5 -AAAACTGCAGTGACGATGAGCCGCTCTTGC-3 ] and pSP72-ptrA. The resulting three PCR products were combined by ligase, and the final linear replacement construct was used for transforming SYS4 strain. The transformation, screening, and characterization of the PT-resistant transformants were performed as described above.

TLC and HPLC Analyses of AFs
After the tip cultures of SYS4 or each of the disruptants and then centrifugation, metabolites in the culture medium were extracted with an equal volume of chloroform, and the resulting extract was analyzed by TLC or HPLC. For detection of metabolites in mycelia, the mycelia in the tip was extracted with 800 µL acetone, and the resulting acetone extract was air-dried by keeping it at room temperature in dark condition. The resulting debris was solubilized by adding 100 µL of benzene:acetonitrile (98:2, v/v), and the extract was analyzed by TLC.
The extracts of culture medium and mycelial extract were spotted onto a silica gel TLC plate (Silica Gel 60; Merck, Darmstadt, Germany) and then developed with solution A containing chloroform:ethyl acetate: 90% formic acid (6:3:1, v/v/v). Fluorescence of the metabolites on the TLC plate was observed under 365-nm UV light and photographed with an Olympus C-745 digital camera. In some cases, the TLC plate was observed under 300-nm UV light and photographed with Fluor-S™ MultiImager (Bio-Rad, Hercules, CA, USA) for the detection of AFs. For the characterization of substances derived from HAMA or metabolites, strain NIAH-26, ordB-, hypA-, and verA-disruptants were also analyzed by TLC using solution B containing benzene:ethyl acetate (8:2, v/v) with 10% acetic acid.

Feeding Experiments for Aflatoxin Production
We used the tip culture method for the feeding experiment ( Figure 3A). Conidiospores (approx. 1.5 × 10 5 ) of either verA disruptant or strain NIAH-26 was inoculated into 250 µL of YES medium supplemented with 30 µM VHA, VA, DMST, ST, or OMST and cultured at 28 • C for 4 days. The AFs in the media were measured by HPLC. For the feeding experiment with HAMA, strain NIAH-26 was incubated by tip culture using YES medium supplemented with TLC-purified HAMA. In addition, NIAH-26, ver-1 disruptant, hypA disruptant, and ordB disruptant were each individually incubated in tip cultures using YES medium supplemented with an equal volume of filter-sterilized 4-day cultured medium of verA disruptant for 4 days. The resulting medium was analyzed by either TLC or HPLC.

Co-Incubation of Two Disruptants among verA, ver-1, ordB, and hypA Disruptants
Conidiospores (approx. 1.5 × 10 5 ) of each two of four verA, ver-1, ordB and hypA disruptants were inoculated into the same tip culture and incubated at 28 • C for 4 days. After the separation of culture medium from mycelia by centrifugation, the culture medium was analyzed by silica-gel TLC.

Preparation of HAMA Pigment
For the preparation of HAMA pigment, the culture medium of the verA disruptant was extracted with an equal volume of ethyl acetate, and the resulting extract was used for TLC or LC-APPI-MS analysis. For the purification of HAMA, HAMA pigment in the culture medium of the verA disruptant was purified by using Diaion HP20 (Mitsubishi Chemical Co. Tohyo, Japan), and the resulting extract was concentrated and used for LC-ESI-MS. For small scale purification of HAMA pigment, the extract of the culture medium of verA disruptant was analyzed by TLC, and the yellow part (HAMA) on the TLC plate was scraped and then extracted with ethyl acetate. The resulting extract was concentrated and then used for characterization of HAMA.

Characterization of HAMA Pigment
For the detection of the formations of DMST and STM from HAMA, we further analyzed HAMA pigment (which had been purified by TLC using solution A) by performing a TLC analysis using solution B again ( Figure 3D). We measured the ethyl acetate extract of the culture medium of the verA disruptant or the TLC-purified HAMA by HPLC using the SCL-10Avp HPLC apparatus equipped with a UV-VIS spectrophotometric detector (model SPD-6AV, Shimadzu, Kyoto, Japan) and an octadecyl silane column (0.46 × 15 cm; STR ODS-II; Shinwa Chemical Industries, Kyoto, Japan) at room temperature. Absorption at 330 nm was monitored, and the solvent system was solution D: 75% methanol aqueous solution with a flow rate of 1 mL min −1 by monitoring absorption at 330 nm. The retention times of DHDMST, DMST, and STM were compared with those of the authentic standards (13.8 min, 15.0 min, and 19.8 min), respectively.
To test the stability of HAMA, we examined the heat stability of the pigment after the culture medium of the verA disruptant was autoclaved at 121 • C for 15 min and then cooled. The [pks-fas] − strain was cultured with the treated HAMA, and the resulting culture medium was analyzed by HPLC. The acid and alkaline stabilities of the pigments were also examined by changing the pH of the culture medium of verA disruptant to pH 1.5 or pH 11.0 by adding 1 M HCl or 1 M NaOH, followed by incubation at room temperature for 3 h. After the pH of the solution was readjusted by adding the same volume of 1 M NaOH or 1 M HCl, respectively, the resulting solution was used for the feeding experiment with strain NIAH 26 using the tip culture method.
For the determination of the effect of the repetition of drying followed by solubilization, we dried the extract of the TLC-purified HAMA after the elution of the TLC spot with ethyl acetate with N 2 gas in a microtube; the debris was then solubilized in methanol. These procedures of drying and solubilization were repeated several times. The resulting solutions were analyzed by HPLC using the ODS column and solution D.
Condition [C]: the LC-MS conditions were the same as those in condition [B] except that the carrier solvent was set as an isocratic mixture of methanol and 5 mM ammonium acetate in water (40:60, v/v).

Substances Accumulated by the stcP Disruptant of A. nidulans
We analyzed the metabolites produced by A. nidulans TAHK64.42 (stcP disruptant) and FGSC26 (the wild strain) as described [37] with minor modifications. Conidiospores (approx. 1.5 × 10 5 ) of either strain were cultured in oat flake medium (3 g oat flakes and 0.5 mL water) at 28 • C for 5 days. Metabolites of fungus ware extracted twice with the same volume of acetone-chloroform solution.
The resulting pooled extract was concentrated and then analyzed by silica-gel TLC using benzene:acetic acid (95:5, v/v). Compounds on the TLC plate were visualized by spraying 10% (wt./vol) aluminum chloride solution in ethanol and heating. For the quantitative measurement of the metabolites, the part corresponding to either spot of STM or DMST was scraped and extracted with ethyl acetate. The extract was then analyzed by HPLC using an ODS column and solution D.

Conclusions
We clarified that verA gene in the aflatoxin gene cluster is involved in the step to make the xanthone structure of demethylsterigmatocystin as well as dihydrodemethylsterigmatocystin in aflatoxin biosynthesis.