Aflatoxins are secondary metabolites produced by Aspergillus flavus and Aspergillus parasiticus. Most of the genes involved in the biosynthesis of aflatoxin are contained within a single cluster in the genome of these filamentous fungi. Studies directed toward understanding the molecular biology of aflatoxin biosynthesis have led to a number of important discoveries. A pair of fatty acid synthase genes were identified that are involved uniquely in aflatoxin biosynthesis. Two genes were also characterized that represent new families of cytochrome P450 monooxygenases. Gene expression is coordinated during aflatoxin production and is under the control of a positive regulatory gene belonging to a family of fungal transcriptional activators associated with various metabolic pathways in fungi.
The sudden appearance of turkey-X disease in 1960, which resulted in the death of over 100 000 turkeys in the United Kingdom, led to the discovery of a family of structurally related metabolites called aflatoxins [1, 2]. Aflatoxins are secondary metabolites produced by the filamentous fungi Aspergillus flavus and Aspergillus parasiticus. These toxins are commonly found prior to harvest in oil seed crops such as corn, cotton, peanuts and tree nuts. Aflatoxins greatly impact the food and feed industry because they are highly toxic and carcinogenic in a variety of animal species . Aflatoxin B1 (AFB1) has been shown to produce G-T transversions at codon 249 of the p53 tumor suppressor gene, whose altered sequence has been associated with a number of human cancers . Due to concern for the potential effects of aflatoxins on human health, most countries have expensive monitoring programs and legislation that restrict marketing of aflatoxin-contaminated grain .
Neither A. flavus nor A. parasiticus has a sexual stage, but a parasexual cycle has been elucidated in each [6, 7]. Genetic studies using the parasexual cycle are difficult due to the similar size of haploid and diploid conidia which precludes easy and accurate determination of ploidy levels. Despite the obstacles, Papa described 24 distinct aflatoxin non-producing mutants and a number of spore color and auxotrophic mutants of A. flavus derived from the common parental line, PC-7 . He used parasexual analyses to map over 30 loci to eight linkage groups. The aflatoxin loci are all recessive in diploids, except for the dominant afl-1 locus. Of the 11 mapped aflatoxin loci, nine are in linkage group VII, one is in linkage group II, and one is in linkage group VIII. In A. parasiticus, several anthraquinone-accumulating mutants as well as spore color and auxotrophic mutants are available. Genetic studies on A. parasiticus are less extensive and the number of linkage groups reported range from six to ten .
The objective of this review is to discuss the genetic organization and function of the genes involved in aflatoxin biosynthesis. The first gene was isolated and described in 1992 [9, 10], and since the last review on the molecular biology of aflatoxin biosynthesis , rapid and significant progress followed the discovery that genes involved in aflatoxin biosynthesis in both A. flavus and A. parasiticus are clustered [12, 13]. The cluster in A. flavus is located on a 4.9 mb chromosome . Prieto et al.  have shown by complementation in a cluster-deletion mutant of A. flavus that the aflatoxin pathway genes reside in approximately 90 kb of cloned genomic DNA. Yu et al.  partially mapped the clusters in A. flavus and A. parasiticus, and they estimated the cluster size to be 75 kb. The relative position of genes within the clusters and the nucleotide sequence of the genes are similar between the two species, although A. parasiticus appears to have an additional copy of several genes indicating that at least part of the cluster is duplicated [17, 18].
Our understanding of aflatoxin biosynthesis was significantly advanced by the discovery that in A. nidulans the genes involved in sterigmatocystin biosynthesis, which is identical to that portion of the aflatoxin biosynthetic pathway, are also clustered [19, 20]. The entire nucleotide sequence of the A. nidulans cluster was recently determined, and 25 transcripts were identified whose expressions correlated with sterigmatocystin production . The arrangement of the genes within the cluster is different from the aflatoxin gene cluster, but the deduced amino acid sequence of the gene products are similar. The genetic experiments that are possible with A. nidulans should facilitate the discovery of additional information about the regulatory circuit for both sterigmatocystin and aflatoxin.
Norsolorinic acid biosynthesis
The polyketide backbone of AFB1 involves the extension of a hexanoate unit by a polyketide synthase (PKS) that assembles seven acetyl units from malonyl CoA without ketoreduction to generate noranthrone [21, 22]. The synthesis of norsolorinic acid (NA), the first stable intermediate, is completed after oxidation of the noranthrone by a proposed oxidase (Fig. 1) [11, 21]. Support for a specialized PKS and fatty acid synthase (FAS) was provided by the cloning and characterization of several A. parasiticus genes including pksA (pksL1) and fas1. Chang et al.  identified pksA when transforming an NA-accumulating strain of A. parasiticus with DNA containing fragments of the aflatoxin gene cluster . Approximately 5% of the transformants lost the ability to produce NA when the transformants were disrupted in the pksA gene. A gene, pksL1 (which is pksA), was identified among several cDNA clones isolated by differential screening . Gene disruption of pksL1 resulted in transformants that did not accumulate aflatoxins or NA suggesting the gene is involved in aflatoxin biosynthesis. The FAS gene fas1 was identified by complementation in A. parasiticus strain UVM8, which produces no aflatoxin or pathway intermediates . Additional gene disruption experiments established that fas1 functioned prior to the formation of NA and that the gene was not vital for fungal growth. A second FAS gene, fas2, has been located adjacent to fas1 (Fig. 1) and is likely the β-subunit of the functional FAS .
Genes from A. nidulans encoding PKS (stcA) and FAS alpha and beta subunits, stcJ and stcK, respectively, were localized to the sterigmatocystin gene cluster . Disruption of stcJ and stcK had no effect on fungal growth, but sterigmatocystin was not produced unless cultures were supplied with exogenous hexanoic acid. These data indicated the specific function of these genes in secondary metabolism. Hexanoic acid feeding data from mutants of both A. nidulans and A. parasiticus disrupted in the FAS genes suggest that PKS and FAS function cooperatively [26, 27].
Analyses of nucleotide sequences indicate that motifs common to known PKS and FAS are conserved in the gene products, namely β-ketoacyl synthase, acyltransferase and acyl carrier protein [12, 23–25]. The FAS genes also contain conserved domains for β-ketoacyl reductase, enoyl reductase and enoyl hydrase [12, 23, 25]. The absence of ketoreductase in the PKS sequence is consistent with a lack of a reductive step in the PKS reaction that produces noranthrone.
Conversion of NA to averatin (AVN) via a ketoreductase has been extensively studied (Fig. 1) . Isolated by complementation in a leaky mutant of A. parasiticus that accumulates NA and aflatoxins , nor1 encodes a 29 kDa protein with amino acid similarity to dehydrogenases that have NADPH-binding motifs . Disruption of nor1 in a wild-type strain resulted in transformants that accumulated NA. Similar to the original NA mutant, these nor1-disrupted transformants were not completely blocked in AFB1 biosynthesis. These data suggested the existence of other NA-reductases. In fact, a second reductase capable of converting NA to AVN in vitro was previously purified to homogeneity and shown to be a single subunit protein of about 40 kDa [21, 29]. Using monoclonal antibodies against this NA-reductase, Cary et al.  identified a cDNA clone corresponding to a gene, norA, located in the aflatoxin gene cluster of both A. parasiticus and A. flavus (Fig. 1). The predicted amino acid sequences of NOR1 and NORA share low identity (22%), but both have motifs indicating a dehydrogenase-type enzyme with adenine nucleotide-binding domains [17, 28]. Attempts to disrupt both copies of the norA in a wild-type A. parasiticus strain failed . In A. nidulans, stcE and stcV have gene products with amino acid sequences similar to NOR1 and NORA, respectively; however, confirmation of the function of these genes is lacking .
Genes involved in the conversion of AVN to versicolorin A (VER A) are the least defined of the genes in aflatoxin biosynthesis. Through the use of mutant strains, pathway inhibitors, and radioactively labelled intermediates, the major metabolites in this conversion have been determined: AVN→averufanin (AVNN)→averufin (AVF)→1-hydroxyversicolorone (HVN)→versiconal hemiacetal acetate (VHA)→versiconal (VHOH)→versicolorin B (VER B)→VER A (Fig. 1) [13, 21]. Using a cDNA clone, Yu et al.  disrupted the corresponding gene in the aflatoxin gene cluster of A. parasiticus and recovered transformants that accumulated AVN. The gene, avnA, encodes a 56.3 kDa cytochrome P450 monooxygenase, suggesting that the gene is involved in the conversion of AVN to AVNN. A similar gene, stcF, has been identified in the sterigmatocystin cluster of A. nidulans[13, 19].
Prieto et al.  transformed a mutant of A. flavus lacking the entire aflatoxin gene cluster  with two overlapping cosmids (5E6 and 8B9) and recovered transformants that accumulated AVF. The addition of a third overlapping cosmid (13B9) restored aflatoxin biosynthesis in this mutant, suggesting that a gene involved in the conversion of AVF to VHA is located on cosmid 13B9. The putative gene, avf1, was mapped to a 7 kb fragment on cosmid 13B9 (Fig. 1) . Sequence analysis indicated that this DNA fragment contains two genes, aflB and aflW (Fig. 1), which encode proteins with similar amino acid sequences to stcB and stcW, respectively, in A. nidulans but with unknown functions (, R. Prieto and C.P. Woloshuk, unpublished).
Conversion of VHOH to VER A is important in aflatoxin biosynthesis because the dihydrobisfuran ring system is formed in this step. The dihydrobisfuran is essential to the mutagenic nature of AFB1 . The double bond at the 2, 3 position in the difuran moiety is the target for renal and hepatic cytochrome P450 enzymes or lipoxygenase that generate a highly reactive epoxide [31, 32]. The formation of the dihydrobisfuran is achieved by the dehydrative cyclization enzyme known as versicolorin B synthase (VBS) followed by the oxidative desaturation of VER B to VER A. The gene encoding VBS, vbs, was isolated with PCR primers designed based on peptide fragments of the purified enzyme . The deduced vbs gene product showed significant identity to many flavin-dependent oxidases and dehydrogenases .
In A. nidulans, Kelkar et al.  recently showed that stcL codes for a cytochrome P450 monooxygenase involved in the conversion of VER B to VER A in the sterigmatocystin pathway. Further biochemical and molecular evidence suggests that the biosynthesis of aflatoxin B2 branches off the pathway at VER B [21, 34]. The structure of aflatoxin B2 is identical to AFB1 except it lacks the double bond in the difuran moiety . Recently, the gene verB (Fig. 1) was cloned which appears to be the homolog of stcL in A. flavus and A. parasiticus (D. Bhatnagar, unpublished). A mutant of A. flavus that accumulates aflatoxin B2 was found to lack verB expression.
Sterigmatocystin is formed from VER A through several enzymatic reactions, including oxidation, ketoreduction, decarboxylation, and methylation. Demethylsterigmatocystin is a known intermediate metabolite in this conversion . Genetic complementation of an A. parasiticus mutant that accumulates VER A led to the isolation ver1A[10, 35], which encodes a NADPH-dependent ketoreductase . A second gene, ver1B, was identified in A. parasiticus that has 95 percent identity to ver1A. ver1B is truncated rendering it non-functional and the gene is not located in the gene cluster . Gene-disruption of stcU, a homolog of ver1A, in a sterigmatocystin-producing strain of A. nidulans resulted in VER A accumulation . Disruption of a second A. nidulans gene, stcS, also resulted in the accumulation of VER A . StcS encodes a cytochrome P450 monooxygenase, which has been assigned to a new cytochrome P450 family, CYP59. A gene, aflS, similar to stcS has been located in the aflatoxin gene cluster (Fig. 1). A third gene, stcP, from A. nidulans most likely encodes an O-methyltransferase . Mutants with a disrupted stcP accumulated demethylsterigmatocystin in the culture medium, indicating that stcP is involved in the conversion of this intermediate to sterigmatocystin.
Sterigmatocystin to aflatoxin
Sterigmatocystin is converted to O-methylsterigmatocystin (OMST) by a 40 kDa O-methyltransferase (Fig. 1). Antiserum raised against the purified protein from A. parasiticus was used to screen a cDNA library and isolate the corresponding gene omt1. Amino acid sequence deduced from the cDNA indicated OMT1 contains conserved amino sequence for S-adenosylmethionine-binding . Extracts from E. coli expressing omt1 converted exogenously supplied sterigmatocystin to OMST requiring only the addition of S-adenosylmethionine. Both A. flavus and A. parasiticus contain single copies of omt1 that are 97 percent identical , but there is no evidence of the gene in A. nidulans.
The gene responsible for the conversion of OMST to AFB1 was identified with an A. flavus strain containing a deletion of the entire aflatoxin gene cluster [15, 39]. Transformants of this mutant containing cosmid 8B9 did not produce aflatoxin or any pathway intermediates but converted exogenously supplied OMST to AFB1 (Fig. 1) . The gene, ord1, responsible for this enzyme activity was mapped to a 3.3 kb genomic DNA fragment on cosmid 8B9 [15, 40]. The product of ord1 is a cytochrome P450 monooxygenase which was assigned to a new cytochrome P450 family named CYP64. Yeast expressing ord1 under the control of the Saccharomyces cerevisiae galactose-inducible gal1 promoter were able to convert OMST to AFB1, indicating that ord1 is sufficient to accomplish the last step of the aflatoxin biosynthetic pathway . In A. parasiticus, ordA was identified by chromosome walking, and this gene shares over 90 percent nucleotide sequence identity with ord1 (D. Bhatnagar, unpublished).
Expression of genes in the aflatoxin biosynthesis cluster is regulated by aflR. The gene was isolated from A. flavus by complementation in a mutant that did not produce aflatoxin and failed to convert exogenously supplied pathway intermediates to AFB1 [26, 41]. In A. parasiticus, aflR was discovered when investigators observed increased metabolite production in an OMST-accumulating strain transformed with a cosmid containing a large DNA fragment from the aflatoxin gene cluster . The additional copies of aflR in the transformants were apparently responsible for an increase in metabolite production. In A. nidulans, aflR was identified and has been shown to be a functional homolog of the A. flavus and A. parasiticus genes . The protein encoded by aflR contains a zinc cluster motif (Cys-Xaa2-Cys-Xaa6-Cys-Xaa6-Cys-Xaa2-Cys-Xaa6-Cys) similar to a family of fungal transcriptional activators associated with several metabolic pathways in fungi [39, 42]. The zinc cluster motif is involved in the binding of the regulatory protein to the DNA target sequence. Recombinant AFLR was shown to bind a palindrome 120 bp upstream of the aflR translation start site, suggesting the autoregulation of aflR expression . There are currently no publications describing the binding of AFLR to the promoters of the pathway genes.
The cluster arrangement of the aflatoxin biosynthetic genes is similar to a growing number of clustered genes involved in secondary metabolite and nutrient utilization pathways in filamentous fungi. A recent review indicated that there are no obvious answers to questions concerning the regulatory and evolutionary significance of these gene clusters . In A. flavus and A. parasiticus, an intact aflatoxin gene cluster is not required for aflatoxin production. Ectopic genomic integration of wild-type genes (both structural genes and AflR) into various mutants results in complementation of the mutated genes with no measurable effects on gene expression [10, 15, 41]. Further, there is little evidence to indicate that gene order within the aflatoxin cluster influences regulation. The evolutionary relatedness between the aflatoxin cluster and the sterigmatocystin cluster is not known. Although clusters contain genes with similar structure and function, the gene order within the cluster and the direction of transcription of some of the corresponding genes are different . Therefore the aflatoxin gene cluster probably does not regulate expression through an interaction with a specific chromatin structure. However, the cluster may stabilize or insure the inheritance of aflatoxin production in a fungal population.
Progress in the molecular biology of aflatoxin biosynthesis should continue at its current pace. Various laboratories are generating gene-knockout mutants of the remaining genes in the aflatoxin cluster. The phenotypes of these mutants should give insight into the functions of the respective genes. One intriguing gene that has proven difficult to define its function is aflJ (Fig. 1) . An aflJ disruption mutant of A. flavus does not produce aflatoxin or convert pathway intermediates to aflatoxin. Analysis of the nucleotide sequence did not reveal enzymatic domains similar to any known gene (G.A. Payne, unpublished).
New areas of research now focus on the genes that reside outside of the gene cluster. Studies of four A. flavus strains from Papa's collection  with mutated loci in linkage group VII but not mapped to the gene cluster suggest that the mutations in these strains affect the regulation of aflR (M. Weaver and C.P. Woloshuk, unpublished). Another of Papa's A. flavus strains with a mutation on linkage group II also appears to affect gene expression prior to aflR (M. Brown and G.A. Payne, unpublished). The isolation of these genes by genetic complementation should lead to new information.
Characterization of the genes involved in sterigmatocystin biosynthesis in A. nidulans with its classical genetic system will contribute significantly to our understanding of aflatoxin/sterigmatocystin biosynthesis. Several mutations have led to the identification of a set of regulatory genes involved in both fungal development and toxin production (T. Adams, unpublished). These studies should soon reveal information about the signal transduction pathways controlling aflatoxin/sterimatocystin biosynthesis.
Little is known about the gene(s) involved in the biosynthesis of the aflatoxin G-family, which are only produced by A. parasiticus. Biochemical evidence suggests that aflatoxin G1 is produced by a separate pathway branching from sterigmatocystin [21, 46]. Currently there is no evidence indicating that the genes responsible for the branched pathway are within the gene cluster of A. parasiticus. However, the genes at the ends of the cluster have not been thoroughly characterized.
Finally, nomenclature for the aflatoxin biosynthetic genes should be examined to achieve consistency in the criteria for gene symbols. Recognized as a problem by Bennett  before the first gene was isolated, no standard convention has been used for naming the genes from A. flavus and A. parasiticus. As a result of the inconsistency in gene symbols and the frequent changes made in the gene symbols, we have created confusion even among those who work in the field. In this review, we have referred only to the current symbols for aflatoxin genes. The molecular biology of aflatoxin biosynthesis has reached a point that the current symbols should be changed to comply with the convention used in A. nidulans.
We thank Deepak Bhatnagar and Larry Dunkle for reviewing this manuscript. We also thank T. Adams, D. Bhatnagar, J. Linz, N. Keller, and G. Payne for sharing their unpublished information. This is Journal Publication No. 15396 of the Purdue University Agricultural Research Station.