Process Biochemistry 41 (2006) 14671474 www.elsevier.

com/locate/procbio

Review

Identication and quantication of mycotoxigenic fungi by PCR

Abstract This review was inspired by an apparent oversight. A report claimed that a gene probe based on a regulatory gene for aatoxins could be used selectively for screening foodstuffs. However, aR also regulates sterigmatocytin production so that many other fungi could provide a positive result. I suggest that aP, or aQ are more logical choices. Other aspects are reviewed including why it is valid to screen for the metabolic pathway rather than marker DNA, and emphasising that the current state of fungal taxonomy does not permit absolute condence in delineations of taxa. Also, the gene sequences determined from very few strains may not represent the situation in nature. Common genes for a wide range of important mycotoxins (e.g. polyketide synthetase) may not be able to be used with authority, and more specic ones are desirable (e.g. isoepoxydon dehydrogenase). Metabolomics may challenge PCR analysis under certain circumstances and the most appropriate technology needs to be considered. Negative PCRs can be false. Quantifying fungi is a surprisingly inaccurate science, and also in relation to mycotoxin concentration. It is noticeable that few strains of taxa have been investigated in many cases. PCR of mycotoxigenic Aspergillus, Fusarium, and Penicillium in particular are reviewed in the present paper. # 2006 Elsevier Ltd. All rights reserved.
Keywords: Fungi; Aatoxin; Ochratoxin A; Patulin; Fusarium toxins; PCR

Contents 1. 2. 3. 4. Introduction . . . . . . . . . . . . . . . . . . . . . PCR . . . . . . . . . . . . . . . . . . . . . . . . . . . Primers for general sequences . . . . . . . . . Primers for mycotoxin pathway sequences 4.1. Aatoxins . . . . . . . . . . . . . . . . . . 4.2. OTA . . . . . . . . . . . . . . . . . . . . . . 4.3. Fusarium . . . . . . . . . . . . . . . . . . . 4.4. PR toxin . . . . . . . . . . . . . . . . . . . 4.5. Patulin . . . . . . . . . . . . . . . . . . . . Conclusions. . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1467 1469 1469 1470 1470 1471 1472 1472 1472 1473 1473 1473

5.

1. Introduction This review was inspired by an apparent oversight. A method was described using primers for the regulatory gene (aR) for aatoxins biosynthesis to detect fungi in food [1]. However, the same gene is involved in sterigmatocystin

E-mail address: russell.paterson@deb.uminho.pt. 1359-5113/$ see front matter # 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.procbio.2006.02.019

production: more on this later. Now to a review of the polymerase chain reaction (PCR) identication of mycotoxigenic fungi where some novel points are raised. These will indicate the advantages and pitfalls of the process ranging from the specicity of particular genes in taxa to what constitutes a fungus species [2] in this wide-ranging review. Mycotoxins are compounds from fungi that contaminate foodstuff and have detrimental effects on humans and animals [10]. They are referred to as secondary metabolites which are

1468

R.R.M. Paterson / Process Biochemistry 41 (2006) 14671474

low molecular weight compounds that do not contribute to the bioenergetics of the producing fungi: The metabolites require energy to be produced. Production is restricted to certain taxa or strains within taxa. They often exert their effects external to the organism, but not exclusively. On the other hand, primary metabolites are the well-known and almost-universal compounds involved in life such as (a) DNA, (b) proteins, and (c) the small compounds involved in glycolysis, and the hexose monophosphate shunt mechanisms. The use of PCR to identify mycotoxin fungi is attracting considerable attention [3,4]. These methods are based on genes separate from mycotoxin biosynthesis. However, there are only a few mycotoxins about which the biochemistry has been determined sufciently to enable the development of gene probes of the pathway [5]. It is worth mentioning that many of the methods are intended ultimately to be used on virgin samples, i.e. those which otherwise have not been examined [6]. The concept Fungi [7] is not as straightforward as it might rst appear. Fungi can be used to refer to the monophyletic true fungi in contrast to fungi which are organisms traditionally studied by mycologists. This is perhaps not the place to enter into the details, except to state that mycotoxigenic fungi are true fungi. Fungi form a separate kingdom and it has been estimated that there are 1.5 million species although only 70,000 species have been named [8]. So any implication that a particular PCR method has screened representative samples of fungi is immediately open to scepticism. The question is raised as to which species are actually being analysed in environmental samples by PCR if so few are known. Inherent difculties with fungal taxonomy must be considered [2,9]. However, it is known is that some fungi produce toxic compounds, which are often associated with food and feed and are referred to collectively as mycotoxins [10]. The estimated cost of control of mycotoxins in the United States of America alone is $ 1.4 (US) billion. Such fungi are also gaining credibility as sources of health problems in houses. Furthermore, in the current state of world concerns with bio safety and bioterrorism, the use of mycotoxins as weapons cannot be ignored possibly in relation to food and/or water supplies. Indeed, the unambiguous identication of mycotoxigenic fungi remains the most critical area of mycological taxonomic research because of the importance of mycotoxins and the somewhat confused state of the systematics. Despite this, a chapter was not devoted to fungal classication in the

recent, otherwise comprehensive, Council for Agricultural Science and Technology (CAST) report on mycotoxins [10]. The best argument for identifying problem fungi is that it indicates control points within the food system as part of a hazard analysis critical control point (HACCP) approach [11]. This assumes there is a close link between fungus and toxin. However, conventional methods for isolating and identifying fungi are time consuming and require admirably dedicated taxonomists [9]. The classication of fungi is an immature science from lack of attention compared to other groups of organisms [2,12]. However, fungi are unique as indicated by the placement of them in a separate kingdom as mentioned previously. For a practical identication of use to the food industry, and to avoid misidentication, it may be more useful to think of a collection of isolates grouped as a single taxon which produces a particular mycotoxin [9,13]. This would support the view that a species name could be used to predict features of the species. Mycotoxins can be neurotoxic, nephrotoxic, hepatotoxic, immunosuppresive, teratogenic, and/or carcinogenic. A demand exists for rapid and reliable techniques to detect mycotoxins and mycotoxin-producers within foodstuffs, homes, etc. It is worth emphasising that PCR methods do not detect mycotoxins. The producing fungi are microscopic lamentous fungi observed commonly on substrates as mould for example. Important genera are Penicillium, Aspergillus and Fusarium. Claviceps spp. with their ergot alkaloids are often considered separately as plant pathogens. However, the involvement in human disease of ergots in the middle ages is fascinating not least because it indicates the importance of mycotoxins in general. Mycotoxins have been responsible directly for deaths in animals including humans. Most countries have statutory limits for these compounds in some foods [10]. It is now worth indicating the information that PCR and alternatives will provide to assist users in deciding what is required for their own particular work (Table 1). Conventional plating out methods will indicate specic viable fungi from a sample. Analysis of the pure culture (e.g. by chromatography) will inform if it can produce particular mycotoxins. PCR analysis will detect relevant genes in a sample (and it is possible to determine whether, or not, such genes are expressed). Chromatography, for example, will determine if a sample actually contains particular mycotoxins and by implication that a producing fungus was present at some stage. This is the eld of metabolomics. The present author has

Table 1 Some merits and demerits of various identication methods particularly as they relate to mycotoxigenic fungi Identication method Traditional colony isolation Chromatographic analyses of cultures PCR of non-mycotoxin genes PCR of mycotoxin genes Culture independent PCR Metabolomic analysis of environmental samples Merit Viability, relates to taxonomy Mycotoxin detected Identity of taxa Indicates potential for toxin production Detects some unculturable fungi, rapid Historical identication possible, determine if complete mycotoxin pathway is expressed, rapid Demerit Unrepresentative, slow Do not know if mycotoxin is simply below detection limits if not detected, slow Unclear how relates to mycotoxin production Unclear how relates to taxonomy Ditto Possibly less sensitive than PCR

R.R.M. Paterson / Process Biochemistry 41 (2006) 14671474

1469

referred it as secondary metabolomics as secondary metabolites are being considered [47]. The levels of detection for mycotoxins are extremely sensitive and now concentrations of as low as 1015 (106 is the mg kg1 level) can be conceived of with NMR and mass spectroscopy. Furthermore, it is possible to determine if all the genes of a pathway can be expressed by incubating a sample and assessing whether concentrations of the mycotoxin increase. It is certainly worth mentioning that taking representative samples [10] from the environment is a science in its own right that has been studied particularly for mycotoxin analysis but applies equally to all the above areas and is often overlooked. Finally, does high resolution DNA melting have a role in systematics [2], or has this been overlooked in the hegemony of PCR? To ensure that future mistakes are minimised there follows a review of the role of PCR in detecting mycotoxigenic fungi and some apparent misconceptions are addressed. 2. PCR This is an elegant, and well-known, technique that has brought previously inconceivable areas of research into reach. There is a large literature on the subject. However, the basic process involves denaturing (separating the individual strands) DNA by heat applied for specic lengths of time. A small segment of DNA which can be referred to as a probe, is targeted to anneal with the piece of DNA of interest (the target). This is extended to equal the number of kb of the product to be amplied and to yield doubled DNA. The process is cycled approximately 40 times to give the desired quantity of DNA product. The use of the thermophilic polymerases involved in the reactions is of course revolutionary. The DNAs are separated on gels often of agar and the negatively charged DNA is separated through the gel on the basis of size. Above certain sizes the DNA does not separate and so techniques such as pulse eld electrophoresis are required. However, this does not need to concern us for the present topic which is large in scope in any case. The technique has had great utility in fungal phylogenetics where determining the DNA structure holds the promise of attaining true evolutionary relationships. However, the DNA sequences do not mean a great deal in isolation and need to be linked to some other relevant attribute(s) [12]. The status of a catholicon has been achieved by PCR, and there are few proposals for funding that would be brave enough not to include a reference to the reaction. It is a breakthrough that small pieces of DNA can be amplied and detected routinely. However, it has pushed some appropriate technology to the background. There has been a renaissance in the eld of metabolites (via metabolomics) which could conceivably compete (see [14]). An embryonic metabolomic approach is provided in [15,16] where more fungal secondary metabolites were detected from poor than better quality food. Each metabolite is capable of being produced by a limited number of fungi, and so fungal identications of a limited sort can be made. Of course, more sophisticated analytical methods [17] would greatly assist this type of work; routine methods are also available for individual mycotoxins [10]. What do PCR methods offer?

It is common to analyse pure cultures of fungi by PCR. This is often undertaken for taxonomic, or more specically phylogenic, purposes [2]. However, in a review such as this it is sometimes necessary to emphasise the potential pitfalls: two bands of equal size on a gel may not represent the same DNA fragment and as the relationship between two analysed fungi increases the chances of this occurring increases; one of the biggest disadvantages with PCR is the risk of (invisible) contamination of the reaction with alien DNA [18,19]. Very small amounts of contamination will be amplied if the correct sequence is present. Conversely, false negative reactions have to be carefully assessed and control PCRs need to be mandatory [20]. It is possible to (re) amplify an apparently negative sample and obtain a positive result [6]. Whether this relates to contaminants, freeze/thaw cycles and/or detection limits needs to be determined. Also, the gene may be present but is it expressed? Apparently single bands on gels may not be: issues of inhibition and nucleic acid damage need to be carefully considered [21]. Finally, does the in vitro work translate to real-life situations? As indicated previously, a logical step is to apply the methodology to environmental samples [6]. This has been referred to culture independent PCR [54] and has been recognised for a number of years previously for fungi (see [21]). Questions of the minimum quantity of propagules that can be detected must be considered (not just conidia) as well as issues of gene copy number in the cells (e.g. some genes may exist as multiple copies in individual cells). As discussed, the number of detected cells is lower than the genome equivalents, and false negative reactions need to be considered [20]. In addition, inhibition may be problematic, as in pure culture [21]. However, this technology gives at least an approximation to the true situation. For mycotoxigenic fungi there are two approaches: (a) the analysis of genes not involved with mycotoxin production and (b) the converse. 3. Primers for general sequences Papers cited in [5] appear to assume that non-mycotoxin biosynthetic pathways will indicate adequately mycotoxin production potential. However, inadequate taxonomy and identications [9] imply there may not be a direct relationship. A great reliance is made on current taxonomies of the target fungi in the cases where general genes are employed. For example, the polygalacturonase gene is claimed to identify Penicillium expansum [3]. This fungus is a member of the terverticillate penicillia which is one of the most variable groups of fungi, and many more representative strains and taxa need to be examined before it can be reliably employed. A partial b-tubulin sequence was used to separate many fungi within this group, although the two ochratoxin A (OTA) species, P. verrucosum and P. nordicum, were virtually identical [22]. So a probe based on b-tubulin could not separate these different species. Peterson [23] considered the whole group to be monophyletic. However, the work of Karolewiez and Geisen indicate that the two species were quite different [24]. This indicates how important it is to use multi-phasic

1470

R.R.M. Paterson / Process Biochemistry 41 (2006) 14671474

approaches. A good tool is described for separating A. niger and other aspergilli based on PCR of the ITS sequence [50]. It is described as an early detection tool for OTA-producing Aspergillus species. However, effectiveness is dependant again on an association with species and OTA production, which is seldom exact and where primers for the metabolic pathway would be more useful. This was extended to include A. carbonarius and A. ochraceus in Patino et al. [51]. It is noted that some strains produced detectable OTA but were negative for the PCR products. The authors provided information on the detection limits of the ITS amplication product which was also interesting. The most extensive single general sequence work is [4], where a generic assay for target species was developed. The tests varied in specicity from species or subspecies to closely related species groups. All the issues concerning the number of genes in fungal structures and how this relates to quantity apply. As does using a gene not involved in mycotoxin production unlike the situations cited below. Of course, the assays were developed on conidial suspensions and how that relates to hyphae and other spores is not recorded but could be signicant nevertheless: Spores of all kinds can be uni- or multinucleate. The nuclei of the latter may be derived from a single parent nucleus or from several nuclei and so may be homokaryotic or heterokaryotic (see also [24]). Therefore PCR analysis of these spores will overestimate the number of fungi (depending upon how fungal quantity is dened which is, in itself, not straightforward). Obviously, environmental samples may contain all structures. The conidial suspensions tested were only 95% pure, and one wonders how many genes are equal to 5% hyphae. Some conidial suspensions were combinations of ascospores and conidia (in addition to hyphae). Further, species assignments were sometimes made from sequences published in Gene Bank which may be problematic although such statements cannot be made from supercial studies (see [25] and correspondence in the same volume of the journal). This of course puts an extremely high premium on existing taxonomies and identications. Finally, non-orthologous genes could affect results. This means that more than one gene in a strain could code for enzymes that perform the same function (e.g. in a metabolic pathway) and so would overestimate the number of fungi present in the sample. Other issues of using the procedures as an estimation of mycotoxin levels is severely problematic as (a) the toxins could be present without the fungi and (b) quantication is not one gene one conidium. It would be optimal to relate a PCR method with quantication of the mycotoxin of interest (see [26] for in vitro comparisons), or a more quantitative nonmycotoxin technique such as ergosterol [9]. Even here the correlation was not direct at all stages as is discussed. Some observations are made in [4] concerning the relative merits of an identication of AspPen compared with their methods. AspPen was the term used to describe a general identication of a fungus that belongs to the aspergilli and/or penicillia. One could argue that an identication based on a sequence which is unrelated directly to mycotoxin genes could be equally misleading (see later). Also, it would be entirely

possible to simply analyse the sample of interest for offending mycotoxins by chromatography, spectroscopy and/or NMR, with or without an incubation period, where quantication can be achieved using well documented protocols. Similar arguments can be made for [3] for P. expansum where the polygalacturonase gene is employed. Two further specic examples are now provided. Problems from using RAPDs are demonstrated in the case of toxigenic Fusarium spp. where species specic primers were not tested against closely related species [5]. This is a general problem and needs to be addressed with any PCR work and could be controlled at a prepublication stage. An extensive study by Jurado et al. [27] claims to distinguish mycotoxin producing species one from the other. However, no attempt was made to relate the results to actual production of the mycotoxins as part of the experimental procedures. Better was the correlation observed for RAPD patterns and secondary metabolite formation patterns in P. roqueforti [28]. b-Tubulin sequences were used to separate P. expansum from P. solitum in Sholberg et al. [48]. However, the two are quite different from morphology and of course P. expansum is associated with patulin production and P. solitum is not: it would be interesting to test P. solitum for the isoepoxydon dehydrogenase (idh) gene as discussed later. The four subclusters indicated for P. expansum merit further investigation. Incidentally, PCR methodology is available for Claviceps [29] although they tend to be treated historically from a plant pathogenic, rather than a mycotoxin perspective. What is the situation regarding the genes controlling the biosynthetic pathways involved in the production of mycotoxins? 4. Primers for mycotoxin pathway sequences In these cases, the metabolic pathway is being assessed of direct relevance to mycotoxin potential. There is no reliance on taxonomy whatsoever if the gene is specic for the chosen mycotoxin (see later). A chance contamination may also represent a problem but this would only be intermittent, and an overall assessment could still be made. Of course the fact that PCR analysis can be coupled to mycotoxin production is a large advantage [30]. (Indeed, the benets of reverse transcriptase PCR are somewhat negated by the possibility of undertaking mycotoxin analysis per se.) This technology will be useful for determining associations between detection of a gene at critical control points in food production (e.g. apple orchards) and quantication of the mycotoxin in the nal product (e.g. apple juice) (see [6]). It has implications for a functional taxonomy of the species involved, and will distinguish between genetically different strains of the same species (cryptic species). So which specic probes are available? 4.1. Aatoxins The biosynthetic pathways for the aatoxins have been determined [31]; they are typical polyketide compounds. Many potential probe sites are available. However, which reaction

R.R.M. Paterson / Process Biochemistry 41 (2006) 14671474

1471

steps are unique to the aatoxins, and would not include sterigmatocystin, another potent mycotoxin? The dozens of fungi which are known to produce the precursors within the aatoxin biosynthetic pathway are provided in [32]. The shared metabolic pathways for aatoxins and sterigmatocystin is what was not realised by [1], and apparently in [33,34]. In [1], only the single aR gene was used which was not tested with recognised sterigmatocystin producers per se. Interestingly, A. sojae contains the gene although it is defective [31]. Chang et al. [35] describe a system using the aR regulatory gene which can distinguish aatoxigenic fungi from other fungi found commonly on grains. However, I would argue that the system could create false positives from sterigmatocystin fungi on the same substrate. Although an intriguing result was presented almost as an aside that F1 and R1 primers did not generate PCR products in A. nidulans, A. versicolor, Chaetomium thielaviodeum single strains, and some A. oryzae. It concluded that if counterparts of aR are present they exhibit low similarity to the sequence for aR. Further, Yu et al. [36] stated that the AR homologs in A. avus and A. parasiticus are almost identical, whereas the A. nidulans one is 31% similar. However, the zinc binuclear region is the most highly conserved having 71% identity amongst the aspergilli. So one of my conclusions is simply that any probe for aatoxin species in food should be tested carefully with numerous exclusively sterigmatocystin fungi. The current author has tested a much wider range of taxa for false positives for the idh gene (see later). Of course idh is not part of the aatoxin pathway and would not be relevant to the situation under discussion above. Furthermore, an enrichment procedure could have been employed to analyse for the aatoxins themselves hence removing any doubt as to whether the genes were expressed. Enrichment is also suggested in [1]. However, clearly aP and probably better aQ, are not relevant to sterigmatocystin biosynthesis [31] and would be a valid choice for specic aatoxin primers for food, etc. If the gene sequences for the specic conversions to individual aatoxins were available they too would be of great utility in this respect. There is more than one aatoxin and there is not as yet a set of primers for one specic compound, the most obvious choice being aatoxin B1the most toxic (see Table 2). Finally, Manonmani et al. [1] also claims that A. avus produces the aatoxin Gs which conventionally is not considered to be the case. It is possible that genes from a morpho-species (a species dened by morphology) would have specic sequences but this

cannot be assumed. In this respect it is informative to consider [37], where a multiplex PCR reaction was developed for nor-1, ver-1, omt-A aatoxin (and sterigmatocystin) genes. Triplets were indeed observed for A. avus, A. parasiticus and the sterigmatocystin-only producing species, A. versicolor. Interestingly, so-called aatoxin producing negative strains of A. avus varied in the patterns, including one from which all three bands were recorded. The system could not distinguish between aatoxigenic and non aatoxigenic strains of A. avus. Could it actually produce small concentrations of the compounds? A consideration of the terms not produced versus not detected is provided in [9]. This has obvious negative implications for genes not even involved in mycotoxin production if the ones that are cannot differentiate (see above). What is the situation with other mycotoxins? 4.2. OTA The steps in OTA biosynthesis have not been established and proposed pathways are hypothetical. A polyketide synthetase (PKS) is probably involved. OTA is a complex molecule combining the amino acid and the polyketide structure basically of the mycotoxin citrinin. Of course OTA is quite different from citrinin. However, would an OTA gene based on the polyketide portion differentiate OTA from citrinin fungi? The answers are no in one case and yes in another [38] where two PKS-based primers are involved, although crucially no OTA-producing penicillia were tested. Furthermore, citrinin is a co-metabolite of dihydrocitrinone and sclerin in, for example, A. carneus and there are other metabolites similar to citrinin which may share PKS genes. All these need to be considered as sources of false positives for OTA fungi. OCallaghan et al. [39] report the rst cloning and characterisation of a gene involved in OTA biosynthesis but reafrm that there was no information involved in the steps from acetate and malonate to the isocoumarin group. Work is continuing on the elucidation of the other genes involved in the pathway. Again, this does not help in the search for a specic gene. Perhaps more interestingly, Karolewiez and Geisen [24] demonstrate that a particular PKS gene can be identied only in P. nordicum and not in P. verrucosum or Aspergillus spp. Although, whether the PKS is involved in biosynthesis can only be surmised at present. The conclusion is that OTA PKSs are potentially useful screening genes. However, are the primers specic to OTA and not other citrinin species (e.g. P. citrinum)?

Table 2 Comparison of the potential problems and solutions of the non-polyketide synthetase gene probes of the principal mycotoxins Mycotoxin Patulin Aatoxins Ochratoxin A Fusarium trichothecenes Gene idh Various (e.g. aR) Various Various (e.g. Tri5) Problem None. Specic. Patulin is a single compound. Precursors (e.g. isopatulin) are not mycotoxins Various exclusively sterigmatocystin fungi detected. All four aatoxins are major concerns Complex molecule. Exclusively citrinin producers detected (e.g. P. citrinum) Cannot detect individual deoxynivalenol or nivalenol producers (i.e. not specic) Solution None required Look for genes upstream of sterigmatocystin; probes for individual aatoxin steps? Probe for the step governing the combination of phenylalanine to the citrinin molecule More work required to target specic genes

1472

R.R.M. Paterson / Process Biochemistry 41 (2006) 14671474

It is worthwhile pointing out that P. nordicum and P. verrucosum were considered different biochemically and genetically. However, they were virtually identical in [22] by b-tubulin sequences. The capacity of fungal strains which had been genetically manipulated to be defective in production of secondary metabolites, but which could in fact still manufacture low quantities is revealing [24]. What does defective mean? Do the genes not work or are they simply impaired? To continue, it would have been instructive to have measured OTA after 10 days in [26]: was the apparent correlation maintained with the real-time PCR? Also, more than one analysis per day would have been benecial. It is difcult to extrapolate as to what may occur on real food. The lack of correlation with such a crucial parameter as temperature is noted. Finally, it is possible to question the need for the realtime procedure when samples can simply be analysed for OTA and at very low concentrations, with incubation if required. Furthermore, some problems exist with other mycotoxin systems. 4.3. Fusarium The real problem with these is that there are so many trichothecenes (see Table 2). Supposedly unique bands for F. avenaceum were observed also from F. tricinctum, and primers for F. moniliforme were not tested unfortunately with the related complex Gibberela fujikuroi (they were demonstrated subsequently to cross-react). Various other problems are reported for PCR of the fusaria [5]. However, PCR to differentiate between the chemotypes of F. graminearum which produce deoxynivalenol (DON), and nivalenol (NIV) is suggested in [40]. The other taxa that produce these, such as F. culmorum, F. cerealis, F. crookwellense, or G. zea were not considered. Further, there are now nine species that are reported to constitute the F. graminearum complex [41] and that species do not conform to mycotoxin patterns [42] making the situation regarding a general probe complex indeed. Basically, the answers to these questions have yet to be found. As suggested by [40], their method may be better utilised to determine two mycotoxin chemotypes of Fusarium in general rather than for F. graminearum. As Edwards et al. [5] stated, any assay to distinguish DON and NIV production (or potentially producingmy addition) needs to be tested against all lineages of both chemotypes. However, to state that PCR allows the detection of trichothecene producing Fusarium species is misleading, and further that such tests could indicate the presence or absence of all Fusarium trichothecenes is untrue as this can only be achieved by non-PCR methodology (often chromatography). Although mainly dealing with non-mycotoxin pathway genes, [43] is worth a special mention as it relates these to the Tox 5 gene in particular and mycotoxin analysis per se. For example, 18% of F. equiseti as dened were positive for Tox 5. However, somewhat ambiguous results were obtained for actual analysis with detection sometimes being recorded. However, to state that a PCR assay has been developed for detection of

trichothecenes (p282) is simply a lack of comprehension, and one wonders why this was not eliminated prior to publication. The biosynthesis of the polyketide compounds moniliformin, zearalenone and fumonisins have not been established. Nevertheless, various PCR methods have been developed and there has been some less-than-robust testing of related species (see above). These compounds are considered briey in [5]. 4.4. PR toxin This is perhaps not one of the major mycotoxins but illustrates some interesting points. Some potential problems mentioned with PCR of the PR toxin genes in reality represent opportunities and provide a model for other more important mycotoxins. Penicillium strains produced PR toxin, although a PCR fragment was not observed. However, the gene was detected by more sophisticated methods. Some species were positive for the gene which had never been reported to produce PR toxin: this indicates that the probe can provide new information to assist with a practical taxonomy such as has been suggested for penicillia in general (see below and [13]). The genes may be present but not expressed. Interestingly, it was postulated that the ari1 gene or similar may be involved in producing other secondary metabolites given the relatively unspecic nature of secondary metabolite genes [49]. A compound which has been investigated in various guises over the decades next will be considered. 4.5. Patulin The useful thing about patulin is that the pathway is so specic without other mycotoxins being involved (see Table 2). The compound is second only to aatoxins in terms of the amount of data that exist because of the antibiotic and model secondary metabolite roles of the compound. It has the clear advantage as a model mycotoxin of its singularity rather than being one of many patulins as end products. Patulin is conceived of as being formed from isoepoxydon ! phyllosphyllostine ! isopatulin ! ascladiol ! patulin where the precursors are naturally similar to patulin. There are dimeric compounds which can be conceived of as being formed from two patulin molecules but (a) the formation of these structures is not well established, (b) they are not co-produced with patulin, and (c) they are not mycotoxins. In comparison, all four important aatoxins mycotoxins have a common precursor and any potential dimeric compounds simply would not be considered within the same category. The idh gene catalyses the conversion of isoepoxydon to phylostine: Much of this work is based on the data from Gauchers group [55]. Similar phenomena to PR toxin of previously unrecorded patulin producers containing the isoepoxydon dehydrogenase (idh) gene of the patulin biosynthetic pathway were not considered to be false claims [30,44,45]. Incidentally [5], ascribed wrongly the possibility of patulin being translocated through soil into apples to Mantle, rather than Paterson et al. [44]. Mantle provided evidence for OTA moving from soil to coffee beans as discussed in [44] and the current author made a simple

R.R.M. Paterson / Process Biochemistry 41 (2006) 14671474

1473

comparison for patulin and apples. Again Paterson et al.s conclusions appear to have been misinterpreted as the idh primers do not simply identify P. expansum and P. brevicompactum from environmental samples as suggested in [5] but is certainly conceived of as being capable of identifying all organisms that possess the idh gene. A new identication procedure of fungi by reducing all idh positive fungal strains to pen idh+ [13] has been suggested, which was based on a premise in [9] for OTA. There are no similar reactions from isoepoxydon to phyllostine described leading to another end-point metabolite, although epoxydon and phyllostine have been reported in Phyllostica spp. [46] which are associated with infection of leaves. It would be interesting to determine if representative strains produce patulin. Of course, other important secondary metabolites could be described in the future in fungi that are produced via isoepoxydon to phyllostine but this is speculation. Paterson et al. [21,30] extended the work to a wide range of penicillia indicating that the gene is widespread. Importantly, negative idh strains and positive patulin detection were not observed. When strains were demonstrated to be idh positive but patulin negative it was concluded logically that (a) some part of the patulin metabolic pathway was not functioning although the idh gene was intact or (b) patulin was being produced at below the detection limits of the chromatography system. Also, the gene and production from other genera were demonstrated [21,44]. There are increasing amounts of information on the sequence of the idh gene [52,53]. Dombrink-Kurtzman [53] observed differences between the idh sequences of P. expansum and P. griseofulvum which were correlated to the quantities of patulin that each species could produce. In White et al. [52] various other genes were also cloned in the search for a molecular based detection method for P. expansum. They demonstrated that patulin was regulated at the transcriptional level and represented the rst report of this phenomenon. The idh gene method was also used by Varga et al. in Aspergillus where it was suggested that the occurrence of patulin in taxa was subjected to evolutionary inuences without considering the possible effects of preservation (see [45]). Furthermore, Paterson [21] introduced the concept of secondary metabolites affecting the analysis by inhibiting PCR. Results from environmental samples indicated that the gene was detected in orchard soils in comparison to a nonorchard soil and HACCP procedures were proposed [6,44]. It is emphasised that the idh gene is not a PKS type: equivalent sites are required to be investigated for some of the other main toxins. 5. Conclusions The promise of PCR to identify fungi has blinded some to its potential pitfalls. Contamination is a real problem for all PCR. DNA from fungi related to the target fungus may interfere with the analysis in the environment. Morphologically identical species to mycotoxigenic fungi without the ability to produce a particular mycotoxin may result in false positives if general genes are employed. In addition, genes which are shared for

different mycotoxins (e.g. sterigmatocystin and aatoxins) would tend to overestimate contamination problems. Nonviable cells or pieces of DNA may provide results. On the other hand, the benets obtained from undertaking mycotoxin analysis per se are often not fully appreciated. Potential mycotoxin production can be detected by PCR which may permit the establishment of critical control points and is a signicant advantage. The detection of unculturable fungi is a large advantage of PCR. In general, a more critical assessment of the benets of PCR methodology would be advantageous. Finally, I have been struck by how few strains of each taxon have been studied in most of the papers. It is possible to inquire just how representative the results are of the situation in nature. Acknowledgements This paper is dedicated to the memory of my colleague and friend Dr. David Brayford, the Fusarium taxonomist. Undertaken as part of grant SFRH/BPD/14923/2004 from Fundacao para a Ciencia e a Tecnologia (FCT), Portugal. References
[1] Manonmani HK, Anand S, Chandrashekar A, Rati ER. Detection of aatoxigenic fungi in selected food commodities by PCR. Process Biochem 2005;40:285964. [2] Burnett J. Fungal populations and species New York: Oxford University Press Inc.; 2003. [3] Marek R, Thiranavakkarasu A, Venkitanarayan K. Detection of Penicillium expansum by polymerase chain reaction. Int J Food Microbiol 2003;89:13944. [4] Haughland RA, Varma MM, Wymer LJ, Vesper SJ. Quantitative PCR analysis of selected Aspergillus, Penicillium and Paecilomyces species. Syst Appl Microbiol 2004;27:198210. [5] Edwards SG, OCallaghan J, Dobson ADW. PCR-based detection and quantication of mycotoxigenic fungi. Mycol Res 2002;106:100525. [6] Paterson RRM. Primers from the isoepoxydon dehydrogenase gene of the patulin biosynthetic pathway to indicate critical control points for patulin contamination of apples. Food Control 2006;7:7414. [7] Lutzoni F, Kauff F, Cox CJ, McLaughlin D, Celio G, Dentinger B, et al. Assembling the fungal tree of life: progress, classication, and evolution of subcellular traits. Am J Bot 2004;91:144680. [8] Hawksworth DL. Fungal diversity and its implications for genetic resource collections. Stud Mycol 2004;50:918. [9] Paterson RRM, Venancio A, Lima N. Solutions to Penicillium taxonomy crucial to mycotoxin research and health. Res Microbiol 2004;155:507 13. [10] CAST. Mycotoxins: risks in plant, animal and human systems Ames, IA: Council for Agricultural Science and Technology; 2003 [2/1/06]http:// www.cast-science.org/cast/pub/Mycotoxins.pdf. [11] FAO. Manual on the application of the HACCP system in mycotoxin prevention and control. FAO Food and Nutrition Paper 73; 2003. http:// www.fao.org/DOCREP/005/Y1390E/y1390e00.htm [2/1/06]. [12] Wheeler QD. Taxonomic triage and the poverty of phylogeny. Philos Trans R Soc Lond 2004;359:57183. [13] Paterson RRM, Venancio A, Lima N. An applied identication scheme for Penicillium strains by patulin and isoepoxydon dehydrogenase gene detection with utility in HACCP control. In: Microbes in a Changing World, XI International Congress of Mycology. 2005. p. 9. [14] Phelps TJ, Palumbo AV, Beliaev AS. Metabolomics and microarrays for improved understanding of phenotype characteristics controlled by both genomics and environmental constraints. Curr Opin Biotechnol 2002;13:204.

1474

R.R.M. Paterson / Process Biochemistry 41 (2006) 14671474 [36] Yu J-H, Butcher RAE, Fernandes M, Keller NA, Leonard TJ. Conservation of structure and function of the aatoxin regulatory gene aR from Aspergillus nidulans and A. avus. Curr Genet 1996;29:54955. [37] Geisen R. Multiplex polymerase chain reaction for the detection of potential aatoxin and sterigmatocystin producing fungi. Syst Appl Microbiol 1996;19:38892. [38] Dao HP, Mathieu F, Lebrinhi A. Two primers to detect OTA producers by PCR method. Int J Food Microbiol 2005;104:617. [39] OCallaghan J, Caddick MX, Dobson ADW. A polyketide synthase gene required for ochratoxin A biosynthesis in Aspergillus ochraceus. Microbiology (UK) 2003;149:348591. [40] Li H-P, Wu A-B, Zhoa C-S, Scholten O, Lofer H, Liao Y-C. Development of a generic PCR detection of deoxynivalenol- and nivalenol-chemotypes of Fusarium graminearum. FEMS Microbiol Lett 2005;243:50511. [41] ODonnell K, Ward TJ, Geiser DM, Kistler HC, Aoki T. Genealogical concordance between the mating type locus and seven other nuclear genes supports formal recognition of nine phylogenetically distinct species within the Fusarium graminearum clade. Fungal Genet Biol 2004;41:60023. [42] Demeke T, Clear RM, Patrick SK, Gaba D. Species-specic PCR-based assays for the detection of Fusarium species and a comparison with the whole seed agar plate method and trichothecene analysis. Int J Food Microbiol 2005;103:27184. [43] Monds RD, Cromey MG, Lauren DR, Menna Mdi, Marshall J. Fusarium graminearum, F. cortaderiae and F. pseudograminearum in New Zealand: molecular phylogenetic analysis, mycotoxin chemotypes and co-existence of species. Mycol Res 2005;109:41020. [44] Paterson RRM, Archer S, Kozakiewicz Z, Lea A, Locke T, OGrady E. A gene probe for the patulin metabolic pathway with potential use in novel disease control. Biocontrol Sci Technol 2000;10:50912. [45] Paterson RRM. The isoepoxydon dehydrogenase gene of the patulin biosynthetic pathway, patulin detection, and the utility of species names in patulin-producing fungi. In: Lima N, Smith D, editors. Biological Resource Centres and the use of Microbes, European Culture Collections Organisation XXII. Braga: Micoteca da Universidade do Minho; 2003. p. 25966. [46] Turner WB, Aldridge DC. Fungal metabolites II London: Academic Press; 1983. p. 631. [47] Paterson RRM. Book review. Handbook of secondary fungal metabolites. Mycol Res 2004;108:721. [48] Sholberg PL, Harlton C, Haag P, Levesque CA, OGorman D, Seifert K. Benzimidazole and diphenylamine sensitivity and identity of Penicillium spp. that cause postharvest blue mold of apples using b-tubulin gene sequences. Postharvest Biol Technol 2005;36:419. [49] Geisen R. PCR methods for the detection of mycotoxin producing fungi. In: Bridge PD, Arora DK, Reddy CA, Elander RP, editors. Applications of PCR in mycology. Wallingford: CAB International; 1998. p. 24366. [50] Gonzalez-Salgado A, Patino B, Vazquez C, Gonzalez-Jaen MT. Discrimination of Aspergillus niger and other Aspergillus species belonging to section Nigri by PCR assays. FEMS Microbiol Lett 2005;245:35361. [51] Patino B, Gonzalez-Salgado A, Gonzalez-Jaen MT, Vazquez C. PCR detection assays for the ochratoxin-producing Aspergillus carbonarius and Aspergillus ochraceus species. Int J Food Microbiol 2005;104:207 14. [52] White S, OCallaghan J, Dobson ADW. Cloning and molecular characterisation of Penicillium expansum genes upregulated under conditions permissive for patulin biosynthesis. FEMS Microbiol Lett 2006;255:1726. [53] Dombrink-Kurtzman MA. The isoepoxydon dehydrogenase gene of the patulin metabolic pathway differs for Penicillium griseofulvum and Penicillium expansum. Antonie van Leeuwenhoek, in press. [54] Jacobsen J, Lydolph M, Lange L. Culture independent PCR: an alternative enzyme discovery strategy. J Microbiol Methods 2005;60:6371. [55] Fedeshko RW. Polyketide enzymes and genes in Penicillium urticae. PhD thesis. The University of Calgary, Canada; 1992.

[15] Friere FCO, Kozakiewicz Z, Paterson RRM. Mycoora and mycotoxins in Brazilian black pepper, white pepper and Brazil nuts. Mycopathologia 2000;149:139. [16] Friere FCO, Kozakiewicz Z, Paterson RRM. Preliminary study of mycoora and mycotoxins from Brazilian cashews. Mycopathologia 1999;145:95103. [17] Nielsen KF, Smedsgaard J. Fungal metabolite screening: database of 474 mycotoxins and fungal metabolites for dereplication by standardised liquid chromatographyUVmass spectrometry methodology. J Chromatogr A 2003;1002:11136. [18] Paterson RRM. Fungus or bacterium and vice versa? Microbiology 2005;151:641. [19] Paterson RRM. The coronamycin producer: a case of mistaken identity? Mycol Res 2005;109:8501. [20] Hoorfar J, Cook N, Malorny B, de Medici D, Abdulmawjood A, Fach P. Diagnostic PCR: making internal amplication control mandatory. J Appl Microbiol 2004;96:2212. [21] Paterson RRM. The isoepoxydon dehydrogenase gene of patulin biosynthesis in cultures and secondary metabolites as candidate PCR inhibitors. Mycol Res 2004;108:14317. [22] Samson RA, Seifert KA, Kuijpers AFA, Houbraken JAMP, Frisvad JC. Phylogenetic analysis of Penicillium subgenus Penicillium using partial btubulin sequences. Stud Mycol 2004;49:175201. [23] Peterson SW. Phylogenetic analysis of Penicillium species based on ITS and LSU-rDNA nucleotide sequences. In: Samson RA, Pitt JI, editors. Integration of modern taxonomic methods for Penicillium and Aspergillus classication. Reading: Harwood Publishers; 2000. p. 16378. [24] Karolewiez A, Geisen R. Cloning a part of the OTA biosynthetic gene cluster of Penicillium nordicum and characterization of the ochratoxin polyketide synthase gene. Syst Appl Microbiol 2005;28:58895. [25] Bridge PD, Spooner BM, Roberts PJ. Reliability and use of published sequence data. New Phytol 2004;161:157. [26] Geisen R. Molecular monitoring of environmental conditions inuencing the induction of ochratoxin A biosynthesis genes in Penicillium nordicum. Mol Nutr Food Res 2004;48:53240. [27] Jurado M, Vazquez C, Patino B, Gonzalez-Jaen MT. PCR detection assays for the trichothecene-producing species Fusarium graminearum, Fusarium culmorum, Fusarium poae, Fusarium equiseti and Fusarium sporotrichioides. Syst Appl Microbiol 2005;28:5628. [28] Geisen R, Cantor MD, Hansen TK, Holzapfel WH, Jakobsen M. Characterization of Penicillium roqueforti strains used as cheese starter cultures by RAPD typing. Int J Food Microbiol 2001;65:18391. [29] Pazoutova S, Olsovska J, Linka M, Kolnska R, Flieg M. Chemoraces and habitat specialization of Claviceps purpurea populations. Appl Environ Microbiol 2000;66:541925. [30] Paterson RRM, Kozakiewicz Z, Locke T, Brayford D, Jones SCB. Novel use of the isoepoxydon dehydrogenase gene probe of the patulin metabolic pathway and chromatography to test penicillia isolated from apple production systems for the potential to contaminate apple juice with patulin. Food Microbiol 2003;20:35964. [31] Yu J, Chang P-K, Ehrlich KC, Cary JW, Bhatnagar D, Cleveland TE, et al. Clustered pathway genes in aatoxin biosynthesis. Appl Environ Microbiol 2004;70:125362. [32] Frisvad JC, Thrane U. Liquid column chromatography of mycotoxins. In: Betina V, editor. Chromatography of mycotoxins techniques and applications. The Netherlands: Elsevier. J Chromatogr Library 1993;54:253372. [33] Farber P, Geisen R, Holzapfel WH. Detection of aatoxinogenic fungi in gs by a PCR reaction. Int J Food Microbiol 1997;36:21520. [34] Shapira R, Paster N, Eyal O, Menasherov M, Mett A, Salomon R. Detection of aatoxigenic molds in grains by PCR. Appl Environ Microbiol 1996;62:32702. [35] Chang P-K, Bhatnagar D, Cleveland TE, Bennet JW. Sequence variability in homologs of the aatoxin pathway gene aR distinguishes species in Aspergillus section Flavi. Appl Environ Microbiol 1995;61:403.