Expanding the Species and Chemical Diversity of Penicillium Section Cinnamopurpurea

A set of isolates very similar to or potentially conspecific with an unidentified Penicillium isolate NRRL 735, was assembled using a BLAST search of ITS similarity among described (GenBank) and undescribed Penicillium isolates in our laboratories. DNA was amplified from six loci of the assembled isolates and sequenced. Two species in section Cinnamopurpurea are self-compatible sexual species, but the asexual species had polymorphic loci suggestive of sexual reproduction and variation in conidium size suggestive of ploidy level differences typical of heterothallism. Accordingly we use genealogical concordance analysis, a technique valid only in heterothallic organisms, for putatively asexual species. Seven new species were revealed in the analysis and are described here. Extrolite analysis showed that two of the new species, P. colei and P. monsserratidens produce the mycotoxin citreoviridin that has demonstrated pharmacological activity against human lung tumors. These isolates could provide leads in pharmaceutical research.


Introduction
Penicillium is a mold genus widely known for production of the antibiotic penicillin by some species, ripening of camembert and blue cheeses by others, and the production of damaging mycotoxins in feeds, forage and foods by yet other species [1][2][3]. Monographic treatments of Penicillium based on morphology and physiology [4] resulted in the splitting of Penicillium into four subgenera, and subsequent phylogenetic study justified moving subgenus Biverticillium species into Talaromyces [5,6]. The genus is also a source for compounds that possess therapeutic activities [7].
Nomenclatural rules of the past have allowed dual naming where the sexual and asexual morphs of a single species were placed in different genera. A new nomenclature [8] rescinds the prior rule and requires a single name for a single species. We follow the examples of others supplemented with 5% NaCl (CYAS). Specific color names are from the Ridgway color guide [33] and are indicated by a parenthetical R and plate number. Microscopy, microphotography and macrophotography were as described [30] using acid fuchsin dye. Photographs were resized and modified for contrast using Adobe Photoshop Elements 10 [34].

DNA extraction sequencing and analysis
Biomass for DNA extraction was grown, DNA was isolated and purified, and loci were amplified and sequenced as described [30]. Sequenced loci were beta tubulin (BT2), calmodulin (CF), nuclear internal transcribed spacer (ITS), minichromosome maintenance factor 7 (Mcm7), DNA dependent RNA polymerase II subunit (RPB2), and ribosome biogenesis protein (Tsr1). Sequences were deposited in GenBank and the accession numbers are listed (S2 Table). An instance of BIGSdb [35] was implemented and populated with sequences from this study and associated GenBank sequences for public fungal identification searches [28].
Maximum likelihood analysis was performed using MEGA ver. 6.06 [36]. Sequences were aligned using MUSCLE, most appropriate model was tested, maximum likelihood was calculated and 1000 bootstrap replicates were conducted using the built-in functions of MEGA. Aligned data files in MEGA format are accessible as supporting information (S1-S6 Datafiles). Tree diagrams were opened in and annotated using CorelDraw X6 [37].

Extrolite detection
Citreoviridin and other metabolites were induced and detected using described methods [38].

Nomenclature
The electronic version of this article in Portable Document Format (PDF) in a work with an ISSN or ISBN will represent a published work according to the International Code of Nomenclature for algae, fungi, and plants, and hence the new names contained in the electronic publication of a PLOS ONE article are effectively published under that Code from the electronic edition alone, so there is no longer any need to provide printed copies.
In addition, new names contained in this work have been submitted to MycoBank from where they will be made available to the Global Names Index. The unique MycoBank number can be resolved and the associated information viewed through any standard web browser by appending the MycoBank number contained in this publication to the prefix http://www. mycobank.org/MycoTaxo.aspx?Link=T&Rec=. The online version of this work is archived and available from the following digital repositories: PubMed Central, LOCKSS.

Results
The datasets examined contained, by locus, BT2, 489 aligned coding and non-coding characters, best model K2+G; CF, 741 aligned coding and non-coding characters, best model K2+G; ITS, 533 aligned characters from ITS1, ITS2 and 5.8S rDNA, best model JC+I; Mcm7, 616 aligned coding characters, best model K2+I; RPB2, 1014 aligned coding characters best model K2+G; and Tsr1, 816 coding characters, best model T92+G. In each of the single locus trees (S1 Fig.) isolates formed terminal groups with strong statistical support. Penicillium cvjetkovicii, P. lemhiflumine and P. fluviserpens are strongly supported at 6 of 6 loci, P. monsgalena, P. monsserratidens and P. idahoense are supported by 5 of the 6 loci, and P. colei and P. salmoniflumine are supported by 4 of the 6 loci. Some species and terminal groups had very low bootstrap support at the ITS locus, but the ITS region is an effective barcode for these species, with the five protein coding loci also useful for genotypic species recognition. Mcm7 and RPB2 loci display the lowest intraspecific variation and would be most suitable for diagnostic purposes. The datasets were concatenated and analyzed providing a fully resolved view of the group (Fig. 1).
GenBank databases were searched prior to writing (accessed 12 July 2014) for additional isolates having sequences fitting with the newly described species. While all six loci were used for BLAST searches, the ITS region returned the most sequences and the most diverse sequences. Those sequences were downloaded, aligned with the study-group sequences and a tree was generated ( Fig. 2; conditions as for ITS data above). Two unidentified endophytes from coffee plants in Colombia [39] were identified as P. fluviserpens. An isolate from an apple in South Africa [40], a cheese contaminant from Spain [41], both previously identified as In the bootstrap, all nodes were above 95%, except the P. fluviserpens subclade NRRL 35844, NRRL 58649 77%; P. cinnamopurpureum subclade NRRL 35501-NRRL 35502 55%; and P. cinnamopurpureum subclade NRRL 162-NRRL 3118-NRRL 3326-NRRL 35503 64%). Outgroup chosen based on prior work [27]. P. ellipsoideosporum is represented only by sequences obtained from GenBank. P. chermesinum, and an unidentified decay mold of Ophiocordyceps sinensis in China [42] were all identified as P. cvjetkovicii. A sequence derived from an uncultured DNA clone from forest soil in Switzerland GB# KC818302 appears to represent an undescribed species from Penicillium section Cinnamopurpurea. Phylogenetic tree based on ITS region sequences, with very similar sequences from GenBank. Unidentified endophytes from coffee plants in South America are P. fluviserpens; a cheese isolate from Europe, an apple core isolate from South Africa, and a fungal epiphyte from China are identified as P. cvjetkovicii although two were initially identified as P. chermesinum using phenotypic identification; and one sequence amplified from soil DNA represents an apparent unknown species. Bootstrap values above 70% are placed on the branches. The haplotypes of each isolate at each locus were compared and arbitrarily assigned letter designations (Fig. 3). In the figure, shading is used to represent groups of isolates with >90% bootstrap support in the multilocus tree (Fig. 1). Haplotypes are shared at some loci between subclades of species, but there are no shared haplotypes between species. The homothallic sexually reproducing species P. cinnamopurpureum has a pattern of haplotypes among the isolates similar to the patterns of the putatively clonal species P. cvjetkovicii and P. fluviserpens. Sharing haplotypes between strongly supported groups could result from incomplete lineage sorting [43] or from hyphal fusion of different genotypes in a homothallic species. Other species have only one or two representatives and patterns cannot be discerned.
Phenotypic data showed that these species grow slowly, rarely attaining a colony diameter exceeding 30 mm in 14d. Colony reverse colors and soluble pigments are consistent within species on particular media but as can be seen from the figures, the expression of these colored pigments is strongly influence by the growth medium. Other factors may influence expression because the colors do not always develop in repeated cultures of an isolate. The species produce mostly small (2-4 μm) conidia that vary from ellipsoidal to spherical with surfaces from smooth to slightly rough and a small number of much larger conidia. Conidiophores are furcate in this clade but P. salmoniflumine and P. monsgalena produce mainly unbranched penicilli. The presence of exudates in the species is variable with the medium and also can be variable among isolates of a single species. Character suites appear sufficiently stable and distinct for morphological diagnostic of the species. Diagnoses are listed with the descriptions of the species in the Taxonomy section.
Detailed extrolite data are listed with each isolate in the Taxonomy section. The in-group clade containing all the new species and P. idahoense have producers of the extrolites called OTO1-OTO6. These metabolites have UV spectra with absorptions at 218 nm (100%), 250 nm (35%) and 324 nm (23%), and retention indices (RI) at 746, 961, 824, 818, 748 and 846, respectively. Isolates in the sub-clade containing P. colei and P. monsserratidens produce citreoviridin and citreomontanin, whereas the sister group to these two, P. idahoense does not produce citreoviridins. One outgroup isolate P. malacaense NRRL 35754 does not produce any of the metabolites produced in the main clade. Several species in the main clade produce the metabolites named PR-828 (RI 828) and PR-878 (RI 878). The UV spectra of these extrolites resemble PR-toxin, sporogen AO1 and petasol, the latter two compounds earlier found in Penicillium sp. JP3430 [44]. These compounds have UV absorption at 251 nm, but mass spectrometric analysis is needed to identify these unknowns. The anthraquinone AQ-840 (RI 840) has approximately the same UV spectrum as carviolin (= roseopurpurin). Carviolin has been found in P. roseopurpureum and P. dravuni previously [45,46]. The extrolite alk-752 (RI 752) had an indole chromophore and was only found in P. lemhiflumine and P. fluviserpens. The profiles of extrolites found in the new species and P. idahoense describe a closely related group of species that are also similar phenotypically.
The metabolite AQ-840 is common to all of the in-group isolates. Other metabolites such as alk-752 may be produced by species of disparate clades (e.g., P. fluviserpens and P. lemhiflumine) or by a clade of species (e.g., metabolite OTO1 present in the clade composed of P. salmoniflumine, P. monsgalena and P. lemhiflumine). Citreoviridin is produced in two of the new species, P. colei and P. monsserratidens that along with P. ellipsoideosporum form a clade in all of the single locus trees (S1 Fig.). Extrolite expression or detection in the isolates has little intra-specific variability (e.g., P. fluviserpens and P. cvjetkovicii, Taxonomy section).

Discussion
Maiden and associates [47] introduced the technique of multilocus sequence typing (MLST) to bacterial systematics. A general theory of how to interpret the multilocus data for the definition of bacterial species is so far lacking. Practical frameworks for species limit designations are based on the amount of sequence difference observed in either typological species or species defined by the 70% DNA-DNA hybridization technique [48]. We provide an identification web site to go along with our phylogenetic taxonomy [28]. Genealogical concordance analysis [20][21][22] is effective in determining the limits of sexually reproducing species. Several factors were considered before applying this technique to this group of primarily anamorphic fungal species. All of the species in this clade are descended from the most recent common ancestor of P. cinnamopurpureum a species known to produce sexually derived ascospores and the ingroup species. One of the in-group species, P. idahoense, makes meiotically produced ascospores. Penicillium colei, P. cvjetkovicii, P. idahoense, P. lemhiflumine, P. monsgalena, P. monsserratidens and P. salmoniflumine all produce a certain proportion of conidia that appear to be twice the size of the commonly observed small conidia. Conidium size has often been associated with ploidy level in fungi [49][50][51][52] and the distinct conidium sizes seen in these species suggests hyphal fusion and ploidy changes, all consistent with processes of heterothallic meiotic recombination. The polymorphisms at the different loci for the in-group are consistent with the polymorphism pattern of the out-group teleomorphic P. cinnamopurpureum. It is also notable that a number of Aspergillus and Penicillium species long thought to be strictly anamorphic have been shown in recent years to possess cryptic sexual states [53][54][55][56][57]. We hypothesize that our in-group species may also have cryptic sexual states.
Interspecific DNA sequence variation between sibling species is present at each locus along with limited intraspecific variation at some, and accurate species recognition by genotype relies on a database of existing species and accurate DNA sequencing of the unknown. The ITS region for barcoding fungi [24] can recognize all species in this group but some species differ by a single base at the ITS locus making absolute accuracy in sequencing essential. The other loci all exhibit larger interspecific differences and are more tolerant of varying sequencing accuracy. Some investigators have used sequences from three loci in their studies, most often ITS, betatubulin and calmodulin (e.g., [58,59]) are chosen. The currently known species could confidently be identified using these three loci in an MLST scheme. The additional loci used in this study are essential for defining the genetic limits of the species using genealogical concordance analysis.
Phenotypic recognition of Penicillium species is complicated by a number of seemingly minor variations in culture conditions and isolate handling that can have large effects on the development of the characters used in phenotypic recognition of the species [60]. Aeration, depth of media in petri plates, size of inoculum, source of the yeast extract and other conditions can all have major impact on the phenotype of a culture [61,62]. Efforts has been made to standardize growth conditions [63]. Certain aspects of culture handling such as initial single spore initiation of cultures [64] are essential for all methods of species identification.
Morphologically the species in section Cinnamopurpurea are quite similar, all producing subglobose to ellipsoidal smooth to finely roughened spores, monoverticillate to divaricate biverticillate smooth-walled conidiophores and quite slow-growing colonies, often with a brown reverse on some media. P. ellipsoideosporum differs from the other species by having longer and more divaricate metulae [65] and P. shennongjianum differs by having a higher proportion of conidiophores with three metulae [66]. The conidiophores of the latter two species are not vesiculate like most other species in the section.
Extrolite diagnosis of species like morphological recognition relies mostly on the predictable expression of the specific metabolites under standardized conditions. Each of these species has metabolites that can be used to identify the species.
Citreoviridin is a human lung tumor inhibitor [67] that may be of interest in finding new species that potentially produce other derivatives of citreoviridin in drug lead searches. Some derivatives of citreoviridin are citreomontanin [68], citreoviridinol and secocitreoviridin [69], neocitreoviridinol and epineocitreoviridinol [70], epiisocitreoviridinol [71], citreoviripyrone A and B [72], citreoviridin C and D [73] and herbarin A and B [74]. Citreoviridin production is not unique to section Cinnamopurpurea. We list all known citreoviridin producers in Table 1.
Two new species, P. colei and P. monsserratidens produce citreoviridin. The species are sibling and produce several other extrolites in common. They are separated because there is very strong statistical support for the concordance analysis that shows NRRL 13013 is neither conspecific with P. monsserratidens nor with the unstudied but related species P. ellipsoideosporum. Wicklow [75] using phenotypic species recognition thought it most likely that NRRL 13013 was a variant of P. citreonigrum in contrast to placement of the isolate in P. charlesii by Cole et al. [17]. Phylogenetic analysis shows that NRRL 13013 is not an isolate of P. citreonigrum, and also shows that it is distinct from its sibling species P. monsserratidens and P. ellipsoideosporum. Naming a species on the basis of a single isolate necessarily cannot describe all the variation (phenotypic or genetic) present in the species. NRRL 13013 produces citreoviridin and its identification has been subject to different interpretation as noted above. Here we resolve its identification by providing the new name P. colei to contain this isolate and establish that the most closely related previously described species is Penicillium idahoense, not P. citreonigrum. Penicillium fluviserpens isolates form two strongly supported subclades (Fig. 1), but were very similar phenotypically, have no outstanding characteristics or metabolites that would argue for distinct names and share some haplotypes, so we retain them in a single species. Penicillium cvjetkovicii isolates occur on a very strongly supported branch (Fig. 1) and share production of the extrolite AQ-840, but the individual isolates all produce additional different extrolites. NRRL 735 is an isolate obtained by Charles Thom from Biourge in 1924 as P. griseo-roseum and has been maintained in culture since. It differs from the other P. cvjetkovicii isolates by producing white, largely non-sporulating colonies. Peterson and Horn [27] noted the unusual growth of this isolate as others have [1,4] and referred to it as Penicillium sp. until this study showed it to be P. cvjetkovicii. Other isolates of P. cvjetkovicii are quite consistent in appearance so the caution showed earlier [1,4,27] was justified. Penicillium salmoniflumine is A. terreus A. terreus (not all strains) [73] Eupenicillium ochrosalmoneum Penicillium ochrosalmoneum [16] Cladosporium herbarum Questionable record, should be confirmed [74] Penicillium aurantiacobrunneum P. aurantiacobrunneum [59] P. cairnsense P. cairnsense [59] P. charlesii P. colei [16,17]  represented by two isolates that produce the extrolite OTO1. NRRL 58001 also makes the unidentified metabolite endphe-1130. Penicillium salmoniflumine isolates are on a strongly supported branch and as a monophyletic group with distinguishing characters merits formal naming. The most problematic new species are P. monsgalena and P. lemhiflumine, each represented by a single isolate. Penicillium monsgalena NRRL 22302 (Fig. 1) could be grouped with P. salmoniflumine but does not share the common phenotype or genotype used to describe P. salmoniflumine. Production of citreoviridin metabolites is not unique to the P. idahoense clade or even to the monoverticillate Penicillium species (Table 1). In Penicillium section Citrina the species P. aurantioacobrunneum, P. citreonigrum, P. gallaicum, P. godlewskii, P. neomiczynskii, P. quebecense and P. vancouverense [59] are producers. Citreoviridin is also produced by Aspergillus terreus [73], A. aureoterreus, A. neoafricanus, A. pseudoterreus, and A. neoniveus [76]. Penicillium pulvillorum was reported to produce citreoviridin [77], but the identification of the producer was later corrected to P. manginii [4,78]. Reported citreoviridin production by Cladosporium herbarum [74] needs to be confirmed. Advanced chemical analytical techniques and genealogical concordance analysis of multilocus DNA sequence data allows confident assessment of the species that produce citreoviridin, and we speculate that the culture of Cladosporium herbarum was contaminated by a Penicillium or Aspergillus culture (see Table 1 for known producers) As noted above, citreoviridin contamination of foods is rarely reported and is not a major issue in food hygiene, however P. ochrosalmoneum can produce citreoviridin in maize, and P. citreonigrum can produce citreoviridin in rice, and has been claimed to induce Beri-beri disease [3,13,14-17, 19, 83-84, 86]. Of the species described here, P. colei was found on molded pecan fragments, and thus citreoviridin may occur in pecans. Generally it is of course not recommended to eat moldy nuts, and such nuts will probably be discarded by the producers during quality control of commercial nuts. P. monsserratidens has only been found in air samples, but we doubt it would give a mycotoxin problem via inhaling spores of this species, as it appears to be relatively infrequent in indoor air. CYA yellow green (Vetiver green R47) in sporulating area with pale yellow margins, velutinous, moderately sulcate or wrinkled, mycelium white, occasionally yellowish, inconspicuous, exudate brownish orange, soluble pigments light-brown, no sclerotia, sporulation heavy, reverse dark brown nearly black centrally (carob brown R14), marginal area buff to nearly mouse gray (R51). MEA velutinous with central button 1-2 mm high, gray-blue-green (Castor gray R52), mycelium white to buff becoming pale yellow at the marginal, sporulation abundant, no exudate, no soluble pigment, no sclerotia, reverse brown-yellow centrally (ochraceous tawny R15) and pale yellow marginally (antimony yellow R15).
P. colei is distinguished from similar species by reverse color carob brown centrally to mouse gray marginally on CYA after 14 d. The closest macro-morphological match is with P. monsserratidens. P. colei produces exudate on CYA and PDA in the center of colony while P. monsserratidens produces an abundance of exudate over the entire colony, mostly peripherally. Also P. colei does not produce exudate or soluble pigment on MEA while P. monsserratidens produces clear exudate and reddish yellow soluble pigment. Micro-morphological difference is that P. colei produces ellipsoidal to spherical 2.5-4 (-11) μm conidia, in contrast to P. monsserratidens that produces spherical to ellipsoidal conidia 2.5-3 (-7) μm.
Penicillium cvjetkovicii S. W. Peterson   Notes: One of us (ZJ) has commonly isolated this species from hospital environments in parts of the US. Despite this species' isolation locales, it does not grow at body temperature. It is distinguished from similar species by producing faint red soluble pigment on MEA and CY20S to very strong, vinaceous to reddish-brown soluble pigments on CYA, PDA and CY5S. Macro-morphological appearance is unique. Growth on MEA with a central button, conidia dark bluish-gray-green, with abundant sporulation that looks almost powdery. At the margins it sporulates in 'veins". Mycelium at the margins subsurface or submerged into the media.
Notes: Penicillium lemhiflumine is distinguished from similar species (P. cvjetkovicii, P. salmoniflumine, P. monsgalena and P. idahoense) that form sulcate colonies when growing on MEA. Penicillium cvjetkovicii growth on MEA occurs with a central button, conidia dark bluish-gray-green, with abundant sporulation that looks almost powdery. At the margins it sporulates in "veins". Mycelium at the margins is subsurface or submerged in the media. In contrast the P. lemhiflumine sporulating area is gray green. Penicillium lemhiflumine grows twice as fast on CYA, MEA, OA, CY20S, PDA and CYAS media in contrast to P. salmoniflumine which has good growth only on CYAS. Penicillium monsgalena has a gray-green conidial area at the center of the colony, while P. lemhiflumine sporulates heavy over the entire colony and radial sulcation is very pronounced. Penicillium idahoense on CYA produces abundant brown to dark brown sclerotia, 200-300μm diameter, clear exudate, abundant; no soluble pigments; reverse purple to dark purplish brown. On MEA it produces brown sclerotia, 100-300 μm in diameter. In contrast, P. lemhiflumine produces pinkish soluble pigment on CYA and does not produce sclerotia on CYA or MEA.
An array of photographs and microphotographs of Penicillium idahoense is included here for comparative purposes (Fig. 11).