Absence of the aflatoxin biosynthesis gene, norA, allows accumulation of deoxyaflatoxin B₁ in Aspergillus flavus cultures

Abstract

Biosynthesis of the highly toxic and carcinogenic aflatoxins in select Aspergillus species from the common intermediate O-methylsterigmatocystin (OMST) has been postulated to require only the cytochrome P450 monooxygenase, OrdA (AflQ). We now provide evidence that the aryl alcohol dehydrogenase NorA (AflE) encoded by the aflatoxin biosynthetic gene cluster in A. flavus affects the accumulation of aflatoxins in the final steps of aflatoxin biosynthesis. Mutants with inactive norA produced reduced quantities of aflatoxin B₁ (AFB₁), but elevated quantities of a new metabolite, deoxyAFB₁. To explain this result, we suggest that, in the absence of NorA, the AFB₁ reduction product, aflatoxicol, is produced and is readily dehydrated to deoxyAFB₁ in the acidic medium, enabling us to observe this otherwise minor toxin produced in wild-type A. flavus.

Introduction

Many species of Aspergillus produce the xanthone metabolite sterigmatocystin (ST) (Scheme 1), but only a few are capable of converting ST to the far more toxic and carcinogenic aflatoxins (AFs: AFB₁, AFB₂, AFG₁, AFG₂) (Frisvad, et al., 2007). Because Aspergillus species are common agricultural contaminants and because ingestion of aflatoxins can lead to hepatocellular carcinoma, a better understanding of the final steps of aflatoxin biosynthesis is needed. For AFB₁ biosynthesis, ST must first be methylated by an O-methyltransferase (OmtA) unique to AF biosynthesis (Bhatnagar, et al., 1987). The resulting methylated intermediate, O-methyl-ST (OMST) (Yu, et al., 1998) is then oxidized by the cytochrome P450 monooxygenase, OrdA (AflQ). Since AFB₁ was produced when either OMST or its presumptive initial oxidation product, 11-hydroxy-OMST (HOMST), was fed to yeast cells expressing the A. parasiticus cytochrome P450 monooxygenase OrdA (Prieto, et al., 1996, Udwary, et al., 2002), it was proposed that OrdA is the only enzyme required for the conversion of OMST to AFB₁. To be consistent with the yeast-feeding experiment, OrdA must also introduce an oxygen atom into HOMST (Scheme 1). The subsequent conversion steps require hydration, ring-opening, cyclization, decarboxylation, and demethylation to produce AFB₁.

Scheme 1.

Known or previously postulated intermediates in the conversion of ST to AFB₁. Molecular weights are shown below each compound.

The oxidative ring cleavage and rearrangement necessary for the formation of the coumarin ring system in AFB₁ must be consistent with the following observations: a) NADPH is utilized in the conversion (Singh & Hsieh, 1976); b) an “NIH hydride shift” occurs so that the C-11 hydrogen is retained (Simpson, et al., 1983); c) an oxygen atom and carbon-11 in the A-ring of OMST are lost as carbon dioxide (Chatterjee & Townsend, 1994); and d) an oxygen atom incorporated into the B-ring (Scheme 1) is retained (Watanabe & Townsend, 1996).

The role of the putative aryl alcohol dehydrogenase NorA (AflE) in AF biosynthesis has not been definitively ascribed, although it was originally thought to function in the reduction of norsolorinic acid to averantin (hence the name “Nor”) (Cary, et al., 1996, Yu, et al., 2004). NorA shares more than 60% amino acid identity with NorB (AflF), an aryl alcohol dehydrogenase shown to be involved in formation of AFG₁ (Ehrlich, et al., 2008). Genes encoding both enzymes are part of the AF biosynthesis gene cluster. The ST gene cluster of A. nidulans possesses only one of these genes, stcV (Brown, et al., 1996). Based on BLAST searches of genome sequence databases, genes encoding aryl alcohol dehydrogenases are common in many filamentous fungi and yeast. In the A. flavus AF gene cluster, the promoter and translation start codon of norB are missing due to a large DNA deletion in this region (Ehrlich, et al., 2004). Therefore the role of norA is most easily examined in A. flavus. We now provide evidence that A. flavus, lacking a functional copy of norA, accumulates a new metabolite, deoxyAFB₁, a shunt metabolite that most likely is formed by dehydration of aflatoxicol (AFOH) in the acidic culture medium.

Materials and methods

Preparation of norA mutants in A. flavus AF13

A vector for insertional inactivation of norA in A. flavus was constructed by PCR with the oligonucleotide primers P1, 5’-acgactacaagaatagcggtgacat and P2, 5’-tattctagagacgcagactcttggtatgg (Genbank Accession #AY510451; 47574 to 48188) and P3, 5’-tattctagagtactgggccgcggtcagtt and P4, 5’-aatggtacctcgagtccgcgacaactaggctcattttg (48516–49107) to amplify 5’- and 3’-portions of AF13 norA, respectively (Fig. 1A). The resulting 614 and 591 bp PCR fragments were cloned into the SphI/XbaI and XbaI/KpnI sites of pUC18, respectively. An XbaI fragment from the niaD-containing plasmid, pSL82 (Chang, et al., 1996), was then inserted into the internal XbaI site of the norA fragments in pUC18 to create the knockout vector.

Fig. 1.

Disruption of norA in A. flavus AF13. (A) Schematic showing design of the knockout vector. Numbers denote the position of the region in AF13 DNA (Genbank accession number AY51051). P1 to P4 are the oligonucleotides primers used for amplification of the 5’- and 3’-ends of the gene. The neighboring genes estA and ver-1 are shown. niaD was obtained from an XbaI digest of pSL82. The construct was prepared in pUC18. (B) PCR with oligonucleotide primers P1 and P4 of AF13 DNA from different transformant clones. Lane 1, 1 kb⁺ marker (Invitrogen); Lane 2, AF13 (pSL82) control; Lanes 3–5, AF13ΔnorA-clones 3, 15, 21, respectively.

Transformation of A. flavus AF13ΔniaD protoplasts was done as previously described using the PEG procedure (Ehrlich, et al., 2004) with 10 µg of XhoI/SphI-linearized plasmid. Confirmation that norA was insertionally inactivated (double crossover event) in the resulting transformants was done by PCR using the outer oligonucleotide primers (P1 and P4, Fig 1B) with DNA from the putative transformants or from pSL82-transformed AF13 as the control.

Thin layer chromatography (TLC) and liquid chromatography/mass spectrometry (LC/MS)

Fungal cultures grown from spores at 30°C for 3 days on potato dextrose agar (PDA, Difco, Voigt Global Distribution, Lawrence, KS) were extracted with acetone and chloroform as previously described (Ehrlich, et al., 2004). Aliquots of the extract were analyzed by TLC on 250 µm silica gel plates (J.T Baker, Philipsburg NJ) developed with toluene:ethyl acetate:acetic acid (TEA) (8:1:1). A prominent blue-fluorescent compound from ΔnorA cultures was partially purified by preparative TLC. The unpurified extract, the TLC-purified metabolite, and authentic standards (AFB₁, synthetic aflatoxicol, synthetic deoxyAFB₁, OMST, and synthetic HOMST) were analyzed in the positive ion mode by LC/MS. The materials were dissolved in methanol, injected on a Luna C18 100×4.6 mm column (5 µm, 100Å, Phenomenex) equilibrated in 10% acetonitrile/0.1% formic acid and 90% aqueous formic acid (0.1%), and eluted with a gradient to 100% acetonitrile/0.1% formic acid over 30 min. Metabolites were monitored by both diode array UV-visible spectrophotometry and quadrupole MS (Agilent 6130).

Feeding studies

Aflatoxicol (AFOH) and deoxyAFB₁ were prepared by zinc borohydride reduction of AFB₁ (Sigma, St Louis, MO) (Hsia & Chu, 1977). Aflatoxicol (AFOH) was partially purified from the reaction mix by preparative TLC. Synthetic AFOH was dissolved in 200 µl dimethyl sulfoxide and added to 3-day mycelial cultures of A. parasiticus SRRC2043 (accumulates OMST only) in low sugar replacement medium (Bhatnagar, et al., 1987) or to S. cerevisiae expressing norA or ordA after induction with galactose (Yu, et al., 1998). Following 4-hr incubation, metabolites were extracted into methylene chloride and aliquots examined by TLC.

BLAST searches

BLAST searches (tBLASTX and BLASTP) were done against the sequenced fungal genome datasets in Pubmed (http://www.ncbi.nlm.nih.gov/blast/Blast.cgi), the Broad Institute fungal database (http://www.broadinstitute.org/annotation/genome/aspergillus_group/MultiHome.html) and the A. flavus genomic sequence (http://www.aspergillusflavus.org/genomics). The cut-off for matches was E -30.

Results

Transformation of A. flavus AF13ΔniaD with the linearized norA knockout vector (Fig. 1A) yielded approximately 60 colonies, three of which had slightly darker orange mycelia when re-grown on PDA plates. The three darker orange transformants were confirmed to be double crossover norA disruptants by PCR (Fig 1B). A 1.5 kb PCR band was obtained for intact norA in the AF13 control strain and an 8 kb product for the positive ΔnorA transformants (Fig 1B). The latter product is consistent with the size expected with the 7 kb niaD selection marker inserted into the norA gene.

Acetone extracts of the norA knockout cultures and cultures transformed with the selection marker only were examined by liquid chromatography combined with mass spectrometry (LC/MS; Fig. 2 and Table 1). A metabolite eluted after AFB₁ (14.1 min compared to 13.7 min) and exhibited a blue-shifted (λ_max = 332 nm) chromophore compared to that of AFB₁ (λ_max = 362 nm). This less polar compound was identified as deoxyAFB₁ by its positive ion mass spectrum (M+H = 297; deoxyAFB₁ M = 296 Da) and its having a retention time and UV-visible chromophore identical to that of deoxyAFB₁ prepared by established synthetic methods (Hsia & Chu, 1977). The LC data showed that deoxyAFB₁ accumulated in at least 20-fold greater amounts in the norA knockout strain than in the selection marker-only transformed strain (Fig. 2).

Fig. 2.

Liquid chromatography profile of the A. flavus AF13ΔnorA extract and the AF13 control extract obtained during the LC-MS analysis.

Table 1.

LC/MS data from extracts of AF13ΔnorA and control cultures

AF13ΔnorA (RTa)	control (RT)	UV λmax	m/zb	Identityc
10.31, 10.48	NDd	330	315	AFOH
10.6	10.6	318, 360she	371
10.9	10.9	360	329
12.4	12.4	318, 360sh	355	HOMST
13.0	ND	*	329
13.8	13.8	362	313	AFB1
14.1	14.1f	332	297	deoxy-AFB1
15.6	ND	448	371
15.9	15.9	312, 340sh	339	OMST

^aRT-retention time (min)

^bM+H ions

^cComparisons were to authenticated standards.

^dND-not detected.

^esh-shoulder

^fControl had 20-fold lower amount of AFB₁ by LC

^*The amount was too low for determination.

Comparison of other metabolites in the acetone extracts of an AF13ΔnorA clone (#15) and the AF13 control with natural or synthetic standards by UV-visible spectrophotometry and positive ion LC/MS confirmed the presence of OMST (15.9 min), HOMST (12.4 min, M+H = 355, M = 354), and AFB₁ (13.8 min) (Table 1). The metabolites shared identical LC retention times, UV-visible chromophores, and mass spectra with their respective standard. Several unknown compounds were also observed in extracts of fungi with both mutant and intact norA. One exhibited a chromophore (λ_max = 318 nm, shoulder at 360 nm; M+H = 371, M= 370) similar to those of OMST and HOMST, suggesting that it could be a related intermediate in the pathway. Two unknown compounds eluting at 10.9 and 13.0 min with the same mass (M+H = 329, M = 328) were found in extracts from control and norA mutant fungi. One of them eluted at 10.9 and 13.0 min, and exhibited a chromophore similar to that of AFB₁ (λ_max = 360 nm). This metabolite was found mainly in the AF13ΔnorA extract. Another compound with M+H = 371, identified only in the AF13ΔnorA extract, eluted at 15.6 min. Taken together, the observed alteration in the metabolic flux between the control and knockout transformants suggests the presence of other minor natural products and intermediates in the biosynthetic pathway to AFB₁.

An ion with the expected mass, elution time, and chromophore for AFOH (314 Da, 10.3 min) was detected in extracts of a 2-day A. flavus norA knockout culture but not in the control culture extract. AFOH, after feeding to a strain of A. parasiticus with defective ordA but intact norA, was readily oxidized to AFB₁ (Fig. 3, lane 3); deoxyAFB₁ was not detected. Similarly, AFOH was oxidized to AFB₁ by yeast cells whether or not they expressed norA or ordA (Fig. 3, lanes 7–9).

Fig. 3.

Feeding of AFOH to A. parasiticus SRRC2043 or S. cerevisiae: Lane 1, A. parasiticus SRRC2043 (accumulates OMST only) extract; Lanes 2 and 6, AFB₁ standard; Lane 3, A. parasiticus SRRC2043 incubated with partially purified AFOH (AFOH is converted to AFB₁); Lanes 4 and 10, partially purified synthetic AFOH; Lane 5, OMST standard; Lanes 7–9, AFOH fed to S. cerevisiae expressing norA, ordA, and non-transformed control, respectively. The dotted line indicates the TLC origin. Fluorescent materials at the top of the TLC plates are not shown.

Discussion

Orthologs of the aryl alcohol dehydrogenase-encoding gene norA are found in the gene clusters of all AF- and ST-producing Aspergillus species (Ehrlich, et al., 2005). The role of NorA in AF biosynthesis has not yet been defined. In previous studies, mutants of norA in A. parasiticus failed to show a detectable phenotype (Cary and Ehrlich, Chang and Ehrlich, unpublished data). Our results show that A. flavus lacking norA accumulate deoxyAFB₁. This is the first time deoxyAFB₁ has been shown to be a natural metabolite of AF-producing Aspergillus cultures. DeoxyAFB₁ most likely results from dehydration of aflatoxicol (AFOH) as had been previously demonstrated in synthetic studies and confirmed here (Lau & Chu, 1983). AFOH is a natural enzymatic reduction product of AFB₁. Therefore we suggest that A. flavus norA mutants lacking the aryl alcohol dehydrogenase accumulate an increased amount of the presumed NorA substrate AFOH, compared to cultures with intact norA, and that AFOH undergoes acid-catalyzed dehydration in the acidic growth medium to give deoxyAFB₁ (Scheme 2).

Scheme 2.

Scheme showing possible intermediates in the conversion of the putative 370 Da lactone intermediate to AFB₁ and deoxyAFB₁. Molecular weights are shown below the compound.

The presence of AFB₁ in AF13ΔnorA mutant extracts indicates that only a portion of AFB₁ is reduced to AFOH in the absence of NorA, suggesting an oxidative role for NorA that minimizes accumulation of AFOH. This provides insight into the previously reported phenomenon that aflatoxin producers and non-producers are capable of interconverting AFB₁ and AFOH (Nakazato, et al., 1990). The counterpart reductive enzymes involved in this oxidation-state balance as well as the underlying ecological rationale for the activity remain undefined. A BlastP search of the translated A. flavus genomic DNA database with the A. flavus NorA sequence revealed the presence of six genes predicted to encode proteins (AFLA_134080, E = 0; AFLA_077060, E = 0; AFLA_124600, E = -175; AFLA_096620, E = −107; AFLA_027250, E = −42; AFLA_093600, NorB, E = −44) with a high degree of homology (E value < −40). It is possible that these homologs could complement the function of NorA to some extent, even in the absence of NorB.

The LC/MS results show that the presumptive first OrdA oxidation product HOMST is present in the extracts of both AF13ΔnorA mutants and the AF13 control strain. HOMST was implicated as a potential intermediate in synthetic feeding studies with either A. parasiticus cultures or with yeast expressing ordA (Udwary, et al., 2002), and this intermediate was confirmed here in our product analysis. Our results indicate that NorA is involved in a catalytic step after OrdA oxidation and are consistent with the route proposed in Scheme 2 where OrdA is predicted to catalyze oxidation of HOMST to a putative 370 Da lactone.

The subsequent rearrangement steps of the presumptive 370 Da lactone are less clear. Ultimately, these are likely to result in formation of the 326 Da methyl enolether shown in Scheme 2, which is likely to be the immediate AFB₁ precursor. Recent results suggest that the AF biosynthesis gene, hypE, encodes a protein with an EthD domain that may be involved in the oxidative demethylation of this methyl enolether (Holmes, 2008). Proteins with an EthD domain, previously only reported in bacteria, are required for oxidative ethyl-t-butyl ether degradation in the presence of a cytochrome P450 monooxygenase (Chauvaux, et al., 2001). Disruption of hypE in A. flavus led to accumulation of a compound with the intense blue fluorescence characteristic of deoxyAFB₁ and aflatoxins, but which migrated faster than AFB₁ on TLC. This new metabolite exhibited a mass of 328 Da, which is consistent with the methyl ether shown in Scheme 2. Oxidation of the methyl ether in either the 326 Da or 328 Da intermediates may occur with HypE and an unknown cytochrome P450 enzyme (possibly OrdA or CypX (AflV)) to cause loss of the methyl as formaldehyde and directly give AFB₁ or AFOH, respectively. AFOH resulting from demethylation of the 328 Da ether would require NorA-catalyzed oxidation to AFB₁.

In the absence of NorA, the 326 Da methyl enolether or AFB₁ may be partially reduced to the 328 Da methyl ether in the reductive metabolic environment of the cell as shown in Scheme 2. As suggested previously, formation of increased quantities of deoxyAFB₁ rather than AFOH in the absence of NorA could be a consequence of the precursor metabolites being produced and isolated under acidic culture conditions. In our studies, synthetic AFOH was found to dehydrate readily under mild acidic conditions. In the fungal cell, the pH is likely to be significantly higher and therefore, if AFOH is formed, it is unlikely that it would be subjected to acid-catalyzed dehydration. The balance in the cellular environment between oxidation and reduction as well as the availability of active transport out of the cell of AFB₁ would be expected to play critical roles in determining the levels of the individual precursors and in maintaining the oxidation state of AFB₁.

Because synthetic HOMST fed to yeast expressing ordA was converted to AFB₁ [5–6], NorA is not required for formation of AFB₁ by yeast or by the norA knockout mutants (Fig. 3). Conversion of AFB₁ to AFOH was found by Nakazato, et al. when AFB₁ was fed to strains of fungi incapable of toxin production (Nakazato, et al., 1990). NorA, therefore, may serve as a maintenance alcohol dehydrogenase to prevent derailment of AFB₁ production. Our study suggests that, while conversion of OMST to AFB₁ may only require a single cytochrome P450 monooxygenase, other enzymes are important to minimize derailment of AFB₁ production.