Unveiling the fungal diversity and associated secondary metabolism on black apples

ABSTRACT Black apples are the result of late-stage microbial decomposition after falling to the ground. This phenomenon is highly comparable from year to year, with the filamentous fungus Monilinia fructigena most commonly being the first invader, followed by Penicillium expansum. Motivated by the fact that only little chemistry has been reported from apple microbiomes, we set out to investigate the chemical diversity and potential ecological roles of secondary metabolites (SMs) in a total of 38 black apples. Metabolomics analyses were conducted on either whole apples or small excisions of fungal biomass derived from black apples. Annotation of fungal SMs in black apple extracts was aided by the cultivation of 15 recently isolated fungal strains on 9 different substrates in a One Strain Many Compounds (OSMAC) approach, leading to the identification of 3,319 unique chemical features. Only 6.4% were attributable to known compounds based on analysis of high-performance liquid chromatography–high-resolution mass spectrometry (HPLC–HRMS/MS) data using spectral library matching tools. Of the 1,606 features detected in the black apple extracts, 32% could be assigned as fungal-derived, due to their presence in the OSMAC-based training data set. Notably, the detection of several antifungal compounds indicates the importance of such compounds for the invasion of and control of other microbial competitors on apples. In conclusion, the diversity and abundance of microbial SMs on black apples were found to be much higher than that typically observed for other environmental microbiomes. Detection of SMs known to be produced by the six fungal species tested also highlights a succession of fungal growth following the initial invader M. fructigena. IMPORTANCE Microbial secondary metabolites constitute a significant reservoir of biologically potent and clinically valuable chemical scaffolds. However, their usefulness is hampered by rapidly developing resistance, resulting in reduced profitability of such research endeavors. Hence, the ecological role of such microbial secondary metabolites must be considered to understand how best to utilize such compounds as chemotherapeutics. Here, we explore an under-investigated environmental microbiome in the case of black apples; a veritable “low-hanging fruit,” with relatively high abundances and diversity of microbially produced secondary metabolites. Using both a targeted and untargeted metabolomics approach, the interplay between metabolites, other microbes, and the apple host itself was investigated. This study highlights the surprisingly low incidence of known secondary metabolites in such a system, highlighting the need to study the functionality of secondary metabolites in microbial interactions and complex microbiomes.

B lack apples represent an unconventional ecological system: dynamic and never quite reaching a state of homeostasis, but following a development path that recurs year-on-year (1,2).These apples harbor a limited array of filamentous fungi and yeasts, following a succession of fungal species over time.Very little is known about the chemical ecology involved in the ensuing microbial interactions and in particular which secondary metabolites are produced and what role they play in developing the system.Such rotten or moldy apples are often removed from apple plantations, as well as from apple trees in private gardens.Of special concern is the potential growth of Monilinia species, including species such as M. fructigena, M. fructicola, M. laxa, M. yunnanensis, and M. polystroma (3,4).Subsequent Monilinia conidium production can give rise to the decay of further apples, if insects, birds, and other animals rupture the apple skin (5).When these apples are left on the tree or the soil they will give rise to the accumulation of Monilinia conidia.The development into black apples usually occurs after infection with several more fungal species after being left on the ground for some time.Even though black apples are not fit for human consumption, they constitute an interesting habitat where other microorganisms and animal species may live.In Nordic countries such as Denmark, apples infected with M. fructigena will often also be infected with Penicillium expansum (blue mold rot), and other species of filamentous fungi.Species of yeasts may also thrive in apples because of the facile availability of monosaccharides and disaccharides (6,7).The low pH of apples, caused primarily by malic acid, will select for acidophilic filamentous fungi, which often produce specialized proteins, such as pectinases, glucose oxidase, and hydrophobins (7,8), that play major roles in apple colonization.This may also result in the release of carbon sources that filamentous fungi can exploit for growth.For instance, Monilinia spp.produce pectinases (pectin lyases and polygalacturonases), proteases, xylanases, α-glucosidases, and some species produce cutinases (9).P. expansum often grow on apples (with or without Monilinia growth), and is well known for its production of the mycotoxin, quorum-sensing inhibitor, and antibiotic, patulin (10,11).Several other species of Penicillium have been reported to grow on apples, including P. crustosum giving rise to a substantial rot, while species such as P. solitum and P. polonicum can cause minor rots in apples (12)(13)(14).Apart from the rot on the surface of apples, other species of filamentous fungi can cause dry rot diseases in apples (at the core) including Fusarium, Alternaria, and Trichoderma species (6).Some of these fungi can inhibit the apple scab disease fungus Venturia inequalis (15).However, these endophytic fungi do not produce a soft apple rot.Other species associated with rot of pomaceous fruits have included Talaromyces minioluteus and Talaromyces rugulosus (16) in addition to Trichothecium roseum (pink rot of apples) (17,18).
Metataxonomic studies that have attempted to examine the fungal taxa of apple communities using next-generation high-throughput DNA sequencing to find both culturable and "non-culturable" species have proven to be inadequate for two main reasons (19).Firstly, DNA sequence-based methods were used to detect the spora (all the spores present on the apple surface) rather than the actual apple funga (the fungi growing on apples) (19).Secondly, the fungi reported were mostly reported at the genus level, and thus neither P. expansum nor Monilinia spp.were reported from the apples examined by Shen et al. (19), even though there is a vast literature on P. expansum and Monilinia spp.being the dominant soft rot fungi on the surface of apples.Furthermore, those that were listed as the "dominant" fungi at the species level have rarely, if at all, been reported as pathogens of apples.Whilst other microbiomes are harder to reproduce in the laboratory, this is less of a problem for apples, and it should be feasible to isolate and cultivate all species naturally present.
Whilst some of these fungal secondary metabolites have reported roles in apples, the vast majority do not.Furthermore, because the vast majority of these metabolites were isolated and characterized from One Strain Many Compounds (OSMAC) studies, it is not known which of these metabolites are realized in their natural environment.OSMAC studies involve the cultivation of a strain on multiple media to encourage the differential production of metabolites and metabolite classes.In combining the total pool of metabolites produced across such an extended suite of media, one can get a more comprehensive overview of the metabolic potential of a given strain.To identify speciesspecific secondary metabolites in black apples, we hypothesize that this OSMAC-generated suite of metabolites can be used to query the total black apple metabolome, while simultaneously chemotaxonomically shedding light on the fungal species present without the need for fungal isolation.However, unlike under laboratory conditions, black apples are not comprised of axenic cultures and instead consist of a diverse-albeit on the lower side compared to other microbiomes-array of microbial species all producing their own set of metabolites in response to the ever-changing dynamics and stimuli of the environment.We hypothesize that the identified fungal secondary metabolites present in black apples must play a role in the system, whether that be for synergistic or antagonistic interactions with other species present, or reasons unknown.By comparing the abundance of metabolites as well as their relative localiza tion in black apples, we hope to shed light on the role these metabolites might play in the invasion and control of other microbes on black apples.

Key fungal species isolated from black apples and their associated secondary metabolites
A key aspect of this study revolved around the fungal diversity of black apples.Over the last 30 years, we have isolated and obtained fungal cultures derived from contamina ted apples (Table S1), in addition to a more recent isolation of black apple-associated strains.All fungal strains isolated were accessioned into the Institut for Bioteknologi (IBT) culture collection.The list only contains relatively few Monilinia strains since our general research interest has been focused on the taxonomy and classification of species belonging to the genus Penicillium.Chemical analysis was conducted on black apples collected during three years (2019-2021), with samples A1-A29 corresponding to whole extractions of 29 different black apples collected in the first two years, and A30-A43 corresponding to small excisions (≈1 cm 2 ) of fungal biomass collected across nine apples in the following year, with in some cases, multiple species adjacent to each other on the same apple.From 6 of these more recent black apples, 10 fungal strains were isolated, corresponding to M. fructigena, P. expansum, P. polonicum, P. thymicola, T. minioluteus, and T. luteus.These fungal isolates were supplemented with an additional five strains comprising M. fructigena and P. expansum, which were either derived from earlier black apple collections or have publicly available genome sequences.

Black apple-associated fungi produce a huge diversity of secondary metabo lites on artificial substrates
In order to address our question regarding the realized potential of fungal secondary metabolites in black apples, we needed to first address the question as to what metabolites or classes of metabolites are produced axenically.Hence, the resulting 15 fungal isolates were profiled metabolically on an extended suite of media (nine substrates) in an OSMAC approach (Table 1).The resulting metabolite profiles were used as an indicative pool of fungal secondary metabolites to query the whole black apples and apple-derived fungal excisions to determine which fungal secondary metabolites are important for invasion and control and to chemotaxonomically determine the species present in black apples.
We performed metabolomics analyses according to the protocol outlined in Fig. 1.Metabolomics studies rely on the chromatographic alignment of chemical featuresunique ion adducts such as [M+H] + and [M+Na] + (at least for the positive ionization mode) as detected in the mass spectrometer-generally relate to a single metabolite in an extract.Since more than one adduct can result from a single metabolite, the number of chemical features detected is usually higher than the number of metabolites in a given extract.When collected, MS/MS fragmentation data supplements these features but does not change the overall number determined at the MS level.Additionally, some metabolites (namely organic acids) may elute over longer than typical retention time ranges, which may artificially inflate the number of features detected based on the retention time tolerance set (in this case, 0.1 min), before being characterized as a new feature.Retention time drift can also occur naturally depending on the state of the instrument and column but can be controlled by running all samples concurrently with appropriate quality controls interspersed.
The OSMAC experiment here revealed 3,319 unique features associated with the 15 fungal isolates tested following the removal of features associated with blanks or primary metabolism.Of these features, 213 (6.4%) were attributable to known compounds as identified by a combination of our in-house library using the Agilent MassHunter PCDL manager, Global Natural Products Social (GNPS) Molecular Networking, and CSI-FingerID results.From this pool of features, it was found that P. expansum represented the most talented producer of secondary metabolites, as distinguished by 554 features uniquely present in at least three samples across the four strains and nine media tested.It was determined that to minimize the potential for misidentification of features associated In addition to the previously reported classes of chloromonilinic acids (33, 35-37) (xanthones) and monilidiols (44) (partly-reduced monoaromatic octaketides) from Monilinia spp., the production of an abundant, and unidentified, suite of secondary metabolites was observed (Fig. S1 to S5).These included the in silico predicted gluconic acid-derived lipids (ID: 5675, C 18 H 36 O 6 ; ID: 10766, C 26 H 52 O 6 ; ID: 11025, C 25 H 50 O 5 ).We also observed the diketopiperazine, cyclo(Phe-Phe) (ID: 5762), from P. expansum (Fig. S6).

A combined targeted and untargeted metabolomic approach highlights a high diversity of fungal secondary metabolites detectable in black apples
In order to determine the chemical diversity of fungal secondary metabolites in black apples, several data processing steps were employed.Following feature detection, dereplication, and alignment of chemical features across the full data set (both extracts of black apples and apple-associated fungal isolates), and the removal of features corresponding to the uncultivated media, uninfected apples and primary metabolism, 4,540 unique features were observed.Of the 1,606 features observed in the black apples, 519 (32%) features could be directly assigned as being fungal-derived, due to their presence in the OSMAC-based metabolic profiling of the apple-associated fungal isolates.The remaining 1,087 (68%) features observed in the black apples corresponded to natural apple metabolites, and fungal secondary metabolites only produced on apples or because of specific microbial interactions or degradative mechanisms.There were 90 (17%) features common to both the whole black apples and excisions, with 216 (42%) features unique to the whole black apples and 213 (41%) features unique to the excisions of black apples, indicating the complementary nature of the two datasets in providing a more exhaustive suite of fungal secondary metabolites in black apples.

FIG 1
Workflow schematic for the metabolomics analysis of black apples.(1) Black apples were initially collected, (2) fungal strains were isolated, (3) fungal strains were cultivated on several media substrates, (4) agar plugs were collected for OSMAC study, and excisions collected of fungal biomass on black apples, followed by extraction using EtOAc:IPA (1:3 vol/vol) + 1% HCOOH.Extraction was also conducted on whole black apples using the same solvent system.( 5) The extracts were subjected to liquid chromatography-electrospray ionization-high-resolution mass spectrometry (LC-ESI-HRMS/MS) analysis in the positive ionization mode and processed in MZMine 3 for feature finding and alignment of samples.( 6) Feature-based molecular networking was conducted using Global Natural Products Social (GNPS) Molecular Networking, followed by (7) in silico classification of compound classes using the CANOPUS package in SIRIUS, with CSIfingerID annotations computed also, and finally (8) all annotations merged into a unified table, compound class annotations propagated according to GNPS clusters, primary metabolites filtered out, and data analysis performed on fungal secondary metabolites (SMs) identified in black apples.
Of the features common to both the black apples and the apple-associated fungi, 27 known compounds were identified, with the first 7 (26%) assigned using our in-house PCDL MS/MS fragmentation database, a further 8 (30%) using the GNPS pipeline or through the propagation of known classes within a GNPS cluster, one (4%) was solely predicted with CSIfinger ID in addition to supplementing the identification of three other compounds previously annotated with the PCDL and GNPS tools, highlighting its validity in reliable prediction of metabolites.Lastly, the remaining 11 (40%) com pounds were identified using manual approaches with reference to the literature such as comparison of MS/MS spectra, if available, or HRMS data for compounds known to be produced by a given species (Table 2).Whilst the identification provided by MS/MS fragmentation matching is not as good as compared to either the isolated compound by NMR or the retention time of the standard under identical HPLC conditions, the tentative assignments ascribed here are of a very high level (B), according to the metabolite annotation guide proposed by Alseekh et al. (45) which we have adapted for this report.The remaining seven (13%) features were annotated to known compounds known to be produced by the producing species by HRESIMS level annotation and/or supplemen ted by investigation of isotopic patterns.This is the lowest level (D) of identification but combined with reported chemistry previously isolated in each species is generally sufficient confirmation.Targeted analysis of the important P. expansum mycotoxin, patulin, suggested its absence in naturally infected black apples, despite being present in detectable quantities in post-harvest infected apples.Further efforts were made in order to minimize the potential methodological hindrances toward the detection of patulin, namely increasing the concentration of specific apple extracts wherein P. expansum was isolated, increas ing the injection volume, and changing the ionization mode.Instead, the biosynthetic precursor of patulin, and its isomer, (E)-and (Z)-ascladiol were observed (46).These derivatives are known biotransformation products of patulin by M. fructigena (47), and by certain yeasts and bacteria (48,49).Hence, due to the late temporal state and altered microbiome of black apples, any produced patulin seems to have been decomposed.

M. fructigena produces mainly non-ribosomal peptide synthase (NRPS)derived compounds on black apples
Due to the vast pool of unidentified secondary metabolites detected, categorization of the specific classes of compounds being produced on black apples was computed, first by using SIRIUS (50) to predict formulae with the ZODIAC tool (51) followed by prediction of the compound class with the CANOPUS tool (52-54) based on compari son of MS/MS fragmentation patterns (Fig. 2).Surprisingly, M. fructigena was shown to produce mostly NRPS-derived compounds, with small peptides accounting for 43% of the features, followed by oligopeptides (22%) rather than the often reported two major types of polyketides.The suite of identified features identified in black apples from M. fructigena included the described diketopiperazines, and some gluconic acid-derived lipids.Compound classes for the other two genera, Penicillium and Talaromyces, were more diverse including sesquiterpenoids (e.g., berkedrimane B), tryptophan alkaloids (e.g., communesins), and cyclic polyketides (e.g., ascladiols).
In addition to comparing the differences in compound classes produced by the different fungal genera on black apples, the differences in compound classes between the different black apple samples were also compared (Fig. 3).Notably, the presence of certain compound classes differed between the whole black apples (A1-A29) and the excisions of black apples (A30-A43), with cyclic and polycyclic polyketides (such as ascladiol and chloromonilinic acid B produced by P. expansum and M. fructigena, respectively) only detectable in the former, and anthranilic acid alkaloids only detectable in the latter.The differences noted here could be due to the localized concentration differences between secondary metabolites or the presence of potentially different fungal species based on the different locations from which the black apple samples were derived.

Differential abundances of fungal secondary metabolites indicate those ecologically important for the invasion and control of black apples
The potential importance of certain secondary metabolites to the black apple system was ascertained by comparing the average normalized abundances produced on black apples to those produced in the laboratory on standard cultivation media.Normalization was performed for each sample according to the maximum peak area in each total ion chromatogram.Interestingly, whilst it was found that known secondary metabolites tended to be produced at equivalent levels or were suppressed in the black apples compared to that observed by the OSMAC approach (Table 3), there were some notable exceptions.Namely, an analog of the antifungal compound, berkedrimane B (55), produced by T. minioluteus was found to be 28 times higher in black apples on average compared to standard cultivation media.Being slightly more non-polar than berkedrimane B, this analog could correspond to a preferred isomer being produced, or a natural decomposition product that has been enhanced in this system.Pinselin and moniliphenone, biosynthetic precursors to the antifungal compound, chloromonilicin (37), by M. fructigena were also 15-25 times more abundant in black apples.Chloromoni licin and chloromonilinic acid A were identified in 9 and 20 of the black apple extracts, respectively, despite not being observed for the OSMAC study, at least for the positive ionization mode.Considering their abundance is significant enough to be detected here suggests their involvement as antifungal compounds is prominent.In contrast, for P. expansum, we do not observe the production of patulin, perhaps the most famous mycotoxin produced by the fungus.However, the biodegradation product of patulin, (Z)-ascladiol, is prominent and seven times more abundant in black apples compared to the OSMAC study.This suggests that the antifungal compound, patulin, was indeed  present at some point prior, and in relatively high amounts compared to that seen in the laboratory, highlighting the ecological push for its production in this setting.Here we can see an interesting case where antifungal compounds are more highly expressed and conversely degraded as a means of control and even protection.Fungal excisions were also compared as a means of investigating the potential microbial interactions in a defined spatial resolution of a given black apple, with for example the adjacent A32 (excision of M. fructigena) and A33 (excision of T. miniolue tus) highlighted in Fig. S9.In this case, it was found that the major differences in the abundance of produced secondary metabolites between excisions were mainly observed for unidentified metabolites.It is tempting to speculate that such unidentified com pounds are related to biosynthetic gene clusters, only expressed in the natural system; however, more investigations are needed to prove this hypothesis.Further differences observed for the secondary metabolites identified in the fungal excisions are likely due to the heightened concentration per unit of apple mass, differences in diffusibility and chemical behavior, and reduced capacity for natural chemical decomposition when compared to the whole black apple extracts.

Metabolic profiling of black apples distinguishes the presence and temporal succession of fungi
Whilst it was observed that M. fructigena was always the first key microbial species observed in the formation of black apples, which species established themselves next and in what order, if indeed there was one, was something we wanted to address in this study.Using chemotaxonomic indicators as a means of identifying species, we found that of the 43 black apple extracts obtained in this study, there were 30 instances where fungal secondary metabolites belonging to one of the key species investigated by the OSMAC approach could be identified (Fig. 4), based on the scoring system applied here.Unsurprisingly, metabolites of M. fructigena signified the bulk of the black apples, with 26/30 indicating its presence.This gives credence to the speculated succession order for the invasion of fungal species giving rise to black apples, with M. fructigena generally observed as the initial colonizer of apples, indicated by brown ringlets on their surface.The next most observed fungal species based on their chemical "footprints" were P. expansum and T. minioluteus, with metabolites detected in 21 and 20 of the black apples extracted, respectively.To find the presence of secondary metabolites belonging to T. minioluteus in such a high proportion of the black apples was indeed surprising, given that its reported isolation frequency is much lower.This is followed by P. thymicola, P. polonicum, and T. luteus.The remaining 13 black apples where no determinable fungal species could be identified are of particular interest, with possible biotransformations of known chemistry, inductions of novel chemistry from known competitors, or the potential presence of more obscure and possibly uncultivable species.However, it is noted that it was evident that the two known metabolites, pyrrolocin A and berkedri mane B, produced in our queried species were identified but collectively with other metabolites did not meet our threshold of six indicative metabolites needed to be present for the confirmed presence of a species in each black apple.Choosing a lower threshold of three indicative features, for instance, altered the detected frequency for metabolites belonging to M. fructigena, where only three samples indicated no prior growth (data not shown).
Interestingly, there was a single black apple, A24, that showed evidence of secon dary metabolites belonging to all six of the fungal species investigated by the OSMAC approach.LC-MS/MS analysis of this apple indicated it to have the highest degree of chemical complexity, with the chemical extract containing 311 features pertaining to secondary metabolites, approximately threefold higher than the mean (109 ± 63), and indeed, far higher than that pertaining to the mean of the black apples where no OSMAC-derived fungal secondary metabolites were detected (81 ± 40).The major metabolite present could be a possible flavonoid with the predicted formula, C 15 H 10 O 8 , which we highlight here due to its absence in any of the blank media or natural apple substrates, and hence could be a biotransformation product.Alternatively, we note that the compound may be instead of fungal origin, with the detection of this compound observed for our single strain of T. luteus on rice substrates.However, rice also contains flavonoids, so the possibility of biotransformation cannot be ruled out.
Other fungal metabolites not detected in the 15 strains selected for OSMAC analysis were detected in black apples, namely mycophenolic acid, a highly biologically active metabolite produced by Penicillium brevicompactum and other fungi (56)(57)(58), aurofu sarin, from the wheat pathogen Fusarium sp. ( 59), the structurally-related xanthoepocin, a common metabolite from the Penicillium genus (60, 61), asperphenamate, derived from several Aspergillaceae (62-66), and berkedrimane A originally described from a P. solitum strain (67).However, we speculate that the latter strain was misidentified since we have never found berkedrimanes to be produced by any P. solitum strain (68).Instead, we have consistently detected berkedrimanes from the Talaromyces species T. amestolkiae, T. purpureogenus, and T. trachyspermus (69).Interestingly, the presence of mycophenolic acid, xanthoepocin, and asperphenamate together, are key indicators that P. brevicompactum may have been present on six of the apples (A3, A6, A7, A23, A25, and A27).Notably, all six of these apple extracts are from whole black apples.Furthermore, the presence of aurofusarin, like in A24 and A38, could indicate chemical footprints of the dry rot fungal plant pathogens, Fusarium sp., species often also commonly associated with the plant leaves or stems.

Networking using GNPS identifies ecologically important secondary metabolite inductions, decompositions, and biotransformations in black apples
To address the vast unknown suite of secondary metabolites in black apples, we hypothesized that the biotransformation, induction, or degradation of several other fungal secondary metabolites was occurring.Networking using GNPS is a useful tool for identifying such molecular transformations since analogs will cluster together due to similarities in MS/MS fragmentation.One example of potential biotransformation can be seen for the identified isochromenone ID: 2983, C 10 H 10 O 5 produced by P. thymicola on jasmine rice, which shares a cluster with a dehydrated analog ID: 3699, C 10 H 8 O 4 only observed in the black apples (Fig. 5A).Reductions of fungal secondary metabolites were also very common, with conversion of ID: 2015, C 15 H 18 O 3 produced by P. expansum only on basmati rice to ID: 5363, C 15 H 20 O 3 in the black apples (Fig. 5B) and the reduction of ID: 14961, C 34 H 56 O 9 produced by T. minioluteus also only on basmati rice to ID: 15389, C 34 H 58 O 9 in the black apples (Fig. 5C).
Whilst the aim here was to focus more on the microbial secondary metabolism associated with black apples, we would be remiss to ignore the secondary metabolism of apples in general.Common apple secondary metabolites that were observed included (+)-catechin, (−)-epicatechin, quercetin, kaempferol, ursolic acid, chlorogenic acid, and other flavanols, phenolic acids, terpenes, carotenoids, and organic acids (70).No obvious microbial biotransformations of known apple secondary metabolites were detected.

DISCUSSION
In this study, the microbial secondary metabolite chemical diversity in black apples was investigated by linking features present in apple-associated fungal isolates grown on standard cultivation media to those detectable in black apples.In so doing, we were able to chemotaxonomically predict which species might have been present in each black apple and link the produced secondary metabolites to potential ecological roles in the invasion or control of a given black apple.Here, it was observed that at least one of our six species selected for OSMAC experiments was present in 30/43 of the black apple extracts examined, based on the criteria of detection of at least six representative chemical features produced by any of our key species in each black apple extract.The likely presence of additional untested fungi, such as Fusarium sp. and P. brevicompactum, can also be seen as determined by the presence of indicative metabolites.Although M. fructigena and P. expansum were confirmed to be the two most prolific fungal species present on black apples among those tested, the frequency of detection of metabolites associated with T. minioluteus (20/43 of the black apple extracts) was higher than expected based on the single observation reported in the literature (16).This could be due to the low number of reports associated with black apples, or it could be a phenomenon more geographically linked to apples produced in Denmark (as indicated by the frequency of isolation in Table S1).Thus, we have shown individ ual microbial and chemical diversity originating from black apples to be understated, indicating a complex network of chemical cues important for establishment, control, and homeostasis.Because of this relatively high observed chemodiversity, the black apple model system represents an interesting environment for the study of the ecological role of secondary metabolites.
Whilst the exact ecological role of specific secondary metabolites is difficult to discern from such a generalized study, it can be seen that many of the known metab olites-which also tend to correspond with those that are highly represented on standard cultivation media-are present in black apples at significantly lower levels than expected, highlighting the specificity of the apple substrate and the interactions between the indicated microbial partners to be highly influential to the expressed apple microbiome secondary metabolome.However, the antifungal, antifungal precursor compounds, or antifungal degradation products, (Z)-acladiol, pinselin, moniliphenone, and a berkedrimane B analog, are present in much higher relative abundances in black apples, highlighting them or their related analogs as potentially important compounds for invasion and control of black apples by the producing fungi.In general, our results indicate that the production of secondary metabolites on black apples is influenced by certain environmental cues (such as low pH), stimuli, and limitations due to the added fight for resources with other competitors present.
One possible explanation for the suppressed production of certain metabolites is that they are being detoxified.Antioxidants, for instance, are known to assist in the detoxifica tion of potentially harmful molecules (71) and are highly abundant in the form of plant flavonoids.This is perhaps reflective of a dual role for plant flavonoids, such as catechin found in apples, for both antibiosis and detoxification of fungal secondary metabolites (72).Previous studies have shown that apples high in procyanidins, dihydrochalcone, flavonols, and hydroxycinnamic acids were more resistant to infection by P. expansum (73).When mycotoxin producers are treated with phytochemicals, the production of mycotoxins is significantly reduced.In Aspergillus flavus, for instance, aflatoxin produc tion is decreased by up to 99.8% with the addition of hydrolyzable tannins, flavonoids, and phenolic acids derived from pine nuts (74).Hence, we hypothesize that the presence of natural antioxidants in apples is likely an influential reason for the lower-than-expec ted metabolic output for known fungal secondary metabolites in black apples.
Not only are fungal secondary metabolites influenced by plant phytochemicals but also the secondary metabolites of other competing species.For instance, mycotoxins are known pro-oxidants and have been shown to decrease the concentration of antioxidants and other biomolecules in some systems (75,76).Whether this is a result of the increased production of reactive oxygen species (ROS), direct oxidation, or both, is something that has not been well studied in natural systems.Even for antioxidants, scavenging ROS has been shown to alter the chemical state, and sometimes even bioactivity, of other secondary metabolites (77).Furthermore, the chemical fate of secondary metabolites can be even more complex, with the low pH of apples (often around pH 3) causing chemical degradation and biotransformations between interacting species.Such biotransforma tions have also been linked to detoxification mechanisms, and are most commonly observed for antibiotic resistance (78) and quorum quenching (79)(80)(81) mechanisms.Some possible biotransformations of fungal secondary metabolites in black apples have been shown here, with dehydration or reduction most observed, although there is the possibility this is due to the low pH environment of apples.To investigate the mechanism behind these chemical transformations, one could take the purified known compounds and systematically treat them with relevant fungal species, or acidified solutions, and with the aid of GNPS clustering determine if the conversion is enzymatic or non-enzymatic.A recent study by Kang et al. systematically investigated the potential biotransformations of phytochemicals by a panel of fungi using GNPS (82).
Another surprising observation was the high abundance of compounds associ ated with M. fructigena present, including several NRPS-derived diketopiperazines and gluconic acid-derived lipids.Gluconic acid has previously been shown to have a role in the virulence of the related M. fructicola on peaches (83), and also by P. expansum on apples (84), and is a means of locally modifying the pH of their environment.On the other hand, the presence of diketopiperazines perhaps reflects on their antimicrobial and quorum-sensing inhibition roles (85), which is notable given that these metabolites are also highly abundant in black apples.Further investigations should be used to elucidate both the structural diversity and the potential ecological role these compounds might play in the invasion and control of M. fructigena on black apples.
In conclusion, this metabolomics study of black apples has provided many new insights into the presence and potential ecological roles of microbially, and in particu lar fungus-produced secondary metabolites in a natural system.Such a microbiome is highly advantageous for the study of secondary metabolites due to the high chemical diversity detectable from a relatively small selection of acidophilic microbial species, and the largely reproducible succession of microbes that is observed annually.The interactions between the invading microbes are dynamic and the resulting black apple metabolome is a reflection of the key secondary metabolites which are important for invasion, control, and temporal fungal progression within such a system.Whilst many of the known metabolites were shown to be diminutively observed, perhaps due to detoxification and degradative mechanisms of phytochemicals and other highly reduced fungal secondary metabolites, enzymatic processes in interacting species, and the low pH environment, there was also a high repertoire of upregulated metabolites arising from inductions and biotransformations.In general, this study focused on the chemical diversity of realized black apple microbiomes rather than examining the role of any one specific molecule, due mostly to the sheer number of unidentified compounds.However, efforts are underway to isolate and characterize some of these key compounds in an attempt to delve deeper into their biological and ecological roles.Furthermore, it will be important to investigate the microbial interactions of this system on a more simplified level, with particular attention to those molecules produced in dual cultures (86) and synthetic communities (SynComs) as has been recently created for many other microbiomes (87)(88)(89).A potential SynCom that would be important to investigate could include a combination of yeasts and the filamentous fungi, M. fructigena, P. expansum, and T. minioluteus.Lastly, in making this data set available publicly (MassIVE: MSV000092823), we believe that collectively as a community, additional insights can be brought to light.

Isolation of apple-associated fungi
Fungal strains were isolated either by classical collection and transfer of fungal spores from black apples into a defined spore solution, followed by streaking (and potentially re-streaking) onto agar plates, and cultivation at room temperature for 5-7 days, until a pure culture was obtained.Alternatively, fungal cultures were isola ted by crushing mummified apples, followed by spreading on V8 and Czapek yeast extract (CYA) with antibiotics.Spore suspensions were made by harvesting spores using 20% glycerol, and subsequently storing them at −80°C.All cultures are kept in the IBT culture collection (https://www.bioengineering.dtu.dk/research/strain-collections/ibt-culture-collection-of-fungi).

LC-MS/MS sample extraction and preparation
All samples used for extraction were HPLC grade or better.For fungi cultivated on agar, three plugs (≈8 mm in diameter) were added to an Eppendorf tube prior to extraction.For fungi cultivated on grains, grains and mycelia totaling a volume of ≈0.5 cm 3 were added to an Eppendorf tube prior to extraction.For excisions of fungi on black apples, excisions of the fungus and shallow penetration of the undried apple exterior totaling a volume of ≈0.5 cm 3 were added to an Eppendorf tube prior to extraction.In each of these cases, the samples were extracted in isopropanol/ethyl acetate (1:3) + 1% HOOH (1 mL).For extractions of whole black apples, the apples were pureed using an IKA dispenser.If needed, a small volume of Milli-Q water was added during homogenization.Following homogenization, the metabolites were extracted with equal volumes (≈3 mL) of isopropanol/ethyl acetate (1:3) with 1% formic acid in a 15 mL falcon tube.The resulting chemical extracts were sonicated for 30 min, and the supernatants dried under nitrogen.Dried samples were resuspended in methanol (1 mL) and centrifuged before analysis by ultra-high-performance liquid chromatography-diode array detection-mass spectrometry (UHPLC-DAD-MS).

Data-dependent LC-ESI-HRMS/MS analysis
Ultra-high-performance liquid chromatography-diode array detection-quadrupole timeofflight mass spectrometry (UHPLC-DAD-QTOFMS) was performed on an Agilent Infinity 1290 UHPLC system (Agilent Technologies, Santa Clara, CA, USA) equipped with a diode array detector.Separation was achieved on a 150 × 2.1 mm i.d., 1.9 µm, Poroshell 120 Phenyl Hexyl column (Agilent Technologies) held at 40°C.The sample (1 µL) was eluted at a flow rate of 0.35 mL min −1 using a linear gradient from 10% acetonitrile (LC-MS grade) in Milli-Q water buffered with 20 mM formic acid increasing to 100% in 10 min, staying there for 2 min before returning to 10% in 0.1 min.Starting conditions were held for 3 min before the following run.
MS detection was performed on an Agilent 6545 QTOF MS equipped with Agilent Dual Jet Stream ESI with a drying gas temperature of 250°C, a gas flow of 8 L min −1 , a sheath gas temperature of 300°C and flow of 12 L min −1 .Capillary voltage was set to 4,000 V and nozzle voltage to 500 V in positive mode.Mass spectra were recorded as centroid data, at an m/z of 100-1,700, and auto MS/HRMS fragmentation was performed at three collision energies (10, 20, and 40 eV), on the three most intense precursor peaks per cycle.The acquisition rate was 10 spectra s −1 .Data were handled using Agilent MassHunter Qualitative Analysis software (Agilent Technologies).Lock mass solution in 70% MeOH in water was infused in the second sprayer using an extra LC pump at a flow of 15 µL min −1 using a 1:100 splitter.The solution contained 1 µM tributylamine (Sigma-Aldrich) and 10 µM Hexakis (2,2,3,3tetrafluoropropoxy) phosphazene (Apollo Scientific Ltd., Cheshire, UK) as lock masses.The [M+H] + ions (m/z 186.2216 and 922.0098, respectively) of both compounds were used.

LC-MS/MS data processing
The mass spectrometry data were centroided and converted from the proprietary format (.D) to the m/z extensible markup language format (.mzML) using ProteoWizard (version 3.0.22112,MSConvert tool) (90).The mzML files were then processed with the MZmine 3 toolbox (version 3.2.8)(91).The .mzML files, metadata (.txt), MZmine batch file (.XML format), and results file (.MGF) are available in the MassIVE data set MSV000092823.The MZmine batch file contains all the parameters used during the processing.Namely, feature detection and deconvolution were performed with the Automatic Data Analysis Pipeline (ADAP) chromatograph builder (92) and local minimum search algorithm.Isotopologs were then identified and filtered, and the remaining features aligned across samples.To reduce the number of features resulting from the same molecule, duplication filtering and metacorrelate were performed.Ion identity networking was also included to identify adducts belonging to the same molecule but this does not reduce the overall number of features.For the final feature quantification table, a .MGF file was exported in both GNPS (93) and SIRIUS relevant formats as well as a .CSV file.We also provide complete computational workflows for the metabolomics analyses conducted (available on GitHub at https://github.com/michael-cowled/black-apples).

LC-MS/MS data annotation
In order to dereplicate known compounds in the chemical extracts, the raw MS/MS data were searched against the in-house library using Agilent MassHunter PCDL manager (Agilent Technologies) (94).Due to the manual nature of this task, this was only carried out for a randomly selected 10% of the samples.
The remaining annotations were performed using the processed mass spectrometry data (.MGF and .CSV files from MZmine).Firstly, GNPS was performed using the featurebased molecular networking workflow (93).SIRIUS (v 5.7.0) (50) was used to compute molecular formulae, through the comparison of experimental and predicted isotope patterns (95), and computation of fragmentation trees (96), and further improved with the ZODIAC module (51).CSI:FingerID (97,98) was employed to conduct in silico structure annotation utilizing structures from biodatabase, which allowed for compound class categorization with CANOPUS (52)(53)(54).SIRIUS was configured with a mass accuracy of 10 ppm for annotation of the ions [M+H] + , [M+K] + , and [M+Na] + .In the case of ZODIAC, the features were divided into 10 random subsets using GIBBS sampling to alleviate the computational burden.Each subset was computed separately, employing a threshold filter of 0.95 and requiring a minimum of 10 local connections.
In order to identify potential microbially produced secondary metabolites, we gathered annotations from spectral library matching (in GNPS), as well as through querying our in-house PCDL database (94).These annotations were subsequently compared against the Natural Products Atlas v2021_08 (99) and MIBiG v3.1 (100) databases pertaining to secondary metabolites.Compound class annotation for unidentified compounds was appended from NPClassifier for which the NPC super class probability ≥70%.Filtration of primary metabolite classes was carried out on the resulting annotated data set.The workflow is akin to and hence was inspired by, that used in a recent report for the Earth Microbiome Project 500 (EMP500), except for propagation of compound classes using GNPS clusters, as we found this removed several features of which we deemed important (101).

LC-MS/MS data analysis
Following the combination of the above annotations, filtering, and computations, the resulting table of features was linked with the corresponding metadata associated with the samples.Features present in blank media controls were subtracted from the data set if present above a threshold of 1,000.At this point, the data were split by whether the samples were derived from black apples or fungal cultivation of standard laboratory media (OSMAC).The resulting two datasets were compared, focussing on the features present in both.Each apple extract was also queried using the suite of metabolites identified in each species (present in at least three samples associated with the species by OSMAC) in order to assign its presence in a given black apple sample with at least six features associated with a species needed in order to confirm its presence.All plots were generated in R using the ggplot2 package (102).

FIG 2
FIG 2 Proportion of fungal secondary metabolites found in black apples associated with the relevant natural product superclasses (>70% confidence predicted by CANOPUS tool in SIRIUS) for the genera: (A) Monilinia, (B) Penicillium, and (C) Talaromyces.

FIG 3
FIG 3 Proportion of fungal secondary metabolites associated with the relevant natural product superclasses for each black apple, shedding light on secondary metabolite classes of ecological significance in invasion, and/or control of apples.

FIG 4 (
FIG 4 (A) Heatmap showing the presence of each fungal species in each black apple.(B) Bar chart showing the associated frequency of each species present in black apples.

FIG 5
FIG 5 Molecular clusters derived from GNPS, highlighting (A) dehydration and (B and C) reduction of fungal secondary metabolites in black apples.Nodes corresponding to features identified in the OSMAC study are shown in blue, and those in the black apples are shown in red.The pie charts show the proportional difference in the sum of peak areas across all samples where the feature is identified.A solid blue or red node means that it is exclusively present in that sample type.The mass difference between two nodes is shown along the edge calculated as the difference between the flat to arrow side.Each node is labeled with the feature ID (inside the circle) and precursor mass (outside the circle).

TABLE 1 List of black apple fungal isolates for which chemical profiling was performed Genus Species Strain Year of isolation Country of isolation Source of isolation
a Morphologically and chemotaxonomically similar.b Selected due to publicly available genome sequence.

TABLE 2
Annotations of detected fungal secondary metabolites identified in black applesID t R (min) Putative metabolite name Molecular formula Adduct Theoret.m/zExp.m/z a Identification level (A, B, C, D): (A) standard or NMR; (B) confident match based on MS/MS; (C) confident match using in silico MS/MS approaches or partial match based on MS/MS; and (D) MS only.bPCDL.cGNPS.d CSI-FingerID.eLiterature/manual dereplication.

TABLE 3
Ratio of the average normalized abundance of the most differentially present (five times higher or lower in apples) known fungal secondary metabolites identified in black apples compared to that produced in the OSMAC study