Extracts of Talaromyces purpureogenus Strains from Apis mellifera Bee Bread Inhibit the Growth of Paenibacillus spp. In Vitro

Honey bees coexist with fungi that colonize hive surfaces and pollen. Some of these fungi are opportunistic pathogens, but many are beneficial species that produce antimicrobial compounds for pollen conservation and the regulation of pathogen populations. In this study, we tested the in vitro antimicrobial activity of Talaromyces purpureogenus strains isolated from bee bread against Paenibacillus alvei (associated with European foulbrood disease) and three Aspergillus species that cause stonebrood disease. We found that methanol extracts of T. purpureogenus strains B18 and B195 inhibited the growth of P. alvei at a concentration of 0.39 mg/mL. Bioactivity-guided dereplication revealed that the activity of the crude extracts correlated with the presence of diketopiperazines, a siderophore, and three unknown compounds. We propose that non-pathogenic fungi such as Talaromyces spp. and their metabolites in bee bread could be an important requirement to prevent disease. Agricultural practices involving the use of fungicides can disrupt the fungal community and thus negatively affect the health of bee colonies.


Introduction
The reproduction of most flowering plants and~30% of all crops is dependent on pollination [1,2]. The estimated value of pollinators to the agricultural economy was USD 164 billion in 2009 [3] and now stands somewhere between USD 235 and USD 577 billion annually [4,5]. This value increases in line with crop production, and so does our dependence on pollinators for the maintenance of food security [2,4,6,7]. However, there has been a recent, dramatic decline in natural pollinator abundance and diversity [8,9]. Accordingly, managed pollinators such as honey bees (Apis mellifera, Linné 1758) (Hymenoptera: Apidae) play a key role in agroecosystems by augmenting wild pollinators and thus ensuring crop quality and yield stability. However, an array of biotic and abiotic stressors associated with agricultural intensification, parasites, and pathogens negatively affect the health of managed bees [10][11][12]. The deterioration in bee health results in the loss of colonies, particularly in the northern hemisphere [13][14][15]. Optimizing the health of individual bees is necessary to improve colony fitness, correlating directly with their benefits to humans, including pollination services as well as the production of honey.

Extract Preparation for Chemical Analysis and Antimicrobial Tests
Extracts were prepared from cultures grown on solid CYA and YES medium for 10 days [52]. Briefly, three plugs were cut from the center of single colonies and transferred to 2 mL screw cap vials. We added 0.8 mL ethyl acetate containing 1% formic acid and incubated in an ultrasonic bath (35 kHz) for 30 min [52]. The liquid phase was transferred to a fresh vial and evaporated under nitrogen, and the residues were redissolved in 40 µL methanol and stored at 4 °C. The fungal strains were also incubated in liquid MP medium and were lyophilized and extracted in methanol as previously described [36]. The crude extracts were redissolved in methanol to a concentration of 100 mg/mL and were stored at -20 °C. Before UHPLC-HR-MS analysis, all samples were centrifuged (8000× g, 1 min, room temperature), and 30-40 µL was transferred to 2-mL HPLC vials with glass microinserts.

Molecular Barcoding and Phylogenetics
We collected mycelia from 7-day-old colonies grown on SDA using a sterile scalpel and homogenized the tissue under liquid nitrogen with a mortar and pestle. The ground tissue (50-80 mg) was mixed with 490 µL 2 × CTAB buffer (2% CTAB, 1.4 M NaCl, 20 mM EDTA, 100 mM Tris of pH 8) preheated to 60 °C. We then added 10 µL of proteinase K (50 mg/mL) and incubated at 60 °C for 1 h on a shaking platform. The lysate was extracted by adding 500 µL 24:1 chloroform:isoamylalcohol, and the aqueous layer (400 µL) containing was transferred to a new 1.5 mL Eppendorf tube. The DNA was precipitated by adding 270 µL isopropanol. After centrifugation (14,000× g, 10 min, room temperature), the pellet

Extract Preparation for Chemical Analysis and Antimicrobial Tests
Extracts were prepared from cultures grown on solid CYA and YES medium for 10 days [52]. Briefly, three plugs were cut from the center of single colonies and transferred to 2 mL screw cap vials. We added 0.8 mL ethyl acetate containing 1% formic acid and incubated in an ultrasonic bath (35 kHz) for 30 min [52]. The liquid phase was transferred to a fresh vial and evaporated under nitrogen, and the residues were redissolved in 40 µL methanol and stored at 4 • C. The fungal strains were also incubated in liquid MP medium and were lyophilized and extracted in methanol as previously described [36]. The crude extracts were redissolved in methanol to a concentration of 100 mg/mL and were stored at -20 • C. Before UHPLC-HR-MS analysis, all samples were centrifuged (8000× g, 1 min, room temperature), and 30-40 µL was transferred to 2-mL HPLC vials with glass micro-inserts.

Molecular Barcoding and Phylogenetics
We collected mycelia from 7-day-old colonies grown on SDA using a sterile scalpel and homogenized the tissue under liquid nitrogen with a mortar and pestle. The ground tissue (50-80 mg) was mixed with 490 µL 2 × CTAB buffer (2% CTAB, 1.4 M NaCl, 20 mM EDTA, 100 mM Tris of pH 8) preheated to 60 • C. We then added 10 µL of proteinase K (50 mg/mL) and incubated at 60 • C for 1 h on a shaking platform. The lysate was extracted by adding 500 µL 24:1 chloroform:isoamylalcohol, and the aqueous layer (400 µL) containing was transferred to a new 1.5 mL Eppendorf tube. The DNA was precipitated by adding 270 µL isopropanol. After centrifugation (14,000× g, 10 min, room temperature), the pellet was washed with 1 mL ice-cold 70% ethanol and dried in a vacuum concentrator for 10 min. The DNA was dissolved in 20 µL of TE buffer with 2 µL of RNase A (10 mg/mL) for 3 h at 37 • C. Samples were then stored at -20 • C. The concentration and quality of DNA was determined using a NanoDrop UV spectrophotometer and by 1% agarose gel electrophoresis.
To create the barcodes, we amplified four DNA regions: β-tubulin (BenA), calmodulin (CaM), the internal transcribed spacer (ITS), and RNA polymerase II second largest subunit (RBPII), which are often used for the molecular identification and phylogenetic analysis of fungal isolates [52,54] (Table 1). Each 20 µL PCR mix contained 10 µL Taq 2× Master Mix, 0.25 µM of each primer, 100-200 ng of template DNA, and PCR-grade water to top up. The reaction conditions are summarized in Table 2. PCR products were purified using ExoSAP-IT (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions but using modified thermal conditions (37 • C for 35 min, 85 • C for 15 min, and 10 • C infinite hold). The cleaned PCR products were sent with forward and reverse primers for bidirectional sequencing (Eurofins, Val Fleuri, Luxembourg). Sequence reads were analyzed and assembled using Geneious v10.2.6, and the assembly consensus sequences of the barcodes were compared to the sequences in the International Sequence Database (INSD) using the GenBank basic local alignment search tool (BLAST) [60]. The search was adjusted for comparison with the type and reference material and the RefSeq Targeted Loci database (available only for ITS sequences). For species assignment based on ITS sequences, we used the recommended identity threshold of ≥97% [61]. Due to the lack of available Ben1A sequences from the type material in the database (December 2022), phylogenetic analysis was carried using the Ben2A, CaM, ITS, and RBPII sequences (Supplementary Table S1).
Sequence alignments were prepared using MUSCLE in Geneious v10.2.6, and single gene trees (Supplementary Figures S1-S5) were constructed using IQtree with default settings [62,63]. From the concatenated alignment, uninformative sites were removed using Gblocks 0.91b [64,65]. Next, we used partition finder in Cipres Science Gateway v3.3 [66] to determine best-fitting models of molecular evolution and partition schemes. Two simultaneous phylogenetic analyses were carried out using MrBayes v3.2 [67][68][69] with four Markov chains (three heated, one cold) for 10 million generations with a sampling frequency of 1000 and a 25% burn-in. From the resulting trees, a 65% majority-rule consensus tree was constructed and visualized using iTol v6 [70].

Detection of Talaromyces spp. at Different Time Points and Locations
Samples were collected in July 2019 and April and July 2020 from five apiaries in the South Bohemia region of Czech Republic as described in Section 2.1. The collection sites were Kamenny Malikov village (KM Zirovnice and KM agro) and three apiaries in Ceske Budejovice (CB campus, CB Litvinovice, and CB Kroclov) from urban and suburban areas and the city periphery, respectively. The locations were ranked by anthropogenic influence, focusing on the level of urbanization and the agricultural landscape within~6 km (Supplementary Figure S6). ITS sequences were used for species assignment.

UHPLC-HR-MS Analysis and Metabolic Fingerprinting
Samples were processed for metabolic fingerprinting as previously described [71]. Briefly, the samples were fractionated on a 1290 UHPLC system (Agilent, Santa Clara, CA, USA) equipped with DAD, ELSD, and a maXis II (Bruker, Billerica, MA, USA) ESI-qTOF-HRMS. We used a gradient of 0.1% formic acid in water (buffer A) and 0.1% formic acid in acetonitrile (buffer B) at a flow rate of 600 µL/min. The gradient began at 95% A and was held for 0.30 min before a transition to 4.75% A over 18.00 min and 0% A over 18.10 min, with a hold for 22.50 min. The gradient then increased to 95% A over 22.60 min followed by a hold for 25.00 min. The column oven temperature was set at 45 • C, and we used an Acquity UPLC BEH C18 1.7 µm column (2.1 × 100 mm) with an Acquity UPLC BEH C18 1.7 µm VanGuard pre-column (2.1 × 5 mm).
For micro-fractionation [42], extracts were injected into the UHPLC-HR-MS system described above. However, the flow path was changed so that 90% of the flow was diverted to a custom-made fraction collector (Zinsser-Analytik, Frankfurt, Germany) while the rest was analyzed in MS/MS mode in the maXis II. Collision-induced fragmentation at 28.0-35.05 eV was achieved using argon at 10 −2 mbar. Micro-fractionation assay plates were prepared by injecting 2 or 5 µL of extract. For each injection, 159 fractions were generated and collected on a 384-well plate (fraction length: 7 s). Before screening, the plates were dried under a vacuum using an HT12-II centrifugal concentrator (Genevac, Ipswich, UK) at 35 • C.
2.6. Evaluation of Antimicrobial Activity 2.6.1. Antifungal Activity Test Antifungal activity was tested against three Aspergillus species associated with stonebrood disease (A. flavus ATCC9170, A. fumigatus ATCC10894, and A. niger ATCC10549) from Fraunhofer strain collection (STO20519-STO29521) using CLSI M51-A parameters [72]. The Aspergillus strains were cultivated on PDA (VWR International, Radnor, PA, USA) for 7 days at 25 • C. The spores were washed by 5-6 mL of 0.05% Tween-80 (Sigma-Aldrich, St. Louis, MO, USA) and filtered through three layers of miracloth to remove the hyphal structures. The inoculum was adjusted to OD 530 = 0.14-0.46 based on spore morphology [73]. The inoculum was spread evenly over the surface of Mueller-Hinton agar (Carl Roth, Karlsruhe, Germany) using a sterile cotton swab. The depth of the medium was consistent in each Petri dish (4 mm). The extracts from the liquid T. purpureogenus strains cultures ( Figure 1) were diluted with methanol to 10 mg/mL, and 25 µL of the crude extract was applied to 6 mm cellulose disks and left in the laminar flow cabinet for 30 min to dry. Methanol was used as a control. The disks were placed on the inoculated media and pressed down with sterile forceps. The zones of inhibition were evaluated in triplicate after incubation at 35 • C for 24 h.

Antibacterial Activity Test
We tested the antibacterial activity of the extracts in a dilution series (25-0.39 mg/mL) in triplicate using a micro-broth dilution test against P. alvei CCM 2051 provided by the Czech Collection of Microorganisms (Masaryk University, Brno, Czech Republic) as previously described [74] with modifications. Briefly, extracts were diluted 1:1 with Mueller-Hinton broth, yeast extract, potassium phosphate, glucose, and pyruvate (MYPGP), and 100 µL of the solution was pipetted to the first row of the microdilution 96-well plate. A two-fold dilution series was prepared by transferring the extract solutions (50 µL) to the next rows containing 50 µL of the MYPGP. Dilutions of methanol were included as a solvent growth control. The lyophilized bacteria were cultivated at 37 • C for~24 or~48 h on MYPGP agar. The inoculum was diluted in sterile water and adjusted to 0.5 McFarland units using the DEN-1 McFarland densitometer (BioSan, Riga, Latvia). The bacterial suspension was diluted 1:150 with the MYPGP (~10 6 CFU/mL), and 50 µL was added to each well within 15 min. Bacterial growth was measured at 625 nm using an xMark spectrophotometer (Bio-Rad, Hercules, CA, USA). Minimum inhibitory concentrations (MICs) were defined as the lowest concentration of extract or standard that inhibited growth by at least 80% relative to the control (bacterial suspension with no extract/solvent).

Bioactivity-Guided Dereplication
The potency of the T. purpureogenus strains extracts was determined using microbroth dilution assays as previously described [75][76][77]. Briefly, extracts were screened in a 12-point dilution series (2-0.001 mg/mL) in triplicate. In addition to four Paenibacillus strains (P. lautus DSM3035, P. lactis FH1832, and two unspecified taxa from the Fraunhofer strain collection ST133196 and ST514408 [78]), we also included the control strain Escherichia coli ATCC35218 and the type strain Bacillus subtilis DSM 10. For all bacteria, the density of the overnight pre-cultures, incubated in cation-adjusted Mueller-Hinton II medium (BD) at 37 • C while shaking at 180 rpm, was adjusted to 5 × 10 5 cells/mL. The assay plates were set up by pipetting 100 µL of adjusted bacterial suspension in each well, except the medium control wells H01-H05 (blank or "low"). Wells H06-H12 contained bacterial suspensions without extracts or antibiotics (growth control, "high"). On each assay plate, we also assessed the MIC of three reference antibiotics (ciprofloxacin 0.5-0.0002 µg/mL, cefotaxime, and gentamicin both 64-0.03 µg/mL). The dilution series of pure MeOH, the sample triplicate solved in MeOH, and the reference antibiotics was prepared by adding additional 98 µL of bacterial suspension plus 2 µL of the respective test samples to the first well of rows A-G. Hence, the maximal solvent concentration per well was 1%. Next, the 1:2 dilution series were prepared by transferring 100 µL suspension. After the last dilution step in column 12, the remaining 100 µL were discarded. Assay plates were incubated for 18 h at 37 • C and 85% relative humidity, shaking at 180 rpm, before we measured the turbidity at 600 nm on a LUMIstar Omega microplate spectrophotometer (BMG Labtech, Ortenberg, Germany) as a proxy for cell growth. The relative growth inhibition was calculated based on the absorbance units (AU) of the sample and the controls (see above) using the following formula: MICs were defined as above. Extracts with bioactivity observed over at least three dilution steps (≤0.5 mg/mL) were selected for micro-fractionation. Selected extracts were injected twice into our UHPLC-HRMS-MS system (2 and 5 µL µL, see Section 2.5). The dried assay plate was re-screened against P. lautus by adding 50 µL of adjusted pre-culture (see above) to each well except column 1 (media blank). Columns 2 and 3 contained no fractions, but a dilution series of gentamicin ranging from 256-0.008 µg/mL. Column 4 contained only bacterial suspension (growth control), while the remaining 320 wells contained 2 × 159 fractions collected from the two injection and an unfractionated extract control of the respective volume. Incubation, read-out, and calculation of the relative growth inhibition was carried out as described above.

Colony Morphology
The use of four standard media enabled us to distinguish the fungal strains based on specific morphological traits: size, sporulation color, and pigment production ( Figure 2). The size of the cultures on MEA was comparable (Table 3) but differences in red pigment production (Monascus-like red pigments) were observed in the fruiting body (B18 > B13 > B195 on reversed MEA and YES) and media (B13 > B195 > B18 on CYA). The production of low levels of acid on CREA was observed only for strain B49 (Figure 2, Table 3).
dried assay plate was re-screened against P. lautus by adding 50 µL of adjusted pre-culture (see above) to each well except column 1 (media blank). Columns 2 and 3 contained no fractions, but a dilution series of gentamicin ranging from 256-0.008 µg/mL. Column 4 contained only bacterial suspension (growth control), while the remaining 320 wells contained 2 × 159 fractions collected from the two injection and an unfractionated extract control of the respective volume. Incubation, read-out, and calculation of the relative growth inhibition was carried out as described above.

Colony Morphology
The use of four standard media enabled us to distinguish the fungal strains based on specific morphological traits: size, sporulation color, and pigment production ( Figure 2). The size of the cultures on MEA was comparable (Table 3) but differences in red pigment production (Monascus-like red pigments) were observed in the fruiting body (B18 > B13 > B195 on reversed MEA and YES) and media (B13 > B195 > B18 on CYA). The production of low levels of acid on CREA was observed only for strain B49 (Figure 2, Table 3).   Our cultivation-dependent method detected isolates of the genus Talaromyces in bee bread from three of the five apiaries, but only in a few hives in spring (Table 4). Based on ITS sequencing, we identified T. purpureogenus in apiaries in areas with a low anthropogenic influence (CB Kroclov and KM Zirovnice) and two species (T. purpureogenus and T. piceae) in urban samples (CB campus). We did not detect any Talaromyces spp. in the other apiaries (KM agro and CB Litvinovice). Images of the isolates are provided in Supplementary Table S2. Fungal isolates representing other genera were not considered.

Molecular and Phylogenetic Characterization
We assigned all strains to the species T. purpureogenus based on ITS sequence similarity, thus excluding other species in the RefSeq ITS database ( Table 5). The sequences were uploaded to the NCBI database (OR192894-OR192900, Supplementary Table S1). By comparing these sequences with the database, we also observed the highest identity (>98%) with T. purpureogenus for the other three barcodes (Ben2A, CMD, and RBPII). Only one Ben1A sequence from Talaromyces spp. type material was available in the database-T. stipitatus (query cover/identity (QC/I) = 88/96.45%; XM_002341495.1). In the standard database, the Ben1A sequences (OR327661-OR327667) showed the highest QC/I to T. marneffei (>99/>93.48%; CP045656.1) and less than 90% identity to T. purpureogenus isolated from medicinal plants (QC/I = >95/89.65-89.82; HM596783.1) [79]. Table 5. Five sequences with the highest identity in the ITS region based on BLAST analysis. The ITS sequence was identical in all seven strains. QC = query cover, I = identity. The resulting phylogeny supports the monophyletic nature of Talaromyces and is composed of several subclades containing different members of this genus (Figure 3). General support for the Talaromyces clade was high, but several of the deepest clades were only weakly supported (see Supplementary Figure S1 for numeric bootstrap values). However, the overwhelming majority of shallow clades relevant for the taxonomic placement of our isolates were sufficiently supported to draw firm conclusions. General support for the Talaromyces clade was high, but several of the deepest clades were only weakly supported (see Supplementary Figure S1 for numeric bootstrap values). However, the overwhelming majority of shallow clades relevant for the taxonomic placement of our isolates were sufficiently supported to draw firm conclusions.  The most ancestral clade within our phylogeny contained two taxa (T. aurantiacus and T. argentinensis) and was followed by a three-taxon clade (T. veerkampii, T. louisianensis, and T. caifornicus). The next clade contained four taxa (T. malicola, T. tumui, T. pratensis, and T. domesticus) with others splitting off subsequently. The first taxon to split off was T. aculeatus, followed by T. sparsus and T. panamensis, then a clade containing T. aspriconidius, T. haitouensis, and T. flavus, and another with T. francoe, T. thalandensis, and T. penicilloides. The remaining taxa included in our analysis formed the two youngest clades. The first contained two taxa (T. zhenhaiensis and T. stipitatus) and was placed as a sister to a clade containing all our isolates and different strains of T. pupurogenus (CBS 184.27, CBS 113158, CBS 122434, CBS 108923, CBS 286.36, CBS 113161, and DTO189A1). This larger clade, containing the T. zhenaiensis/T. stipitatus clade plus the T. purpureogenus clade, received substantial support (99%). This also applied to both included subclades, which received 100% support for their monophyletic composition. Similar topologies of our isolates in close proximity to T. purpureogenus were consistently recovered across all single-gene trees (Supplementary Figures S2-S5). Overall, our phylogenetic analysis supports the assignment of our isolated fungi to the species T. purpureogenus as suggested by our ITS DNA barcoding data ( Table 5).

Metabolic Fingerprinting
The extracts showed a broad distribution of compounds over the polarity range ( Supplementary Figures S7-S9). The pairwise cosine similarities of all samples were plotted as a heat map (Figure 4). The main grouping correlated with the solid and liquid media, as expected. The clustering order in the solid media was less influenced by the medium (YES and CYA) and the metabolic features of the solid medium controls were therefore very similar (cosine 0.93). The medium controls and strains B11, B13, B30, and B195 clustered in pairs (consisting of one sample each from CYA and YES). Strains B49 and B69 were merged into one cluster. Strains B18 and B195 clustered together in YES, but the metabolic profile of B18 on CYA was distinct from the other strains in this cluster.
In liquid media, the highest similarity in metabolic features was observed between two pairs -strains B30 and B49 (0.91) and strains B11 and B69 (0.83). Whereas all five strains (B11, B13, B30, B49, and B69) were merged into one cluster, strains B18 and B195 were clustered more closely (0.60) and the metabolic features were more distant from those of the other strains. The clustering order of the strains and the corresponding pairwise cosine similarities are summarized in Supplementary Table S3.

Antimicrobial Activity and Bioactivity-Guided Dereplication
The crude extracts showed no ability to inhibit the growth of any of the Aspergillus species (no zone of inhibition) at a concentration of 0.25 mg of per disc. Extracts B195 and B18 inhibited the growth of P. alvei at an MIC of 0.39 mg/mL, whereas extracts B11 and B69 showed activity only at higher concentrations, with MICs of 6.25 and 3.13 mg/mL, respectively (Table 6). Table 6. Minimal inhibitory concentrations of crude methanol extracts from T. purpureogenus strains tested at 0.001-2 mg/mL against B. subtilis, P. lautus, and P. lactis, and at 0.39-6.25 mg/mL against P. alvei (* using different protocol).  In liquid media, the highest similarity in metabolic features was observed between two pairs -strains B30 and B49 (0.91) and strains B11 and B69 (0.83). Whereas all five strains (B11, B13, B30, B49, and B69) were merged into one cluster, strains B18 and B195 were clustered more closely (0.60) and the metabolic features were more distant from those of the other strains. The clustering order of the strains and the corresponding pairwise cosine similarities are summarized in Supplementary Table S3.

Antimicrobial Activity and Bioactivity-Guided Dereplication
The crude extracts showed no ability to inhibit the growth of any of the Aspergillus species (no zone of inhibition) at a concentration of 0.25 mg of per disc. Extracts B195 and B18 inhibited the growth of P. alvei at an MIC of 0.39 mg/mL, whereas extracts B11 and B69 showed activity only at higher concentrations, with MICs of 6.25 and 3.13 mg/mL, respectively (Table 6). To support the results of the antimicrobial activity against the bee pathogen P. alvei, we expanded the screening panel to include further Paenibacillus species as well as the quality-control strain E. coli ATCC35218 and type strain B. subtilis DSM 10. Interestingly, extract B18 inhibited the growth of P. lautus and P. lactis at MIC = 0.5 mg/mL ( Table 6). The previously observed ability of extract B195 to inhibit P. alvei was also observed against P. lautus and P. lactis, but only at the highest tested concentration (MIC = 2 mg/mL). Extract B195 also displayed moderate activity against B. subtilis at 2 mg/mL. None of the extracts inhibited E. coli at the concentrations we tested. The full dataset is provided in Supplementary Table S4. Based on the above results, extract B18 was used for microfractionation and re-screened against P. lautus, for which optimal growth conditions have been established in our laboratory.
The B18 extract yielded five zones of consecutive fractions that inhibited the growth of the test strain, namely fractions 26-28, 30-31, 33-34, 39, and 114 (Table 7). The fractionation, MS, and MS/MS spectra are provided in Supplementary Figures S10-21 Neither compound aligned with any reference compound spectrum. The predicted molecular formula search of the major ion (C 18 H 32 O 2 ) retrieved >100 candidates in the DNP, including ubiquitous linolenic acid. However, the nonspecific fragmentation did not allow structural assignment. Table 7. Summary of the dereplication of active fractions derived from the methanol extract of the T. purpureogenus strain B18. The structures are shown in Figure 5.

Discussion
The role of non-pathogenic fungi as regulators of pathogens in the hive and in bee bread has been addressed in several studies [32][33][34]. Indeed, bee bread is colonized by many filamentous fungi whose potential as biological control agents is yet to be explored. Here, we characterized seven strains of Talaromyces recently found in honey bee bread and tested the antimicrobial activity of methanol extracts against bacteria associated with EFB (P. alvei) and opportunistic pathogens of the genus Aspergillus.
Although few studies have reported the presence of Talaromyces in bee bread, its presence is not surprising because this ubiquitous genus can be collected by bees from various

Discussion
The role of non-pathogenic fungi as regulators of pathogens in the hive and in bee bread has been addressed in several studies [32][33][34]. Indeed, bee bread is colonized by many filamentous fungi whose potential as biological control agents is yet to be explored. Here, we characterized seven strains of Talaromyces recently found in honey bee bread and tested the antimicrobial activity of methanol extracts against bacteria associated with EFB (P. alvei) and opportunistic pathogens of the genus Aspergillus.
Although few studies have reported the presence of Talaromyces in bee bread, its presence is not surprising because this ubiquitous genus can be collected by bees from various sources [54]. Bees passively or actively collect fungi during foraging and incorporate them into the corbicular pollen [82,83]. Interestingly, some fungi are probably introduced or eliminated by the bees during the collection and storage of pollen [30]. We isolated Talaromyces strains from some of the hives in one rural location, an urban area, and the city periphery. Seasonal factors may be relevant because we found the Talaromyces strains only in spring. Other studies exploring the composition of fungi in bee bread in Europe did not find any Talaromyces species, but samples were collected in the summer [29,84], when the absence of Talaromyces is consistent with our results. In worker bees, the prevalence of fungi was higher in fall and winter [32]. The composition of fungi in bee bread during months with no flight activity has not been studied.
We did not find any Talaromyces species in apiaries located directly on agricultural land. The foraging distance of A. mellifera ranges from several hundred meters to several kilometers, depending on the available pasture [85,86]. The presence of agricultural land (such as a canola field) within the foraging range increases the risk of fungicide residues in the food stores that influence the fungal community. The negative effect of agricultural pressure and fungicide contamination on the bee bread fungal community has been confirmed by sequence-based [29] and cultivation-based analysis [87]. Our results suggest that seasonal effects and agricultural pressure determine the presence of Talaromyces in bee bread. However, more research is needed to correlate the presence and seasonal prevalence of T. purpureogenus and other Talaromyces species in bee bread.
Our in vitro assays revealed that extracts of Talaromyces strains B18 and B195 showed antibacterial activity against the pathogen P. alvei and other tested strains of P. lautus and P. lactis. Bioactivity-guided dereplication in assays against P. lautus revealed multiple fractions and corresponding inhibitory compounds (siderophores, diketopiperazines, and three unknown compounds) that are also likely to show activity against the other Paenibacillus species. Siderophores are metal chelators that facilitate the uptake, intracellular transport, and storage of iron in plants, fungi, and bacteria, making them useful for both medical and environmental applications [88]. Most fungal siderophores are acylated hydroxamates [89], such as the coprogen-type talarazines produced by Talaromyces [90]. The biosynthesis of siderophores plays a key role in fungal virulence and influences their interactions with other microbes [91]. The ability to produce iron-chelating molecules can starve some microbes of this essential nutrient and thus cause growth suppression while providing a source of iron for others, such as yeasts, thus promoting their growth [91]. Several fungal compounds have been associated with regulation of the bee bread microbiome, including organic acids [34] and lovastatin [92]. Given that siderophores inhibited the growth of bacteria in our study, it would be interesting to explore the use of siderophores to regulate EFB and AFB (American foulbrood) in vivo. The unknown active compounds should also be investigated as novel anti-infectives.
Most sequence-based studies identify fungi to the genus or species level. Our results highlight the metabolic diversity of different strains of the species T. purpureogenus. Depending on the strain and cultivation conditions, T. purpureogenus isolated from the bee bread can affect colony health and pathogen resistance in different ways, such as the production of antimicrobial compounds as discussed here, as well as mycotoxins such as the rubratoxins found in our previous study [36]. The genus Talaromyces may consist of an assemblage of species/strains with various metabolic activities, as described for bee-related Aspergillus species [93]. These fungi are found in the bee hive as a mixture of atoxigenic and toxigenic strains, and some of them are opportunistic pathogens [93]. Similarly, strains of other genera associated with bee bread, such as Penicillium and Alternaria [94,95], are assumed to be beneficial but can produce both the antimicrobial compounds and mycotoxins that are lethal to bees. Therefore, the nature of the bee bread mycobiome depends on the balance between different fungal strains, which under ideal circumstances are beneficial to the health of bees and the colony as a whole.

Conclusions
Fungi in bee bread play an important role in the regulation of pathogens, conferring resistance and promoting colony survival. Many bee bread fungi, including T. purpureogenus, can produce protective antimicrobial compounds as well as lethal mycotoxins. The strain-level balance is therefore important for the beneficial function of the mycobiome. Anthropogenic activity, such as the use of fungicides in agriculture, can disrupt this balance and negatively affect colony health.
Supplementary Materials: The following supporting information can be downloaded at: https://www. mdpi.com/article/10.3390/microorganisms11082067/s1, Table S1: Sequences used for phylogenetic analysis. Table S2: Talaromyces spp. isolates from A. mellifera bee bread collected from different locations in South Bohemia. Table S3: Cosine similarities and grouping of the T. purpureogenus strain extracts on different cultivation media. Table S4: Minimum inhibitory concentration (µg/mL) of the crude methanol extracts from T. purpureogenus strains. Table S5: Minimum inhibitory concentration (µg/mL) of the reference antibiotics. Figures S1-S5: Phylogenetic trees. Figure S6