OUP user menu

The aflatoxin pathway regulator AflR induces gene transcription inside and outside of the aflatoxin biosynthetic cluster

Michael S. Price , Jiujiang Yu , William C. Nierman , H. Stanley Kim , Bethan Pritchard , Carrie A. Jacobus , Deepak Bhatnagar , Thomas E. Cleveland , Gary A. Payne
DOI: http://dx.doi.org/10.1111/j.1574-6968.2005.00084.x 275-279 First published online: 1 February 2006

Abstract

Aflatoxin contamination of food and feed is a major concern due to the carcinogenic properties of this mycotoxin. Previous studies using classical approaches have identified a cluster of genes responsible for aflatoxin production under the control of the pathway-specific transcriptional regulator aflR, but it is unknown whether aflR controls expression of other genes within the genome. Transcription profiling comparing wild type and ΔaflR strains of Aspergillus parasiticus grown under conditions conducive for aflatoxin production identified only 23 upregulated genes in the wild type. These included 20 genes in the aflatoxin biosynthetic cluster, and three additional genes outside the aflatoxin biosynthetic cluster (nadA, hlyC, and niiA), all with AflR binding sites. This report is the first to demonstrate genes outside the biosynthetic cluster as being associated with aflR expression.

Keywords
  • transcription profiling
  • microarray analysis
  • aflatoxin
  • secondary metabolism
  • Aspergillus flavus
  • Aspergillus parasiticus

Introduction

The biosynthesis of the potent carcinogen, aflatoxin (AF), has been studied extensively and molecular analysis has identified a biosynthetic gene cluster (Yu et al., 2004a) regulated by the transcription factor AflR (Yu et al., 1996; Flaherty & Payne, 1997). AflR is transcriptionally and post-transcriptionally regulated (Shimizu et al., 2003), and is believed to regulate itself (Ehrlich et al., 1999). AflR is absolutely required for AF biosynthesis. It is hypothesized that AflR regulates only genes in the biosynthetic cluster, but until now it has not been possible to test this hypothesis.

In this study we sought to identify genes regulated by AflR via gene expression comparisons between Aspergillus parasiticus SU1 and a ΔaflR mutant created from the same A. parasiticus strain. A cDNA microarray representing c. 40% of the A. flavus transcriptome was hybridized with cDNA made from RNA of both A. parasiticus strains grown under conditions conducive for AF production. This comparison is possible because A. flavus and A. parasiticus share nearly identical sequence and conserved gene order in the AF biosynthetic cluster.

Materials and methods

Strains and culture conditions

Aspergillus parasiticus (Speare) strain SU1 (ATCC 56775) and a ΔaflR derivative of SU1 (courtesy of Dr Jeff Cary, USDA-ARS, SRRC) were grown for biomass production in media conducive for AF production as described previously (Price et al., 2005). Cultures to be harvested for RNA extraction were incubated at 28°C, 200 r.p.m. for 8, 16, or 24 h. Five cultures were inoculated per strain for the 8 h timepoint, two cultures per strain for the 16 h timepoint, and one culture per strain for the 24 h timepoint in order to procure sufficient tissue for RNA extraction. Two biological replications of the experiment were performed, with the resulting data combined and analyzed.

RNA preparation and microarray analysis

Tissue handling and RNA preparation were performed as previously described (Price et al., 2005), with the following modifications: after isolation of total RNA with Trizol (Invitrogen Life Technologies, Carlsbad, CA), the RNA was precipitated overnight in 2 M LiCl. After centrifugation, the RNA was washed once with 75% ethanol, centrifuged, and allowed to air dry for 20 min. The RNA pellet was resuspended in 50 μL DEPC-dH2O with 40 units RNasin™ RNase inhibitor (Promega Corporation, Madison, WI) and quantified by spectrophotometry. RNA preparations were visualized by gel electrophoresis to ensure quality.

The microarrays used in this study were printed at The Institute for Genome Research (TIGR) with amplicons (c. 530 bp) generated from genomic DNA using primers selected from EST sequences (Yu et al., 2004c). A total of 5002 genes, including 31 known AF and sugar utilization cluster genes, were arrayed in three duplications (or more) for a total of 17 991 spots. Total A. parasiticus RNA from each treatment studied was converted to cDNA and labeled as previously described (Price et al., 2005).

Comparisons between treatments were made using a loop design (Fig. 1). This design allowed in silico comparisons between each node in the loop (Churchill, 2002). Furthermore, each treatment was labeled with each dye, removing effects on measurements caused by the individual dyes. The hybridized slides were scanned using a Perkin Elmer ScanArray Express Lite scanner (Perkin Elmer Life and Analytical Sciences Inc., Boston, MA). Spot intensity data were extracted from the images using UCSF-Spot (Jain et al., 2002). The resulting spot-intensity data were then analyzed statistically using the mixed procedure in SAS (SAS v8, SAS Institute, Cary, NC) as described previously (Price et al., 2005). Briefly, least squares estimates of gene-specific treatment effects between pairs of treatments were obtained for each gene under each treatment condition. Differences between treatment effects (least-squares estimates) for pairs of conducive and nonconducive conditions can be considered as log2-transformed fold changes (Wolfinger et al., 2001). Genes were considered significantly up- or downregulated in the individual microarray experiments if they possessed P-values less than (0.001). The value P<0.001 was chosen based on the maximum P-value exhibited by the AF pathway genes included on the array.

1

Loop design for microarray analysis of the ΔaflR strain vs. wild type. Comparisons were made between strains at each timepoint. RNA samples from each strain/timepoint combination were isolated and cDNAs were labeled with Cy3 (solid dot) or Cy5 (arrowhead) and hybridized to the microarray. In this design, each sample is labeled once with each dye to correct for bias in dye incorporation. The construction of this comparison loop allows for in silico comparisons between any two treatments.

Quantitative PCR confirmation of microarray results

Real time quantitative PCR analysis was used to support the gene expression data obtained from the microarray experiments. Primers were designed to amplify 65–72 bp amplicons of genes that were differentially expressed at 24 h: aflD (5′-GCTGCAGCAGTCCAAGCAA-3′ and 5′-CATGTTGGTGATGGTGCTGATC-3′), aflM (5′-GTGGGCCTCCCTGTGGAT-3′ and 5′-CACTTACCCATTCGGCTGTGT-3′), hypB (5′-CGACAAGTTGACCCGACTGA-3′ and 5′-GAGCGTCTATGGGCTTGCA-3′), aflO (5′-GCTGGGATGATCTGCTTCAAG-3′ and 5′-ATTTGCGTCATGTCTTCCATGA-3′), aflP (5′-TGAAGCGCTGCGAATCCT-3′ and 5′-AGCAAGTCGCGCATCCTT-3′), nadA (5′CTCCCAGGACGCGGTAGAT-3′ and 5′-ACGGTCAATCGCCTTTCG-3′), niiA (5′-GCGTAATTTCGAGCTCAATG-3′ and 5′-AAGTTGCGATATTCATAGCCTCATC-3′), and hlyC (5′-TGACTGTTTGCTCGGAGAATGA-3′ and 5′-GCTAGCCATCTGTTCACGATAGC-3′). The 2-ΔΔCT method was used to analyze the real time quantitative PCR data (Livak & Schmittgen, 2001). For each strain, 18S control values were subtracted from raw CT values to normalize the data. Normalized CT values for each gene in the aflR deletion strain were subtracted from the corresponding wild type normalized CT values to obtain a log2 value. The fold change was calculating by raising 2 to the power represented by the log2 value for each gene.

Results and discussion

AF production in ΔaflR strain and SU1

Aflatoxin was detected in SU1 cultures at all three timepoints, reaching a concentration of 2010 ng mL−1 at 24 h. (Fig. 2). The time course for AF production was similar to that observed in previous studies (Skory et al., 1993). As predicted, no AF was detected in cultures of the ΔaflR mutant at any of the three time points examined.

2

Aflatoxin production in cultures of Aspergillus parasiticus SU1 and ΔaflR. Cultures of A. parasiticus SU1 and ΔaflR mutant strains were grown for 16 h at 28°C for mycelia production. In all, 20 mL aliquots of these cultures were then inoculated into daughter cultures and incubated for 8, 16 and 24 h at 28°C, 200 r.p.m. for the production of tissue and aflatoxin. RNA samples were prepared from these tissues for microarray analysis.

Transcription profiling of SU1 and ΔaflR strains

An analysis of gene transcription at each of the three time points using the microarrays identified only 23 genes more highly expressed in SU1 than the ΔaflR mutant at every timepoint examined (Table 1, Fig. 3). The trends in expression for eight of these genes were confirmed by quantitative PCR (Table 2). Eighteen of the genes differentially expressed on the microarrays are reported AF biosynthetic genes, three of these genes (hypB, aflY, and nadA) are in or adjacent to the established AF cluster (Yu et al., 2004a, 2004b), and the last two genes (hlyC and niiA) are located outside the AF biosynthetic cluster. All of the AF biosynthetic genes with a putative consensus AflR binding site (5′-TCGSWNNSCGR-3′) in their promoters (Ehrlich et al., 1999), except for aflR, were upregulated in SU1 (Table 1). In addition, aflY, which was listed in the cluster by (Yu et al., 2004b) but not assigned a function, was shown to be upregulated in SU1 and to have a consensus AflR binding site. Recently, Ehrlich and colleagues reported that aflY is involved in the conversion of versicolorin A to sterigmatocystin (Ehrlich et al., 2005). Statistical evidence supports the functionality of the consensus AflR binding sites in these genes. In all, 3647 putative AflR binding sites are identified in the A. flavus genome sequence (unpublished data). This number is similar to the 3398 putative NirA binding sites (the nitrogen assimilation pathway-specific regulator), but more than the number of AflR binding sites that would be expected by chance in the A. flavus genome (2211, P=0.00006).

View this table:
1

ESTs upregulated in Aspergillus parasiticus SU1 at all timepoints

3

Relative expression levels for genes upregulated in Aspergillus parasiticus SU1. Least-squares means estimates of expression obtained from the mixed-model analysis of the microarray data (see Materials and methods for explanation and for relationship to fold changes) were plotted for genes upregulated in A. parasiticus SU1.

View this table:
2

qPCR confirmation of microarray results

Four of the genes that have not been previously shown to be involved in AF biosynthesis were upregulated in SU1 (hypB, nadA, niiA, and hlyC) and possessed consensus AflR binding sites upstream of their putative start codons (Table 1). One of these genes (nadA) was previously identified (Yu et al., 2000) as part of the sugar utilization cluster neighboring the AF biosynthetic cluster. Yu (2000) stated that the genes of the sugar utilization cluster in A. parasiticus did not possess consensus AlfR binding sites near the translational start sites (100–300 bp), but were likely related to AF production due to their location neighboring the AF biosynthetic cluster. The AflR binding site in the intergenic region of the divergently transcribed nadA and aflY genes (752 bp to nadA and 124 bp to aflY start codons) may be shared by both genes. The reduced expression of nadA in the ΔaflR strain and the similarity of its expression to other pathway genes in SU1 in these experiments (Fig. 3) suggest that nadA may in fact belong to the AF biosynthetic cluster. A role for nadA in AF biosynthesis is not known. It is possible that nadA, encoding a putative NADH oxidase, increases reducing power in the cell, and thus increases the available NADH required energy conservation in the cell. A previously undiscovered gene, hypB (for second hypothetical protein in AF cluster), was shown to be expressed and upregulated. hypB was located in the AF biosynthetic cluster between aflI and aflL (data not shown). This gene has two putative AflR binding sites located 100 bp and 1.3 kb upstream of the putative coding region.

Two genes were not located near the AF biosynthetic cluster. One of these genes, nitrite reductase (niiA), is located in the nitrate assimilation cluster, and is divergently transcribed from the same promoter as niaD, the gene encoding nitrate reductase. No putative AflR binding sites were observed in the niiA 5′-untranslated region. Most of the functional AflR binding sites reported reside within 100–300 bp of the transcriptional start site (Ehrlich et al., 1999), although functional AflR binding sites beyond 300 bp have been observed (Fernandes et al., 1998). Interestingly, niiA possessed a consensus AflR binding site approximately 2.3 kb upstream, within the coding region of niaD (data not shown). The prospects for aflR control of nitrite reductase expression, in light of nitrate repression of AF production, compel further investigation of this putative interaction. The hlyC gene, encoding a homolog of α-hemolysin from Aeromonas hydrophila, was located approximately 1.5 Mb from the AF cluster according to A. flavus genome data (data not shown) and has a putative AflR binding site approximately 1.8 kb upstream of the putative coding region. hlyC has no apparent role in AF production but may play a role in animal pathogenesis by Aspergilli.

In conclusion, our data support previous data showing that aflR is required for expression of the AF biosynthetic genes under conditions conducive for AF production (Price et al., 2005). They also support the inclusion of nadA into the AF biosynthetic cluster, but confirmation of its role in AF biosynthesis must await functional analysis of this gene. Furthermore, these data argue that AflR regulates genes outside the cluster. The number of such genes is likely small, as only two were identified using an array that contains approximately 40% of the transcriptome. It will be interesting to learn if these genes are in some way involved in AF production. It is tempting to speculate that both nadA and niiA may be involved in metabolic regulation of the pathway by influencing reducing power and nitrogen metabolism, respectively.

Because nadA and niiA contained the consensus AflR binding site upstream of their coding regions, it may be fruitful to examine the genome (once it is available) for additional genes with the binding site. Clearly the AflR binding site is not required for the activation of all AF biosynthetic genes as several without the consensus site were upregulated in the wildtype strain (Ehrlich et al., 1999). Finally, this study shows that DNA microarrays can be a powerful tool to identify genes regulated by transcriptional factors and points to the need for whole genome arrays for Aspergillus flavus.

Acknowledgements

The authors wish to thank Greg O'Brian, Rob Holmes, Sheri Denslow, and Ryan Georgianna for their critique of this manuscript. This research was supported by Grant 2002-35201-12562 from the USDA/NRI Competitive Grants Program.

References

View Abstract