Penicillium roqueforti has a central role in the production of interior mould-ripened cheeses such as Gorgonzola, Roquefort, and Stilton where the fungus is critical for flavour and texture development through its enzymatic activity. In addition, asexual sporulation of the fungus in cavities of the cheese results in the characteristic blue-veined appearance, with commercial strains being sold partly on the basis of colour development. Variation in both spore colour and colony texture has been described for worldwide isolates from cheese and non-cheese substrates2,7. Despite the importance of colour in P. roqueforti, the genetic basis of spore pigmentation in this species has not previously been elucidated. In the present study we drew on knowledge of pigment development in other ascomycete fungi17–20 to identify a DHN-melanin biosynthetic pathway in P. roqueforti and used experimental approaches to confirm the functional activity of the pathway. Results also demonstrated the possibility to generate novel coloured strains via pathway disruption, of potential public and commercial appeal, which was further confirmed by the use of UV mutagenesis to induce a range of colour mutants.
The blue-green appearance of P. roqueforti conidia suggested that a DHN-melanin biosynthesis pathway like that described for A. fumigatus17,32 might be present. The availability of the P. roqueforti FM164 genome8 allowed BLAST analyses, which led to identification of a canonical DHN-melanin biosynthetic pathway in P. roqueforti, comprised of six genes (alb1, ayg1, arp2, arp1, abr1 and abr2) whose sequential enzyme activity has previously been shown to lead to synthesis of DHN-melanin13,17,20−22. The same set of genes were detected and sequenced from the 74–88 industrial isolate of P. roqueforti used for experimental work in the present study. These results are consistent with reports of the presence of DHN-melanin pathways in related Penicillium and Aspergillus species, which show evidence of genome clustering of pathway genes thought to be linked to pathway evolution and regulation18–20, 22,23. However, P. roqueforti FM164 is unique amongst so far reported penicilia in that the laccase gene abr2 is located on a separate super contig (HG792016.1) from the other five clustered pathway genes, whilst there was a relatively large distance (over 100 kb) between alb1 and the four other members of the main cluster. Such separation of DHN-melanin pathway genes has also been observed in Botrytis cinerea and Alternaria alternata although any functional significance is unclear27.
A combination of biochemical enzyme inhibition, heterologous gene expression and gene modification (GM) approaches were applied to evaluate whether the putative DHN-melanin biosynthetic pathway was indeed functional in P. roqueforti. First, known enzyme inhibitors of the DHN-melanin pathway were used to confirm gene function33,34. Addition of the Arp2 inhibitors tricyclazole and pyroquilon resulted in production of reddish-pink-brown colonies, consistent with arp2 being a key component in the DHN-melanin pathway of P. roqueforti.
Second, the first two putative pathways genes alb1 and ayg1 were expressed in an A. niger heterologous expression system16,28 to identify the metabolites produced. Transformants expressing the P. roqueforti alb1 gene were found to produce the heptaketide YWA1, and when alb1 was co-expressed with P. roqueforti ayg1 then 1,3,6,8-THN was produced. These results not only confirmed that the genes encode functional proteins, but also demonstrated that these key first stages of pigment biosynthesis in P. roqueforti follow the DHN-melanin biosynthesis pathway described for A. fumigatus17. It is noted that pigment formation could be attributed solely to expression of the P. roqueforti alb1 and ayg1 genes because the endogenous DHN-melanin pathway of A. niger is not expressed under the liquid growth conditions of the assay. It is also noted that heterologous expression of further elements of the pathway was considered problematic due to the requirement for multiple gene constructs and need for suitable cellular localization of later stages of the pathway, and was therefore not pursued.
Third, deletion cassettes were successfully integrated into the genome to individually remove all six putative DHN-melanin pathway genes. Gene deletion resulted in a dramatic change in conidial pigmentation for each step in the pathway. Removal of the alb1 gene produced white (albino) colonies with conidia that lacked any visible pigmentation, indicating lack of polyketide synthesis as described in other Aspergillus and Penicillium species17,21. Deletion of ayg1 resulted in yellowish-green colonies, most likely due to accumulation of the heptakide YWA1 and other alb1 intermediate products. Deletion of arp2 and arp1 each resulted in reddish-pink-brown colonies, likely due to the accumulation of 1,3,6,8-THN and scytalone, respectively. Deletion of abr1 and the final laccase abr2 gene both resulted in colonies with a brown colouration, likely linked to accumulation of vermelone or 1,8-DHN and other pathway intermediates, respectively15,17,21. Whilst results were consistent with previous reports and the heterologous expression studies, confirmation of the putative chemical intermediates will require further biochemical work. Indeed, deletion of genes in the DHN-melanin biosynthetic pathway can lead to several alternative pathway byproducts such as formed by auto-oxidation rather than enzyme action alone34. To further confirm gene function, complementation studies were undertaken with the polyketide synthase (alb1) and laccase (abr2) genes as representatives at the start and end of the DHN-melanin pathway. In both cases, rescue with the parental gene restored wild-type conidial pigmentation in the complemented strains.
It was realized that the ability to produce colour derivative strains of P. roqueforti via disruption of the pigment biosynthetic pathway provided the exciting possibility of producing new commercial strains of the fungus with alternative spore colours rather than just the traditional blue-green colouration. However, it was important to understand any possible impact of disruption of the DHN-melanin pathway on the physiology of such colour mutants. Therefore, studies were undertaken to assess the impact on mycotoxin levels and flavour volatile production in the various gene deletion strains.
Regarding mycotoxins, the impact on mycophenolic acid (MPA) and roquefortine C was investigated as two key secondary metabolites of safety concern25,26. There was no significant increase in MPA production between the 74–88 parent and the gene deletant strains. Perhaps surprisingly, a significant increase in roquefortine C production was observed in many of the gene deletion strains, in particular a ca. 3-fold increase in the Δayg1 strain. The reasons for this increase were unclear given that regulation of secondary metabolite pathways are complex. It has recently been shown in Alternaria alternata that the biosynthetic pathway for altertoxin production utilises many of the enzymes of the DHN-melanin pathway, indicating this pathway as a source for fungal toxin production27. Despite the increased roquefortine C levels observed in certain DHN-melanin gene deletants, it is important to note that levels still remained relatively low and of the same overall magnitude of the 74–88 production strain. Furthermore, production was induced by the presence of sucrose, which is absent during cheese production.
Regarding volatile production, the impact of gene deletion was investigated on 26 different flavour compounds associated with ripening and aroma development in blue cheese30,31. No significant change in the levels of ketones, principally associated with blue-cheese flavour, was seen in any of the deletion strains. By contrast, lower levels of certain esters and 2-alcohols were found in all the deletion strains. Results therefore suggest that DHN-melanin pathway mutants of P. roqueforti might have slightly altered aroma profiles if used in cheese production whilst retaining the key ‘blue-cheese’ flavour. However, it is cautioned that the present study used a model milk system for volatile production30,31 and results cannot be directly extrapolated to whole cheese production. Elsewhere there has been reported to be much variation in flavour volatile production between isolates of P. roqueforti12.
Encouraged by these results, we then investigated if it were possible to create pigment mutants of P. roqueforti by classical UV-mutagenesis given that it is not possible to utilize GM strains in commercial cheese manufacture under current food legislation, but UV-derivatives are allowed. Indeed, a UV-derived white-spored strain of P. roqueforti has previously been commercialised35,36. A series of colour mutants were successfully produced following UV irradiation of the 74–88 production or B20 and A22 ascospore-derived strains. For most colour mutants, those producing colonies with a particular non-standard colour phenotype were found to contain UV-induced mutations in the corresponding gene in the DHN-melanin pathway. For example, the reddish-brown colour mutant 74-88-1 had a mutation in the arp2 gene (T290A), the green mutant 74-88-5 had a mutation in the ayg1 gene (L258P), and the white (albino) mutant B20-1 had a mutation in the alb1 gene (L102S). However, interestingly the hue and depth of colour of the UV mutants sometimes differed slightly from the respective whole gene deletant strain. This suggested that some of the UV-mutated genes might still be producing a protein product with partial functionality, although the impact of UV-mutations elsewhere in the genome cannot be discounted. Furthermore, some colour mutants (e.g. greyish green and intense blue) were produced for which no mutation could be found in any of the coding regions of the DHN-melanin pathway genes. This indicates that elements in the promoter region and/or genetic elements other than the DHN-melanin pathway can mediate pigment production and colour in P. roqueforti. Thus, UV mutagenesis might enable the generation of a wider spectrum of colour strains than targeted DHN-melanin pathway gene deletion alone. It is noted that slightly different pathways for DHN-melanin synthesis have been reported for Aspergillus nidulans and Aspergillus flavus18 and that a non-canonical melanin biosynthetic pathway has been described from A. terreus16, as well as synthesis of pyomelanin in A. fumigatus32.
Certain colour derivative mutants were then used as starter cultures in cheese production. The arising cheeses indeed showed striking differences in the colour of veining of the cheese compared to the parental strain. Levels of mycotoxins produced in cheeses by colour derivatives were comparable or lower than those produced by the parental isolates, indicating likely safe use for food production although this finding will require confirmation with a larger number of strains. This was also consistent with results of mycotoxin production on sucrose-inducing media by representative UV-colour mutants where levels were comparable or below the 74–88 production strain, again indicating safe use for food production. It is noted that mycotoxins have previously been reported in blue cheeses, but the levels produced are not considered a health risk and mycotoxins may be unstable in the cheese matrix26,37. However, it is cautioned that P. roqueforti can produce additional mycotoxins which were not examined in the present study6,12.
Although P. roqueforti is best known for applications in cheese production, the species is a relatively common saprotrophic mould found on various natural substrates6. Therefore, the presence of the DHN-melanin pigment in the spore wall can be explained partly by the fact that melanin can provide protection against environmental UV radiation13,34,38. Indeed, we found that deletion of representative genes from the DHN-melanin biosynthetic pathway (Δalb1, Δayg1 and Δarp1) resulted in significantly lower spore survival than those of the parental 74–88 isolate when exposed to UV stress (Fig. S12), consistent with parallel findings on P. roqueforti melanin-deficient conidia exposed to UV radiation38. In addition, fungal melanins may confer protection against oxidative and temperature stress34 and are required as structural components for correct assembly of the fungal cell wall14.
To conclude, work presented in this study has revealed the principal pigment pathway used by P. roqueforti for production of DHN-melanin, involving a canonical set of six genes (alb1, ayg1, arp2, arp1, abr1 and abr2) most of which were clustered together in the genome. Disruption of pathway genes by gene deletion or UV mutagenesis resulted in strains producing spores with novel pigmentation. Such strains might be of public and commercial interest for manufacture of new ‘non-blue’ coloured mould-ripened cheese and represent a further step in the domestication of the species. Market acceptance will reveal the future of such products. UV studies indicated that additional genetic elements other than the DHN-melanin pathway alone can also mediate pigment production and colour in P. roqueforti. Future genome sequencing of such non-standard colour mutants of P. roqueforti is envisaged to provide insights into additional pigment pathways.
The present study also found that deletion of certain genes of the DHN-melanin pathway of P. roqueforti resulted in unexpected changes in both the production of the mycotoxin roquefortine C and also certain flavor volatile compounds. This warrants further work to investigate links between DHN-melanin biosynthesis and other cellular processes.