Introduction

The mycotoxin patulin is produced by 14 species of the genus Penicillium and is most likely to be found in fruits, silage and dung (Frisvad et al. 2004). Fungi capable of producing patulin can also be found in both soil and indoor air. In most cases, production of secondary metabolites (extrolites) is consistent from isolate to isolate in a species (Frisvad and Filtenborg 1983 1989), suggesting that species-specific probes might be useful for establishing the level of patulin contamination. For example, all isolates of Penicillium griseofulvum examined produce patulin, griseofulvin, cyclopiazonic acid and roquefortine C. Penicillium expansum, the species most commonly linked to patulin production and the species most often found on apples and other pomaceous fruits, also produces other toxins, such as roquefortine C and chaetoglobosin C, but these mycotoxins have not been found in rotting pomaceous fruit (Frisvad and Filtenborg 1989).

The most important species of Penicillium associated with food and described as patulin producers include, in addition to P. expansum, P. griseofulvum, Penicillium carneum, Penicillium paneum and Penicillium sclerotigenum. The foods associated with these latter four species are cereals and pasta (P. griseofulvum), beer, wine, rye-bread and dry meat products (P. carneum) (Frisvad and Filtenborg 1989), rye-bread (P. paneum) (Boysen et al. 1996) and P. sclerotigenum (yams) (Frisvad and Samson 2004); although patulin is not normally found in these foods, patulin is able to be produced when these fungi are grown on laboratory media. The remaining patulin-producing species can be isolated from soil, are primarily dung-associated and are not commonly found in foods, although they could be acquired if fruits, for example, apples, are dropped on the ground and are used in processing. These latter species include the following: Penicillium clavigerum (in soil, not food) (Svendsen and Frisvad 1994); Penicillium concentricum (rarely found in foods) (Leistner and Eckardt 1979); Penicillium coprobium (rarely found in foods) (Frisvad and Filtenborg 1989); Penicillium dipodomyicola (in rice, chicken feed, soil) (Frisvad et al. 1987); Penicillium formosanum (probably in dung); Penicillium gladioli (in Gladiolus bulbs) (Frisvad and Samson 2004); Penicillium glandicola (in silage, soil) (Frisvad and Filtenborg 1989); Penicillium marinum (in coastal sand); and Penicillium vulpinum (in dung, insects and soil) (Frisvad and Samson 2004). A compendium of information on species of Penicillium and the mycotoxins they produce has been described (Frisvad et al. 2004).

The patulin biosynthetic pathway contains at least ten enzymes in the direct pathway (Fedeshko 1992) and as many as a total of 19 enzymes when including those in side pathways. A number of biochemical studies to elucidate the patulin biosynthetic pathway have been performed (Sekiguchi and Gaucher 1978, 1979; Sekiguchi et al. 1983; Fedeshko 1992). The nucleotide sequence coding for the first pathway enzyme, 6-methylsalicylic acid synthase (6-MSAS), has been sequenced by Beck et al. (1990). The enzyme 6-MSAS is responsible for the condensing one molecule of malonyl-CoA and 3 molecules of acetyl-CoA; after extensive multi-step processing of 6-MSA, patulin is formed. Isoepoxydon dehydrogenase (IDH), the seventh enzyme in the pathway, is responsible for conversion of isoepoxydon to phyllostine in the terminal portion of the patulin biosynthetic pathway (Fedeshko 1992). P. griseofulvum strains had been used in the research of Beck et al. (1990) and Fedeshko (1992). Recently, both the full length idh gene and a fragment of the 6-msas gene have been cloned from P. expansum (White et al. 2006). Sequences of only these two genes (6-msas and idh) of the patulin pathway enzymes are present in GenBank.

In the present study, the sequence of the idh gene was determined for all of the species listed above, except for P. formosanum and P. marinum, to compare the extent of amino acid conservation at the site of catalytic activity. The phylogenetic relationships among the species were determined for both rDNA and idh gene sequences.

Materials and methods

Fungal isolates

Selection of patulin-producing species of Penicillium for determination of the nucleotide sequence of the idh gene was based on the mycotoxin determinations of Frisvad and Thrane (1987), as well as from suggestions by Jens C. Frisvad (personal communication). Twelve of the 14 patulin-producing species listed by Frisvad et al. (2004) and used in the study are listed in Table 1 and are maintained in the Agricultural Research Service Culture Collection (NRRL), Peoria, Illinois, USA.

Table 1 Penicillium species for which the isoepoxydon dehydrogenase gene sequence was determined
Full size table

DNA extraction

Cultures of different species of Penicillium were grown on PDA plates for approximately 14 days at 25°C. To each plate was added, 2–3 ml of 70% ethanol and surface growth was brought into suspension using a sterile rod, bent at a 90° angle. The suspension was put into a 1.5 ml Eppendorf tube and centrifuged for 5 min to precipitate the cellular material. The supernatant was discarded and 375 μl of DNA extraction buffer (200 mM Tris, 250 mM NaCl, 25 mM EDTA, pH 8.5, 0.5% SDS) and 125 μl of glass beads (dia. 0.5 mm) (Scientific Industries Inc., Bohemia, NY) were added to a pellet. The tube was vortexed on the TurboMix (Scientific Industries Inc.) for 5–10 min; 350 μl of 2×CTAB (hexadecyltrimethylammonium bromide) buffer was added to the tube of broken cells, which was vortexed again for approximately 30 s, followed by addition of 350 μl of chloroform. The tube was carefully vortexed and then microfuged for 10 min at maximum speed to separate the emulsion. The upper aqueous phase was carefully removed, placed into a new 1.5 ml Eppendorf microfuge tube and 500 μl of isopropanol ( − 20°C) was added to precipitate the DNA. The tube was centrifuged for 5 min and the supernatant was discarded. The DNA pellet was washed with 1 mL of 70% ethanol, centrifuged for 3 min and the supernatant was discarded. The DNA pellet was then dissolved in 100 μl of TE buffer (10 mM Tris, 1 mM EDTA, pH 8.0). A 1:10 dilution of stock DNA in TE/10 buffer was the working stock of DNA.

PCR amplification

Before beginning PCR, concentrations of DNA and primers were determined and verified by agarose gel electrophoresis. Amplification of ITS and 28S rDNA domains D1/D2 was done using primers previously described (Peterson 2000). Primers used for determining the idh sequences are shown in Table 2. Symmetrical amplification was performed in a 96-well plate. The PCR mixture per well contained 6.3 μl of dH2O, 4 μl of 10 ×  PCR buffer, 7.2 μl of deoxynucleotide (dNTP) mix (1.25 mM), 1.0 μl of external 5′ end primer (10 pmol/μl), 1.0 μl of external 3′ end primer (10 pmol/μl), 0.4 μl of Taq Polymerase (Sigma Chemical Company, St. Louis, MO) and 20 μl of genomic DNA in TE/10 buffer. The plate was covered with a rubber mat, vortexed, and spun briefly in a centrifuge to collect the liquid in the bottom of the wells. The plate was put into a PTC-100TM Programmable Thermal Controller (MJ Research, Inc., Waltham, MA) and run using the following program: 35 ×  (94°C, 1 min; 50°C, 55 s; 72°C, 2 min), followed by 94°C, 1 min; 50°C, 55 s; 72°C, 10 min (final extension step). After amplification, all PCR products were visualized by agarose gel electrophoresis. If bands of varying sizes were observed, agarose gel purification of the appropriate sized bands was performed. If a single band of the appropriate size was observed, the PCR products were transferred to a MultiScreen plate (MANU03050, Millipore). Approximately 50 μl of dH20 were added to each well and PCR products were dried under vacuum. An additional wash of 100 μl of dH20 and drying followed. Purified DNA was solubilized in 50 μl of dH20 and the plate was placed on a shaker for 20 min. DNA was then transferred to an unused PCR plate, which was covered with a foil seal and placed in the freezer until cycle sequencing.

Table 2 Oligonucleotide primers used in this study for sequencing the isoepoxydon dehydrogenase gene
Full size table

Nucleotide sequencing

Sequencing of both strands of PCR products was performed using an Applied Biosystems MicroAmp 96-well plate and the BigDye Terminator Cycle Sequencing Kit (Applied Biosystems, Foster City, CA). To each well was added: 3 μl of dH2O, 2 μl of BigDye, 2 μl of Primer (1 pmol/μl) and 3 μl of DNA. The plate was covered with a wax film, vortexed and spun. The thermal cycler program had 35 cycles, consisting of a denaturation step at 94°C for 30 s, an annealing step at 50°C for 15 s, and an elongation step at 60°C for 4 min. The cycling reaction product was purified and prepared for sequencing. To each well was added: 34 μl of 100% ethanol and 4 μl of 10 M ammonium acetate; contents of the wells were mixed with a pipette tip. The plate was covered with a rubber mat and placed on a shaker for 20 min. The plate was centrifuged at 3,600 rpm for 45 min. The rubber mat was removed and the plate was inverted on a paper towel; the plate was then pulse spun. To each well was added 150 μl of 75% ethanol, the rubber mat was replaced and the plate was centrifuged for 10 min. After centrifuging, the plate was allowed to drain over a paper towel and then pulse spun. The plate was dried in a swinging bucket vacuum centrifuge for 20–30 min. Ten microliters of formamide (Hi-Di) were added to each well for sequencing. Samples were sequenced on an ABI 3730 DNA Analyzer or ABI 3100 DNA Analyzer. Sequence analysis was conducted using Sequencher software v4.2 (Gene Codes Corporation).

Sequence analysis

Phylogenetic relationships were determined from rDNA sequences and idh sequences using the maximum parsimony (MP) program of PAUP* 4.0 (Swofford, 1998) with heuristic searches. Further comparisons were made with the neighbor-joining (NJ) program included in PAUP* 4.0 using the Kimura 2-parameter distance correction. Confidence limits for phylogenetic trees were estimated from bootstrap analyses (1,000 replications).

Results

The nucleotide sequences of the idh gene (either full length or partial idh sequence) for all 28 strains of the Penicillium species were obtained and translated to determine the amino acid sequences. Following removal of the two introns present in the idh sequence, amino acid sequences were compared (Fig. 1). Additionally, ribosomal DNA (ITS1, 5.8 S, ITS2 and partial sequence of 28S rDNA) for all 28 isolates was also sequenced and compared to verify that species had been correctly identified. The idh sequence of P. griseofulvum NRRL 2159A was used as a reference because it had been the first idh sequence deposited in GenBank (accession AF006680). It was also used to align the idh sequences of the Penicillium species and to determine the location of the two introns (present at nucleotides 2,086–2,139 and 2,383–2,435 for the idh sequence in GenBank). In Fig. 1, the nine conserved amino acids in the IDH sequence described by Fedeshko (1992) are indicated in bold; eight of the nine conserved amino acids were identical in all of the Penicillium species examined in this study. The ninth conserved amino acid, lysine (a basic amino acid) at position 46, was present in all strains of P. griseofulvum and P. dipodomyicola examined, but lysine was replaced by threonine (a polar amino acid) in all of the other Penicillium species. Amino acids present at other positions for many of the Penicillium species had common substitutions: glutamine substituted for glutamic acid (position 31); tyrosine substituted for phenylalanine (position 48); aspartic acid substituted for glutamic acid (position 50); tyrosine substituted for histidine (position 52); serine or proline substituted for alanine (position 53); alanine substituted for proline (position 56); leucine substituted for isoleucine (position 86); lysine substituted for glutamine (position 88); lysine substituted for arginine (position 125); threonine substituted for isoleucine (position 204); and isoleucine substituted for valine (position 216). Additionally, there were certain substitutions not present in any of the other Penicillium species that were shared by all of the strains of P. expansum; this included serine substituted for alanine (position 85), valine substituted for isoleucine (position 191), and asparagine substituted for alanine (position 229) (Fig. 1).

Fig. 1
figure 1figure 1figure 1

Multiple sequence alignment of deduced amino acid sequences of the idh sequences from P. griseofulvum (Pgri), P. expansum (Pexp), P. clavigerum (Pcla), P. carneum (Pcar), P. concentricum (Pcon), P. coprobium (Pcop), P. dipodomyicola (Pdip), P. gladioli (Pgld), P. glandicola (Pgln), P. paneum (Ppan), P. sclerotigenum (Pscl) and P. vulpinum (Pvul). Identical sequences are indicated with a dot, except where indicated by an alternative amino acid residue. Partial and complete isoepoxydon dehydrogenase sequences are given. Conserved amino acids (Fedeshko 1992) are shown in bold. Underlined residues are basic amino acids and are found only in NADP+ specific dehydrogenases (Scrutton et al. 1990)

Full size image

Phylogenetic trees showing placement of strains of the Penicillium species are shown for the idh sequences (Fig. 2) and for the rDNA sequences (Fig. 3). The nucleotide sequences of the idh gene and the rDNA genes of Byssochlamys nivea (NRRL 32565T) (Dombrink-Kurtzman and Engberg 2006) were used as outliers for the respective trees. B. nivea was chosen as an outlier because it is capable of producing patulin and it is not a member of Penicillium subgenus Penicillium. When two or more strains of a given species were analyzed, the strains usually had identical nucleotide sequences. Species relationships on phylogenetic trees determined from the idh sequences (Fig. 2) and from the rDNA sequences (Fig. 3) were similar on branches with strong bootstrap support. In addition, results were nearly identical whether analyzed from maximum parsimony or neighbor-joining analysis (data not shown).

Fig. 2
figure 2

Phylogenetic tree showing placement of Penicillium species as represented by 1 of 4 most parsimonious trees derived from maximum parsimony analysis of the idh gene with introns 1 and 2 removed. Tree length = 683, consistency index (CI) = 0.638, retention index (RI) = 0.786, rescaled consistency index (RC) = 0.502, homoplasy index (HI) = 0.362. Branch lengths are indicated by the marker bar, and numbers at nodes are bootstrap values determined from 1,000 replications. Bootstrap values below 50% are not displayed. Byssochlamys nivea NRRL 32565T was chosen as the outgroup species to root the tree

Full size image
Fig. 3
figure 3

Phylogenetic tree showing placement of Penicillium species as represented by 1 of 18 most parsimonious trees derived from maximum parsimony analysis of rDNA (ITS1, 5.8 S, ITS2 and partial sequences of 28S rDNA). Tree length = 157, consistency index (CI) = 0.815, retention index (RI) = 0.809, rescaled consistency index (RC) = 0.660, homoplasy index (HI) = 0.185. Branch lengths are indicated by the marker bar, and numbers at nodes are bootstrap values determined from 1,000 replications. Bootstrap values below 50% are not displayed. Byssochlamys nivea NRRL 32565T was chosen as the outgroup species to root the tree

Full size image

Discussion

The present study was focused on Penicillium species capable of producing the mycotoxin patulin. Research by Frisvad and collaborators was used as a resource for information on the production of secondary metabolites (extrolites) by species of Penicillium and has been essential for understanding which species produce patulin (Frisvad and Thrane 1987; Frisvad et al. 1987; Frisvad and Filtenborg 1989; Frisvad and Sampson 2004). Patulin production, determined under laboratory conditions, ranged from 65.72 to 294.05 μg/ml for strains of P. expansum and from 812.3 to 1,927.65 μg/ml for strains of P. griseofulvum (Dombrink-Kurtzman and Blackburn 2005). Surprisingly, P. griseofulvum produced approximately 10-fold more patulin than strains of P. expansum, the species most cited as the primary fungus associated with patulin contamination of fruits, primarily apples (Dombrink-Kurtzman and Blackburn 2005).

Comparison of the amino acids sequences of all P. griseofulvum and P. dipodomyicola strains examined in the present study revealed them to be identical (Fig. 1), although some nucleotide differences existed at the third position coding for an amino acid. P. griseofulvum and P. dipodomyicola are closely related and are commonly found on dry cereals and seeds (Frisvad et al. 2004). P. griseofulvum and P. dipodomyicola were also the only Penicillium species examined to have the conserved amino acid lysine present at position 46 in the IDH protein. In all of the other species, the conserved lysine is replaced by threonine. The strains having lysine at position 46 produced the most patulin of the strains examined. The significance of this amino acid substitution relates to the inability of the enzyme IDH to function optimally if the conserved amino acid (lysine) is not present to participate in binding the cofactor NADP+. The conserved basic amino acid lysine is thought to be involved in stabilizing the extra negative phosphate charge on the NADP+ molecule in dehydrogenases requiring the cofactor NADP+ as a reducing coenzyme (Scrutton et al. 1990). Replacement of a conserved amino acid by another amino acid can have different effects in the resulting protein (isoepoxydon dehydrogenase), depending upon whether the substitution involves a similar or different type of amino acid. If the type of amino acid substitution is radically different, the conformation of the protein may be affected, leading to a perturbation of the active site of the enzyme and possibly lower functionality of the resulting protein.

Alignments of introns 1 and 2 were done for all 28 of the Penicillium strains; phylogenetic trees were determined for introns 1 and 2 (data not shown) and analyzed by maximum parsimony. The trees of introns 1 and 2 were congruent with trees based on the idh nucleotide sequences (Fig. 2), suggesting that the introns and exons had evolved in tandem. Intron 1 contained 50–69 nucleotides. The largest number of nucleotides was present in P. coprobium, due to the presence of a stretch of adenine nucleotides. Intron 2 contained nucleotides ranging from 49 to 54.

Interestingly, the respective nucleotide sequence for introns 1 and 2 (idh nucleotide sequences for all strains have been deposited in GenBank) were highly conserved for strains of a particular species. When analyzed by maximum parsimony, the idh (Fig. 2) and rDNA (Fig. 3) sequences were also congruent. P. griseofulvum and P. dipodomyicola, which are both members of the series Urticicolae, were found to be grouped together for both the idh and rDNA sequences. This would suggest that for the different Penicillium species examined in this study, lateral gene transfer had not occurred among the different strains.

P. carneum and P. paneum can occur in very acidic products, such as silage (Frisvad and Samson 2004). Results of the maximum parsimony analyses in the present study indicated that P. carneum and P. paneum were closely related (Figs. 2 and 3). Although P. carneum and P. paneum have been reported to be indistinguishable using traditional morphological and physiological characteristics, they have a 12 nucleotide difference in the ITS regions (Boysen et al. 1996); results, such as these, have led to the conclusion that molecular characterization is essential for correct identification of fungal strains.

It is anticipated that the information gained from this study of idh sequences of Penicillium species with the ability to produce patulin will be beneficial for the design and use of molecular probes for identification of these mycotoxigenic fungi. Work in progress has identified unique species-specific sequences present in intron 2 for the design of 5′ primers. When utilized with specific 3′ primers, this will allow for generation of species-specific amplicons for rapid, accurate identification of patulin-producing species of Penicillium.

Nucleotide sequence accession numbers

The nucleotide sequences for the idh sequences determined in this study have been deposited with GenBank and have been assigned the following accession numbers: P. carneum NRRL 25168 (DQ343637) and NRRL 25170 (DQ343638); P. clavigerum NRRL 1003 (AY885572) and NRRL 1004 (AY885571); P. concentricum NRRL 2034 (DQ343630); P. coprobium NRRL 13626 (DQ343633); P. dipodomyicola NRRL 35582 (DQ343643) and NRRL 35583 (DQ343644); P. expansum NRRL 32293 (DQ343640), NRRL 32289 (DQ343639) and NRRL 35259 (DQ343642); P. gladioli NRRL 938 (DQ343625) and NRRL 939 (DQ343626); P. glandicola NRRL 985 (DQ343627) and NRRL 2036 (DQ343631) (an extra guanine at position 83 was believed to be a sequencing artifact and was not included in the submission to GenBank); P. griseofulvum NRRL 35258 (DQ343641); P. paneum NRRL 25159 (DQ343635) and NRRL 25162 (DQ343636); P. sclerotigenum NRRL 3461 (DQ343632) and NRRL 22813 (DQ343634); and P. vulpinum NRRL 1002 (DQ343628) and NRRL 2031 (DQ343629). Nucleotide sequences for the idh gene that were previously deposited in GenBank and included here for comparisons are P. griseofulvum NRRL 3523 (AY885565.1), NRRL 5256 (AY885566.1) and NRRL 2159A (AY885567.1); and P. expansum NRRL 2304 (AY885568), NRRL 6069 (AY885570) and NRRL 35231 (AY885569.1) (Dombrink-Kurtzman 2006).

The nucleotide sequences for the ITS and 28S rDNA domains have been deposited in GenBank and have been assigned the following GenBank numbers: P. carneum NRRL 25168 (DQ339564) and NRRL 25170 (DQ339566); P. clavigerum NRRL 1003 (DQ339555) and NRRL 1004 (DQ339560); P. concentricum NRRL 2034 (DQ339561); P. coprobium NRRL 13626 (DQ339559); P. dipodomyicola NRRL 35582 (DQ339550) and NRRL 35583 (DQ339570); P. expansum NRRL 2304 (DQ339556), NRRL 6069 (DQ339562), NRRL 32289 (DQ339547), NRRL 32293 (DQ339552), NRRL 35231 (DQ339558) and NR RL 35259 (DQ339548); P. gladioli NRRL 938 (DQ339563) and NRRL 939 (DQ339568); P. glandicola NRRL 985 (DQ339573) and NRRL 2036 (DQ339565); P. griseofulvum NRRL 3523 (DQ339549), NRRL 5256 (DQ339557), NRRL 2159A (DQ339551) and NRRL 35258 (DQ339553); P. paneum NRRL 25159 (DQ339554) and NRRL 25162 (DQ339571); P. sclerotigenum NRRL 3461 (DQ339567) and NRRL 22813 (DQ339574); and P. vulpinum NRRL 1002 (DQ339572) and NRRL 2031 (DQ339569).