Penicillium Amapaense sp. nov. Exilicaulis Section and New Records of Penicillium Labradorum in Brazil Isolated from Amazon River Sediments with Potential Applications in Agriculture and Biotechnology

The genus Penicillium is recognized for its ability to produce bioactive molecules with a wide range of biotechnological applications. Currently, the genus is distributed in 28 sections, with more than 50 species in the Exilicaulis section. Representative species of this section are responsible for the production of antimicrobial compounds, but they can also produce genotoxic compounds that affect commercial mushroom production or cause disease in immunosuppressed animals. In the present study, based on morphological characters such as the length of the conidia, phialides and stipes, as well as sequence analysis of the ITS region and partial sequence of CAM, TUB2 and RPB2 loci, we describe a new fungal species denominated Penicillium amapaense and report for the rst time the occurrence of Penicillium labrodorum in Brazil, both of which were isolated from sediments of the Amazon River. The isolates obtained in this study for each species were submitted to antibiosis assays against 12 phytopathogenic fungi that affect important agricultural crops in Brazil and showed inhibition of 11 out of 12 of them. The production of amylase, cellulase and siderophore as well phosphate solubilization was also detected. Metabolomic analysis indicates the ability of P. labrodorum and Penicillium amapaense sp nov. to produce polyketides such as known curvularins and anthraquinones. In addition to these, unknown compounds were also detected. These results indicate the biotechnological and agricultural potential of P. labradorum and P. amapaense.


Introduction
Species of the genus Penicillium (Ascomycota; Trichocomaceae) are present in different environments such as soils, water and air, and can be found living in plants as endophytes, while some species can also be plant pathogens and even cause disease in humans and other animals (Yadav et al. 2018). Penicillium is recognized for its wide range of biotechnological applications, due to the production of several secondary metabolites with antimicrobial, antioxidant, neuroprotective, insecticide, cytotoxic, anti- The genus Penicillium was rst described by Link et al. (1809) and has since undergone several taxonomic revisions based on morphology (Thom 1930;Raper andThom 1949: Pitt 1979;Ramírez 1982). A taxonomic review, based on ITS (Internal transcribed spacer) data, began in the 1990s with revisions by Pitt and Samson (1993). In 2011, a review based on the evaluation of its morphological characters and on multilocus phylogenetic analysis, using data from partial gene sequences: RPB1 and RPB2 (subunits of the catalytic core of RNA polymerase II), TSR1 (ribosome biogenesis protein) and CCT8 (component of the chaperonin complex) created a new classi cation system based on the 25 distinct monophyletic clades formed, which came to be treated as sections (Houbraken et al 2011). Visagie et al. (2014) provided the databases of the Penicillium type-species with partial sequence data for the cam, TUB2 and rpb2 genes and provided guidelines for the identi cation and nomenclature of the genus.
Recently, a new series classi cation system was proposed that was based on morphological data, secondary metabolism and molecular data of the ITS barcodes cam, TUB2 and rpb2. In this update, the genus Penicillium now has 2 subgenera (Aspergilloides and Penicillium), 32 sections, 89 series and 483 valid species (Houbraken et al. 2020).
The Exilicaulis section has 58 accepted species distributed in six series: Lapidosa, Corylophila, Restricta, Citreonigra, Alutacea and Erubescentia. Species of the Exilicaulis section have monoverticillated and some biverticillated conidiophores with stipes without a terminal vesicular swelling (Pitt et al. 1980;Houbraken et al. 2011;Houbraken et al. 2020; Labuda et al. 2021). Although the species in this section are promising in the production of new bioactive metabolites against pathogenic strains of fungi and bacteria, they also deserve attention for the production of toxic and genotoxic compounds, such Penicillium canis and the recently described species Penicillium labradorum both of ser. Erubescentia species, which can affect commercial mushroom cultures, and cause pathologies that are responsible for the death of immunosuppressed dogs, (Stewart et  During the isolation of fungi from sediments collected from the Amazon River in the state of Amapá in Brazil, four isolates stood out for their high production of exudates. These presented potential for biosynthesis of siderophores, phosphate solubilization, and production of compounds with antifungal activity and enzymes of industrial interesting. The phylogenetic analysis using the ITS regions CAM, RPB2 and TUB2, together with morphological characters, supports the proposition of a new species named Penicillium amapaense that belongs to section Exilicaulis ser. Erubescentia. Furthermore, two of these isolates cluster with P. labradorum, and is the rst report of the occurrence of this species in Brazil.

Isolation and culture conditions
For isolation, sediments from the Amazon River were collected in the city of Macapá in the state of Amapá, Brazil and kindly sent to Laboratory of Molecular Biology at Embrapa Western Amazon in Manaus by Cristóvão Tertuliano de Almeida Lins. The isolation was performed by serial dilution of 1 gram of sediments at a concentration of 10 3 and inoculated using the spread plate technique in ISP2 (Yeast Extract-Malt Extract Agar) medium. All culture mediums are presented in SI 1.
The isolates obtained were submitted to the monosporic cultivation technique in order to obtain pure cultures in potato dextrose agar (PDA) medium and were then preserved in potato dextrose medium containing 20% (v/v) glycerol at -80°C. The isolates were deposited in the biological collection at INPA (National Institute for Amazonian Research) under codes: INPA-AP07, INPA-AP25, INPA-AP10 and INPA-AP10a. Morphological analysis Isolates were grown for 7 days at 25°C in PDA, MEA, YES, OA and CYA media. Additional plates of CYA at 30°C and 37°C were also observed (Supplementary information SI 1) (Visagie et al. 2014 Micromorphological characters, such as conidiophore rami cations, conidia texture and number of phialids per conidiophore were analyzed using scanning electron microscopy (JSM-IT500HR) at the Multiuser Center for the Analysis of Biomedical Phenomena at the Amazonas State University (CMABio).
For this, the microculture was prepared in MEA at 28°C for 7 days, xed with Karnovisky, and then dehydrated in increasing concentrations of alcohol (30%, 50%, 70%, 80%, 90%, 100%) for 15 min each. The step with absolute alcohol was repeated twice. Drying was performed in a critical point dryer (Leica EM CPD300) and submitted to sputter coating (DII-29010SCTR Smart Coater). The measurement of length and width of 30 conidia and phialides and the length of the stipes was performed using an optical microscope with a Carl ZEISS Axio Imager v2 camera.

DNA extraction, PCR ampli cation, and sequencing
The isolates were grown in PD medium at 28°C with rotation at 150 rpm for 7 days. The mycelial mass that was obtained was ltered and DNA extraction was performed with cationic detergent CTAB 2% according to Doyle and Doyle (1987).
PCR ampli cations of the four loci (ITS, TUB2, CAM and RPB2) were carried out using an Easytaq® kit (Synapse Biotechnology) following the manufacturer's instructions. Thermocycling conditions were as follows: initial denaturation at 95°C for 3 min, 35 cycles of denaturation at 95°C for 45 sec, annealing according to each primer-speci c temperature (Supplementary information SI 2) for 45 sec, extension at 72°C for 1 min. Final extension at 72°C for 5 min.
The PCR products were resolved on a 1.5% (w/v) agarose gel stained with ethidium bromide and photodocumented under ultraviolet light using a molecular imaging system (Loccus Biotechnologic L-Pix. Chemi). The amplicon sizes were compared with the 1 kb plus marker (Invitrogen -Catalog number: 10787018). PCR products were puri ed enzymatically using the ExoSAP-IT reagent (USB, Cleveland, OH, USA) following the manufacturer's instructions. Sequencing reactions were carried out in a 10 µL reaction volume which contained 5 µL of the puri ed PCR product, 0.5 µL of BigDye terminator v3.1 (Thermo Fisher), 1.5 µL 5X buffer, 2 µL ultrapure water and 3.2 pmol of each primer. Thermocycling conditions were as follows: 96°C for 1 min, followed by 35 cycles at 96°C for 15 sec, 50°C for 15 sec, and 60°C for 4 min. Sequencing reactions were visualized in a genetic analyzer (ABI 3500, Thermo Fisher).
The ML analysis included 1,000 replicates (bootstrap), and used all sites, with the best model selected by IQtree. Bayesian inference (BI) was based on the model adopted in PAUP* 4.0 and MrModeltest v2 (Nylander 2004). All sites in the loci were considered and the analysis was performed for ten million generations, with the rst 25% of the trees discarded and burned using the MrBayes v 3.7 tool available from CIPRES (https://www.phylo.org/). Posterior probability (PP) and tree topology were visualized with FigTree v1.1.2 (Rambaut 2009). The consensus tree of the ML and BI analyses was generated manually from the topology obtained using FigTree in BI analysis with the posterior probability values, plus the bootstrap values generated by the maximum likelihood analysis, using the CorelDraw editing package.  Enzymatic activity and in vitro physiological traits for promotion of plant growth Qualitative analysis of the enzymatic activity was performed for the production of amylase, cellulase, chitinase, lipase, protease. We also performed in vitro evaluation of plant growth promoting (PGP) traits such as phosphate solubilization and siderophore production. The culture media used for each test are presented in Supplementary Information SI 3. Bioassays were performed in triplicate and incubation was at 28°C for 7 days, and the presence of halos around the colonies indicated a positive result for the selected enzymes. The colonies that showed a clear halo on NBRIP (National Botanical Research Institute's phosphate growth) medium were marked as positive for phosphate solubilization and those with an orange halo on CAS agar medium were considered as positive for siderophore production.
Chemical pro le and compound dereplication analysis A 4 mL sample of resuspended crude extracts at 2 mg mL −1 in acetonitrile (ACN) were injected in methanol and analysed via high resolution (5000) UPLC-MS analysis. The UPLC-MS analysis was performed using a UHPLC (Acquity Waters) coupled to a mass spectrometer (Q-ToF micro TM Micromass, Waters) using an electrospray ionization (ESI) source with a UPLC BEH C18 1.7 mm column (2.1x100 mm, Acquity Waters). The mobile phase A was milli-Q water with 0.1% of formic acid (FA), and mobile phase B was ACN with 0.1% of FA. The ow rate of 0.35 mL min −1 was used following the gradient: 0-16 min, 20% B to 70% B; 16-20 min 70% B to 100% B; 20-23 min, 100% B (clean-up) and 2 min of post time at 20% B for column re-equilibration. The mass spectrometer voltages and temperatures were set as follows: capillary 4000 V; sample cone 40 V, extraction cone 2 V; gas source temperature at 150 o C; desolvation temperature at 10 o C; drying gas at 600 L hr −1 , and the fragmentation was performed using collision energy of 25 and 35 eV. Mass spectra were acquired in pro le and positive or negative ion mode and the acquisition range was 100-1000 m/z. Data were treated using MassLynx V 4.1 and Mzmine 2.53. Compound dereplication was performed using the GNPS platform via FBMN and by manual interpretation of MS/MS spectra.

Phylogeny and nucleotide variation
The alignment resulted in 2,649 characters, which included gaps in which the ITS locus contributed 580 characters and CAM, RPB2 and TUB2 contributed 559, 978 and 532, respectively. The best model adopted by PAUP*4.0 for BI data for all loci was GTR+I+G, while for the ML analyses the best t model was TIM2e+I+G4 for ITS and TNe+I+G4 for CAM, TUB2 and rpb2. The ML and BI analyses revealed that all isolates in the present study were grouped in the ser. Erubescentia species. The isolates INPA-AP10 and INPA-AP10a clustered with P. labradorum, while isolates INPA-AP07 and INPA-AP25 formed a sister group with P. rubidurum, with 87 bootstrap support in the ML analysis and 0.51 of PP in the BI analysis (Fig. 1). center and slightly grooved, irregular edges, absence of exudate, presence of soluble strong orange pigment (ISCC-NBS 50); reverse: smooth, strong yellowish brown center (Fig. 3a, 3b, 3c). In addition to the conditions described above, the strain was analyzed at temperatures of 30°C and 37°C, and showed the same characteristics mentioned above, with the exception of the size of the colony, which presented sizes of 15 mm and 5 mm, respectively. Notes Penicillium amapaense sp. nov. differs from Penicillium rubidurum by having a smaller stipe length and due to its ability to grow on CYA medium.
In all, eleven out of the twelve phytopathogenic fungi tested presented inhibition of mycelial growth for at least one of the four Penicillium isolates obtained in this study (Fig. 4a). Among the phytopathogenic fungi tested, the isolates of P. amapaense and P. labradorum showed potential for inhibiting mycelial growth of up to 49 and 58%, respectively. Among the phytopathogenic fungi inhibited by all the Penicillium isolates, C. guaranicola presented a mycelial growth inhibition rate (IR) that ranged from 34.28 to 58.53%, Ps. gilvanii ranged from 44.10 to 58.09%, and Fusarium sp. ranged from 21.42 to 48.88%, while F. decemcellulare ranged from 32.22 to 47.77% (Fig. 4a). Some phytopathogenic fungi were not inhibited by all the Penicillium isolates; C. spathulatum and Colletotrichum sp. were not inhibited by the P. amapaense isolate INPA-AP07, C. siamense was not inhibited by the P. amapaense isolate INPA-AP25, and C. scovillei was not inhibited by the P. labradorum isolate INPA-AP10a. Conversely, N. formicarum was inhibited only by the P. labradorum isolate INPA-AP10, while none of the four Penicillium isolates showed inhibition of mycelial growth against Rhizoctonia sp. (Fig. 4a).

Analysis of secondary metabolites
The investigation of the secondary metabolites obtained from the different strains was carried out manually via the analysis of accurate mass, fragmentation pro le and speci c annotations in the databases used. In the process of chemical dereplication of extracts, the isolate P. labradorum INPA-AP10 stood out for producing the largest number of compounds previously identi ed (Fig. 5). In total, fourteen compounds were detected, which corresponded to the polyketide subclasses of the curvularins (compounds 1a, 1b, 1c, 1d and 2) and presented as accurate mass m/z 307.1190, m/z 307.1183, m/z 307.1169, m/z 307.1181, m/z 305.1497 and anthraquinones (compounds 3, 4 and 5) with m/z 299.0198, m/z 318.9999, m/z 332.9801 (Table 2).
In general, the P. labradorum strains presented a similar metabolic pro le, but with the absence of anthraquinones in the extract obtained from the isolate INPA-AP10a. The specimens of the species P. amapaense (INPA-AP25 and INPA-AP07) also showed high similarity in the compounds produced by the two strains. Coincidentally, the production of anthraquinones was detected only in the P. amapaense isolate INPA-AP25, with the most signi cant distinction found between the compounds produced by this strain and P. amapaense INPA-AP07. Interestingly, when comparing the metabolic pro les of P. labradorum and P. amapaense species, the main metabolic difference occurred in the production of curvularins. As seen in Fig. 5, none of the isomers referring to the accurate mass of m/z 307 (1a-1d isomers) were identi ed in the P. amapaense strains, while the production of compound 2 was signi cantly observed (Fig. 5).
The analysis of the LC-MS data using the GNPS platform indicated that peak 1 is the curvularin, 5,13,15trihydroxy-9-methyl-10-oxabicyclo [10.4.0]hexadeca-1 (12),13,15-triene-3,11-dione ( Table 2). The cosine value of 0.92 was obtained when comparing the MS/MS spectra for compound 1a and for curvularin, which was indicated by the platform, thus suggesting high similarity in the fragmentation pro le. This data is corroborated by the comparison of the spectra ( Supplementary Information SI 8). The molecular formula C 16 H 20 O 6 , obtained from the accurate mass of compound 1a ( Supplementary Information SI 9), is also in accordance with the curvularin indicated on the GNPS platform. Furthermore, for the remaining peaks referring to the possible curvularin isomers (1b-1d), lower values of cosine were obtained, suggesting other possible identities for them; all are found in the Supplementary Information SI 14. The query of the molecular formula C 16 H 20 O 6 was performed in other databases (see Table 1) and indicated possible hits that correspond to other known curvularins. The hits obtained can be seen in Fig. 5b.
The molecular formula C 16 H 22 N 2 O 4 , suggested for the accurate mass m/z 305.1490 (2) (Supplementary Information SI 10), was consulted in the databases used and no hit was found, thus suggesting that it is a metabolite that has not yet been reported in the literature. Although there are two nitrogen units in the molecule, there is evidence that this compound has a biosynthetic origin that is related to the class of curvularins, due to the similarity in retention time and the observed hydrogen de ciency index (see Table  1) and, mainly, to the MS/MS spectrum when compared to that of compound 1a ( Supplementary   Information SI 11).
Regarding the anthraquinones observed in the extracts of P. labradorum INPA-AP10 and P. amapaense INPA-AP07, data processing on the GNPS platform and database consultation allowed the identi cation of emodic acid for compound 3, 5-chloro-ω-hydroxyemodin for compound 4 and 2-chloroemodic acid for compound 5. In all cases, the molecular formulas were obtained with low values of deviation from the respective exact masses (see Table 2), as well as the observed isotopic pro le for 4 and 5 (Supplementary Information SI 12 and SI 13), which corroborated in con rming the identity of the compounds (Figs. 5a and 5b).  The production of amylase, cellulase and siderophores were detected in all isolates of P. amapaense and P. labradorum. Only in Penicillium amapaense (INPA-AP07 and INPA-AP25) was it able to solubilize phosphate. No enzymatic activity was observed for chitinase, lipase and protease for all isolates analyzed (Table 3, Fig. 6).

Discussion
A total of four isolates of the genus Penicillium were recovered from sediments of the Amazon River, which is a rich but still underexplored source of microorganisms. These isolates were subject to morphological characterization as well as sequence analyses of ITS, CAM, TUB2 and RPB2 loci. The isolates INPA-AP25 and INPA-AP07 were characterized as a new species and named here as P. amapaense sp. nov, while the other two isolates INPA-AP10 and INPA-AP10a were identi ed as P.
labradorum. These data also support that both species are members of the sect. Exilicaulis and ser. Erubescentia.
The phylogenetic inference of P. amapaense sp. nov, based on four loci (ITS, CAM, TUB2 and RPB2), had low bootstrap support (ML = 87) and posterior probability from Bayesian inference (PP = 0.51). Despite the relatively low support achieved by bootstrap and PP, P. amapaense sp. nov. could be differentiated from the four most phylogenetically related species (P. rubidurm, P. menorum, P. guttulosum, and P. pimiteouiense) due to the presence of 15 SNPs identi ed in ITS, CAM and TUB2 loci. By narrowing the comparisons to the sister clade, P. rubidurum, 22 SNPs were identi ed only in the partial sequence of the calmodulin (CAM), thus evidencing the molecular differences between the two species (Fig. 2). In addition, P. amapaense differs from P. rubidurum by having smaller stipe length (9.6 -20.6 µm vs. 15 -60 µm) and also by growing on CYA medium (Peterson et al. 1999).
Phylogenetic analysis of Penicillium sect. Exilicaulis ser. Erubescentia evidenced the lower resolution achieved by the current recommended barcode, which, for most of branches, produced values lower than 80 and 0.95 for bootstrap and posterior probability from Bayesian inference, respectively (Visagie et al. 2016). Recently, a phylogenetic tree of Penicillium species was also constructed with bootstrap values below 70 and posterior probability from Bayesian inference below 0.95, but still allow the characterization of P. hermansii as a new species (Houbraken et al. 2019). Taken together, our results and the literature indicate the need to incorporate new loci, or even the development of new algorithms for a more accurate taxonomic analysis to better resolve the evolutionary relationships of fungal species (Choi & Kim 2017).
In the future, we believe that unambiguous and highly supported phylogenetic analysis will be obtained for Penicillium species based on whole genome sequencing (WGS) drafts. Similar analyses have already been carried out for species of the genus Fusarium for which 59.1 Kb was obtained based on 19 housekeeping genes, thus allowing high support that included highly related members of the F. solani species complex (FSSC) (Geiser et al. 2021). In the near future, the lower cost of obtaining fungal genomes will allow us to perform WGS-based phylogenetic analysis, as well as the creation of an automated high-throughput platform for taxonomic analysis of fungal species, as well as the Type Strain Genome server (TYGS) platform, which is now available for taxonomy of bacteria based on bacterial genomes (Meier-Kolthoff & Göker 2019).
Herein, we examined the biotechnological and agricultural potential of P. amapaense sp. nov.. The two isolates of this species (INPA-AP25 and INPA-AP07) showed the ability to produce compounds of biotechnological interest, as well as possible metabolites (Fig. 5). Moreover, the in vitro antibiosis assays demonstrated the ability to control phytopathogenic fungi that affect important agricultural crops in Brazil. The results showed that isolates could produce compounds that inhibited up to 58% of mycelial growth of 11 out of the 12 phytopathogenic fungi tested. In addition, the isolates demonstrated the ability to produce enzymes such as amylase and cellulase, as well as the ability to produce functional traits related to plant growth promotion such as siderophores and phosphate solubilization. These results are the rst indication that this new species has both biotechnological and agricultural potential that needs to be further investigated.
The LC-MS data allowed the annotation of nine compounds, which were all of a polyketide nature.
Among the species studied, there is a metabolic proximity that is corroborated by the proximity in the phylogenetic analysis. The differences observed refer to the presence or absence of anthraquinones, especially chlorinated ones. In general, curvularins and anthraquinones produced by Penicillium species are bioactive (Bladt et al. 2013 P. labradorum has recently been described as causing disease in an immunosuppressed dog in Florida/USA (Rothachker et al. 2020) and, since this event, only one report in Genbank was found in November, 2020, in Los Angeles,USA, which was isolated from the mushroom Panaeolus cinctulus and identi ed by ITS region (accession number: MW241166.1). These two reports are an indication that this species should be further studied as to its ability to cause diseases in other animals or even in humans, as well as its ability to contaminate edible mushroom species. Penicillium hermansii, which is also a member of the ser. Erubescentia, has been reported to cause problems in mushroom cultivation (Houbraken et al. 2019).
In this work, P. labradorum was isolated for the rst time as a saprophyte, and this is the rst study that reports its ability to control phytopathogenic fungi, to produce enzymes of industrial interest and to possess traits related to plant growth promotion, in addition to the rst report of the identi cation of extrolites produced by this species. The results reveal that P. labradorum showed the best performance against the 12 phytopathogens evaluated when compared to P. amapaense. The isolate INPA-AP10 showed the best performance in growth inhibition rates (IR) for most of the tested phytopathogens, which were mostly species of the Colletotrichum genus, in addition to being the only one with the ability to inhibit N. formicarum (Fig. 4a). This fact is possibly due to the greater diversity of compounds produced by this isolate in relation to the others (Fig. 5).
Our report presented two physiologically interesting Penicillium species with potential for production of enzymes of industrial interest, but which are also promising from a chemical point of view due to the production of compounds with antifungal activity against different phytopathogens. Furthermore, the ability to produce siderophores and phosphate solubilization are interesting characteristics to be further investigated for these isolates. New fungal species, such as P. amapaense or yet unexplored species such as P. labradorum, are a potential source for identi cation and characterization of new molecules and for developing new products and biotechnological processes that together are the key to the new bioeconomy and a new form of ecologically sustainable agricultural development.

Declarations
Author contributions All the authors contributed equally to the study conception and design. All the authors commented on the previous versions of the manuscript. All the authors read and approved the nal manuscript.  Polymorphism identi ed in ITS CAM and TUB2 loci in Penicillium amapaense sp. nov. and in closely related species; Asterisk indicates single nucleotide polymorphism (SNP) present only in P. amapaense species.