Introduction

Doenjang, a Korean traditional fermented soybean paste, is made using meju, in which fermentation is mainly dependent on natural inoculation of various microorganisms such as bacteria, yeasts, and molds [17,21,24]. In meju, Aspergillus spp. have been the primary molds used for soybean fermentation. Supporting this, A. oryzae was dominantly found in various types of meju, regardless of regional origins and fermentation time [14].

Koji using a starter has been developing for industrial production of doenjang because it is easier for mass production with uniform quality [27]. A. oryzae has been included in most starters, and other molds are also additionally used to take advantage of their beneficial characteristics. For example, Aspergillus sojae has been used because of its high activities of protease and amylase [16], and Aspergillus usami and Rhizopus sp. are included to speed up the maturation process [1]. Most molds used as starters for doenjang fermentation are based on isolates that originate from meju.

Because of the long history of doenjang consumption, molds involved in fermentation get little attention with regard to safety issues. However, the World Health Organization (WHO) and United Nations Food and Agriculture Organization (UN FAO) recommend continuous monitoring of mycotoxins because they could cause various diseases [13,25,28]. According to some reports, the aflatoxigenic Aspergillus flavus was isolated from Korean traditional fermentation products [20]. The aflatoxigenic A. flavus comprised 2.2% of the molds isolated from home-made meju [12].

In many cases, A. oryzae-like molds are used without thorough evaluation of safety after simple identification, and sometimes even without identification. However, because fungi belonging to Aspergillus section Flavi are so phylogenetically close that they may sometimes be misidentified, it is possible that molds producing mycotoxins may accidentally be used.

A. oryzae is generally regarded as safe (GRAS); however, A. flavus, which is very closely related to A. oryzae, is a producer of aflatoxins (AFs). AFs are known to be the strongest naturally found carcinogens. There are four main types of AFs-AFB1: AFB2, AFG1, and AFG2. AFB1 is more prevalent than other AFs in contaminated food grains and has been considered to be the most mutagenic and carcinogenic [6]. Even the identification of a strain as A. oryzae does not guarantee its inability to produce AFs or other toxic metabolites. Some A. oryzae produce cyclopiazonic acid (CPA), which is an indole-tetramic acid mycotoxin that could be a health threat to humans [26] . Although CPA has not been considered to be a serious mycotoxin and its contamination has not been regulated for food safety, there have been reports on the toxicity of CPA. CPA co-contamination with AF is considered to be a cause of “X-disease,” which has killed ducks [32]. CPA also causes “kodua poisoning,” which is accompanied by symptoms of dizziness and vomiting [29]. Therefore, the toxigenic potential must be specifically determined for individual strains of A. oryzae.

Although A. oryzae has mainly been used for a starter in doenjang koji, thorough evaluation of its mycotoxin production is required, because it may produce CPA and the strain used may not be A. oryzae because of its very close relationship to aflatoxigenic A. flavus. In this study, we isolated and identified molds from industrial doenjang koji and evaluated their AF and CPA production.

Materials and Methods

Chemicals and Media

AF mix (B1, B2, G1 and G2), CPA, and Ehrlich solution were purchased from Sigma Aldrich Chemical Co. (St. Louis, MO, USA). All HPLC-grade solvents were from Samchun Pure Chemical Co. (Pyeongtaek, Korea). Potato dextrose agar (PDA), malt extract agar (MEA), dichloran rose bengal chloroamphenicol agar (DRBC), and dichloran 18% glycerol agar (DG18) were from Acumedia (Baltimore, MD, USA). Doenjang koji used for manufacturing SNU doenjang was provided by Prof. S. H. Choi (Seoul National University, Seoul, Korea).

Fungal Strains and Culture Conditions

A. flavus NRRL 3357 (AF+) and A. oryzae SRRC 266 (AF-) obtained from Prof. Nancy P. Keller (University of Wisconsin, Madison, WI, USA) were used as reference strains to identify the unknown strains and for AF production analysis. As for reference strains for CPA production analysis, A. oryzae NBRC 4177 (CPA+) and A. oryzae RIB 40 (CPA-) obtained from KACC (Korean Agricultural Culture Collection, Suwon, Korea) were used.

Spores of strains were harvested using a 0.1% Tween 80 buffer after culturing the strains on PDA plates for 3 days at 30℃.

Isolation of Filamentous Fungi from Doenjang Koji

To isolate residing filamentous fungi from doenjang koji, a sample was diluted from 10-1 to 10-7 using 0.1% peptone in water. Diluted samples were inoculated onto PDA, MEA, DRBC, and DG18 and incubated at 30℃ for 5 days. Colonies with different morphological characteristics were pure-cultured and further analyzed.

Morphological Examination

The isolates and two reference strains, A. flavus NRRL 3357 and A. oryzae SRRC 266, were grown for 5 days as 1-point inoculations (1.0 × 104 spores in 10 μl) on PDA. The morphology and color of colonies were noted. For conidial head and mycelia observation, spores were inoculated onto a PDA agar block (1 cm2) on a glass slide, cultured for 5 days, and then observed using a stereoscope (Olympus, Tokyo, Japan). The size of the conidia and the shape of the mycelia were observed microscopically. For sclerotia formation, spores (1.0 × 104 spores in 10 μl) were 1-point inoculated onto Czapek Dox medium (sucrose 3%, NaNO3 0.3%, MgSO4·7H2O 0.05%, KCl 0.05%, K3HPO4 0.1%, agar 1.5%, pH 6.2) and cultured at 30℃ for 15 days.

Analysis of the ITS Region, and of the AF and CPA Gene Clusters

Fungal strains were cultured in PDB at 30℃ and genomic DNA was isolated as previously described [31]. Fragments containing the internal transcribed spacer (ITS) region were amplified using the primers ITS1 and ITS4 [35]. The amplification program consisted of pre-denaturation at 94℃ for 2 min; 30 cycles at 94℃ for 30 sec, 48℃ for 30 sec, 72℃ for 30 sec; and a final incubation at 72℃ for 5 min to complete the final extension. PCR was performed using a C1000 Thermal Cycler (Bio-Rad, Hercules, CA, USA). Sequence analysis was performed at Macrogen Inc. (Seoul, Korea). To identify the isolates, the sequences were analyzed using the NCBI BLAST program (National Center for Biotechnology Information, Bethesda, MD, USA).

The restriction enzyme fragment polymorphism analysis of genomic DNA was performed to differentiate A. oryzae and A. flavus [19]. About 5 μg of genomic DNA was digested using SmaI and analyzed by agarose gel electrophoresis, according to standard protocols [30].

The five AF biosynthetic genes (norB-cypA, aflR, omtA, nor-1,and ver-1) and three CPA biosynthetic genes (moaA, dmaT, and pks-nrps) were analyzed by PCR using the primers listed in Table 1. The β-tubulin gene was used as a template control. The same PCR conditions applied for the ITS region were used, except for the time for extension (30 sec for AF genes and 1 min 30 sec for CPA genes) and the annealing temperature (55℃ for norB-cypA, 68℃ for aflR, 69℃ for omtA, 69℃ for nor-1, 68℃ for ver-1, 55℃ for moaA, 56℃ for dmaT, 56℃ for pks-nrps, and 55℃ for β-tubulin).

Table 1.PCR primers used in this study.

Analysis of AF Production

For the rapid evaluation for AFB1 production of the isolates, the ammonium vapor method (AVM) was carried out as previously described [15]. Ten microliters of spores (1 × 104 spores) was inoculated at the center of the PDA and yeast extract sucrose agar medium (YES; 2% yeast extract, 15% sucrose, 1.5% agar) at 30℃ for 3 days. The cultured strains were exposed to a drop of 25% ammonium hydroxide in the lid of an inverted petri dish. In each colony, color change to red was noted.

For the evaluation of AF production by TLC and HPLC, 0.5 ml of spores (1.0 × 107/ml) was inoculated in 50 ml of PDB and incubated at 30℃ with shaking at 200 rpm for 5 days. The cultures were filtered with Whatman No. 1 filter paper (Whatman PLC, Buckinghamshire, UK). The extraction and TLC analysis of AF were performed as described previously [15]. The filtrate was mixed with an equal volume of chloroform and vortexed for 20 min. The chloroform phase was separated from the aqua phase by centrifugation at 1,750 ×g for 3 min. Chloroform in the AF extract was removed using a rotary evaporator (EYELA, Tokyo, Japan) in a 40℃ water bath, and the samples were redissolved in 1 ml of chloroform. Five microliters of each sample and AF standard (B1, B2, G1, and G2; Sigma, St. Louis, MO, USA) was spotted onto a Silica 60 F254 plate (Merck, Darmstadt, Germany) and the plate was developed using acetone:chloroform (15:85). The developed plate was dried, and the spots were visualized under UV light (365 nm).

The AF extracts were filtered using a 0.22 PTFE membrane filter (Target, National Scientific Co., Rockwood, TN, USA), and 20 μl was injected into the HPLC system (Agilent Technologies, Wilmington, DE, USA) equipped with a Hypersil Gold column (C18, 4.6 mm × 150 mm, 5 μm; Thermo, San Jose, CA, USA). The mobile phase was acetonitrile:methanol:water (10:40:50), pumped at a constant flow rate of 0.5 ml/min for 30 min.

Analysis of CPA Production

For the TLC and HPLC analyses of CPA production, 0.5 ml of spores (1.0 × 107/ml) was inoculated in 50 ml of CAM media (15% sucrose, 2% yeast extract, 1% peptone, pH 6.0 [26]) and incubated at 30℃ with shaking at 200 rpm for 5 days. The culture supernatant prepared by the same method for AF extraction was mixed with an equal amount of chloroform. The CPA extraction and sample preparation for TLC and HPLC were the same as that for AF analysis. A sample was reconstituted in 1 ml of methanol and 10 μl was spotted onto a Silica 60 F254 plate (Merck, Germany) and developed using ethyl acetate/methanol/ammonium hydroxide (85:15:10). The developed plate was dried, and a CPA spot was visualized under UV light (254 nm) after staining with Ehrlich’s solution [4].

The extraction samples were filtered using a 0.22 PTFE membrane filter (Target, National Scientific Co., Rockwood, TN, USA) and 20 μl was injected into the HPLC system equipped with a Hypersil Gold column (C18, 4.6 mm × 150 mm, 5 μm; Thermo, USA). The mobile phase was acetonitrile: 0.1% trifluoroacetic acid water (50:50 (v/v)), pumped at a constant flow rate of 1 ml/min. CPA was detected using a UV detector (254 nm).

Results and Discussion

The safety of a starter for industrial doenjang fermentation is a very important issue. Many starters have been developed based on molds isolated from soybean fermentation products without safety evaluation. Although A. oryzae is the mold most commonly used for a starter and regarded as safe, care should be taken as it is hard to distinguish from A. flavus, a well-known carcinogenic AF producer. In addition, some A. oryzae are known to produce an indole-tetramic acid mycotoxin, CPA.

Isolation of Fungi from Industrial Doenjang Koji

SNU doenjang has a good reputation because of its high quality. To evaluate fungal flora involved in SNU doenjang fermentation, fungi were isolated from SNU doenjang koji by a plating method using PDA, MEA, DRBC, and DG18 media. DRBC and DG18 media were used to ensure that no slow-growing and/or xerophilic fungi were missed. The dilution of a doenjang koji sample at 10-6 yielded about 100 colonies after a 3-day culture in all the media used. Only two types of fungi appeared: one formed greenish-colored colonies (SNU-G) whereas the other formed white colonies (SNU-W) (Fig. 1A). About 90% of the colonies were SNU-G and about 10% of them were SNU-W.

Fig. 1.Morphological characteristics of A. flavus NRRL 3357, A. oryzae SRRC 266, SNU-G, and SNU-W. (A) Colony morphology in PDA medium. (B) Aerial hyphae and conidia head examined using a stereoscopic microscope. (C) Conidia head on slide culture. (D) Conidia size: 2-3 μm diameter for NRRL 3357; 3-5 μm for SRRC 266; 3-5 μm for SNU-G; and 3-5 μm for SNU-W. (E) Sclerotia formation. NRRL 3357 only formed sclerotia in PDA medium.

Identification of SNU-G and SNU-W

To identify the two isolates, their morphological characteristics were examined and compared with known A. oryzae SRRC 266 and A. flavus NRRL 3357. SNU-G formed floccose colonies (similar to A. oryzae SRRC), which is the typical colony characteristic of A. oryzae [7,11]. Supporting the floccose colonies were obvious aerial hyphae in SNU-G (Fig. 1B). The colony color of SNU-G was yellow-green, while that of SNU-W was white during a 2-3-day incubation, changing gradually to an olive color (Fig. 1A). The spore size of SNU-G and SNU-W was 1.5 times bigger (3-5 μm) than that of A. flavus (2-3 μm) (Fig. 1D), and the spore size is known to be bigger in A. oryzae than in A. flavus [7]. SNU-G and SNU-W did not form sclerotia, which are very common in A. flavus but rare in A. oryzae (Fig. 1E). Altogether, SNU-G and SNU-W had more similar morphological characteristics to A. oryzae than to A. flavus.

The identification of Aspergillus at the genus level is carried out using the ITS regions [9]. The ITS regions of SNU-G and SNU-W were amplified by PCR using ITS1F and ITS4R primers, both of which yielded 578 bp fragments. Sequence analysis revealed that SNU-G and SNU-W showed 97% and 95% sequence similarity to A. oryzae/flavus, respectively. However, analysis of the ITS sequence did not differentiate whether SNU-G and SNU-W were A. flavus or A. oryzae. The method to distinguish A. oryzae from A. flavus has been developed depending on the SmaI digestion pattern of genomic DNA [8,19]. The 3.8 kbp fragments were only observed in A. flavus NRRL 3357, whereas both 2.7 kbp and 1.0 kbp fragments were in SNU-G, SNU-W, and A. oryzae SRRC 266, as expected for A. oryzae (Fig. 2). Therefore, both SNU-W and SNU-W were identified as A. oryzae based on the morphological characteristics and the SmaI digestion pattern of genomic DNA. Supporting this identification, SNU-G and SNU-W had the similar AF gene cluster to A. oryzae and did not produce AF (see the following section). Therefore, SNU-G and SNU-W were designated as A. oryzae SNU-G and A. oryzae SNU-W, respectively.

Fig. 2.SmaI digestion pattern of genomic DNA. Lanes: M, molecular size markers (1 kbp ladder); 1, NRRL 3357 (3.8 kbp band pattern); 2, SRRC 266 (2.7 kbp and 1 kbp band patterns); 3, SNU-G; 4, SNU-W.

Evaluation of AF Production in A. oryzae SNU-G and A. oryzae SNU-W

Although both SNU-G and SNU-W were identified to be A. oryzae, their aflatoxigenicity was evaluated to ensure that they are non-AF producers. Presently, various methods are used to distinguish aflatoxingenic from non-aflatoxigenic strains. The most common methods are PCR of the AF biosynthetic genes and AF detection by TLC and HPLC [10,15,22].

A. oryzae is very closely related to A. flavus, a notorious AF producer in various contaminated foods. Most A. oryzae strains possess the AF biosynthesis gene cluster as in A. flavus, but do not produce AF because of deletions or mutations in the gene cluster [10,18]. The AF gene cluster is composed of 25 genes, among which the deletion or mutation has been reported in the norB-cypA, aflR, omtA, nor-1, and ver-1 genes. When evaluated by PCR, the expected sizes of fragments from the aflR, omtA, nor-1, and ver-1 genes were amplified in A. flavus NRRL 3557, A. oryzae SRRC 266, SNU-G, and SNU-W. However, for the norB-cypA gene, the type I-deleted form of the norB-cypA gene (400 bp) was detected in both SNU-G and SNU-W as well as in A. oryzae SRRC 266; however, it was not detected in A. flavus NRRL 3557, in which the type II-deleted form (800 bp) was amplified [2] (Fig. 3B). Supporting this analysis, AF was not detected by AVM, TLC, and HPLC (Fig. 4). AVM is a quick test for AFB1 production that is based on versicolorin, a precursor of AFB1, which can turn to red under alkaline conditions [17].

Fig. 3.PCR analysis of the AF and CPA gene clusters. (A) AF and CPA gene clusters in A. flavus NRRL 3357, A. oryzae NBRC 4177, and A. oryzae RIB 40. Solid and dotted arrows indicate the target AF and CPA genes for PCR amplification, respectively. (B) PCR amplification of AF genes. 1, NRRL 3357; 2, SRRC 266; 3, SNU-G; 4, SNU-W. Amplicon sizes were 400 bp (I) and 800 bp (II) for norB-cypA, 400 bp for nor-1, 1,032 bp for aflR, 537 bp for ver-1, and 797 bp for omtA. (C) PCR amplification of the CPA genes. 1, NBRC 4177; 2, RIB 40; 3, SNU-G; 4, SNU-W. Amplicon sizes were 1,205 bp for maoA, 765 bp for dmaT, and 986 bp for pks-nrps.

Fig. 4.Evaluation of AF production. (A) AVM. (B) TLC. (C) HPLC. AF: AF standard; B1: AFB1; B2: AFB2; G1: AFG1; G2: AFG2; NRRL 3357: A. flavus NRRL 3357; SRRC 266: A. oryzae SRRC 266; SNU-G; and SNU-W. AFB1 was only detected in A. flavus, with a retention time of 8.3 min.

Evaluation of CPA Production in A. oryzae SNU-G and A. oryzae SNU-W

Some A. oryzae strains are known to produce CPA. Although CPA has not been considered as serious as AF, its toxicity is clearly documented. For example, “kodua poisoning,” which is accompanied by symptoms of dizziness and vomiting, is caused by CPA [29]. To evaluate CPA production of A. oryzae SNU-G and SNU-W, the CPA biosynthetic gene cluster was analyzed. The CPA biosynthetic gene cluster is located next to the AF gene cluster [4]. At present, three genes (maoA, dmaT, and pks-nrps) are known to be important for CPA biosynthesis. Some A. oryzae, such as A. oryzae RIB 40, do not produce CPA because of deletion in pks-nrps [4,23,32,33]. In the PCR analysis of the maoA, dmaT, and pks-nrps genes, A. oryzae SNU-G and SNU-W showed the amplification pattern of A. oryzae NBRC 4177, which is a CPA producer (Fig. 3C). However, CPA was not detected in the culture extracts of A. oryzae SNU-G and SNU-W as in A. oryzae RIB 40, which is a non-CPA producer (Fig. 5). The reasons why A. oryzae SNU-G and SNU-W do not produce CPA despite the presence of all three CPA biosynthetic genes remain unclear. One possibility is that we may not detect CPA because production levels are too low to detect at our conditions. Another possibility involves some deletions or mutations in the regulatory elements for CPA biosynthesis. Currently, there is no report on the regulatory genes in CPA production.

Fig. 5.Evaluation of CPA production. (A) TLC. (B) HPLC. CPA: CPA standard; NBRC 4177: A. oryzae NBRC 4177; RIB 40: A. oryzae RIB 40; SNU-G; and SNU-W.

Because of a long history of consumption, doenjang production often excludes the issue of food safety. However, mycotoxins have been detected, suggesting that its safety should not be taken for granted. In particular, starters that are increasingly used in the industrial production of fermented foods should be evaluated for safety, because the misuse of starters could cause serious problems for consumer health. In this study, we evaluated the mycotoxin production potential of starters isolated from industrial doenjang koji that have previously been used without proper species identification. Such evaluation would guarantee the safety of products manufactured using these starters. In addition, the two strains identified in this study can safely be used as fungal candidates for the development of other starters for soybean fermentation.