Volatiles from the hypoxylaceous fungi Hypoxylon griseobrunneum and Hypoxylon macrocarpum

The volatiles emitted by the ascomycetes Hypoxylon griseobrunneum and Hypoxylon macrocarpum (Hypoxylaceae, Xylariales) were collected by use of a closed-loop stripping apparatus (CLSA) and analysed by GC–MS. The main compound class of both species were polysubstituted benzene derivatives. Their structures could only be unambiguously determined by comparison to all isomers with different substitution patterns. The substitution pattern of the main compound from H. griseobrunneum, the new natural product 2,4,5-trimethylanisole, was explainable by a polyketide biosynthesis mechanism that was supported by a feeding experiment with (methyl-2H3)methionine.


Introduction
Fungi release a large number of different volatiles that belong to all kinds of natural product classes [1]. Many of these compounds are of interest, because they are markers for the production of fungal toxins and thus can help to distinguish between toxigenic and closely related non-toxigenic species. For example, the sesquiterpene trichodiene (1, Figure 1) is the precursor of the trichothecene family of mycotoxins [2], a class of highly bioactive secondary metabolites that belong to the strongest known inhibitors of protein biosynthesis in eukaryotes [3]. Similarly, the sesquiterpene aristolochene (2) is the parent hydrocarbon of PR toxin [4,5] and has been used as a marker to differentiate between toxin producing and non-producing Penicillium roqueforti isolates [6]. On the other hand, fungal volatiles are interesting, because they contribute with their aroma to the flavour of many edible mushrooms. One of the first identified and certainly most widespread compounds is matsutake alcohol, (R)-oct-1-en-3-ol (3), that is produced inter alia by Tricholoma matsutake [7], a highly sought delicacy in the Japanese cuisine, the bottom mushroom Agaricus bisporus, and the penny bun Boletus edulis [8], as the name indicates a European equivalent to Matsutake in high-class cooking. Volatile organic compounds are also important in the interaction be- tween different species, e.g., between ophiostomatoid fungi and conifer bark beetles that show different behavioural responses to fungal volatiles [9]. Fungal volatiles can also be of importance in the interaction between plants and fungi. In some cases, fungal volatiles seem to be involved in the plant pathogenicity of fungi, as recently observed for 3,4-dimethylpentan-4-olide (4), a volatile from the ash pathogen Hymenoscyphus fraxineus that currently threatens the European ash population [10]. Both enantiomers of this lactone were found to inhibit ash seed germination and to cause necrotic lesions in the plant tissue. In other cases, fungal volatiles can have beneficial effects and may even be involved in the induction of systemic resistance in plants, as can be assumed for 6-pentyl-2H-pyran-2-one (5) that is produced by many fungi from the genus Trichoderma [11,12].
Fungal volatiles can be efficiently analysed by trapping, e.g., on charcoal filters with a closed-loop stripping apparatus (CLSA) that was developed by Grob and Zürcher [13], followed by filter extraction and GC-MS analysis of the obtained headspace extracts [14]. The unambiguous compound identification requires a good match of the recorded electron impact (EI) mass spectrum to a database spectrum and of the retention index, a standardised GC retention factor that is calculated from the retention times of the analytes and of n-alkanes [15], in comparison to an authentic standard or published data. A peculiar problem in the analysis of aromatic compounds with multiple substituents is that constitutional isomers with the same types of substituents, but different substitution patterns often have very similar mass spectra. Furthermore, some of the isomers may also have similar retention indices, and therefore it is mandatory for unambiguous structure elucidation to compare analytes that fall into this class to all the possible isomers. A similar problem can apply to the structural assignment of compounds with multiple stereocentres based on GC-MS data, because the various possible diastereomers usually also produce very similar mass spectra [16], a phenomenon that is also reported for E and Z stereoisomers and can lead to wrong structural assignments, if no authentic standards are used for comparison [17]. We have recently reported on two chlorinated aromatic compounds from an endophytic Geniculosporium sp. [18] and on a series of structurally related phenols, benzaldehydes and anisole derivatives from Hypoxylon invadens [19] that could only be identified with certainty following this approach of extensive compound comparisons. Members of the family Hypoxylaceae are regarded to be extremely rich in secondary metabolites [20], but not much is known about volatiles from these fungi [21]. In continuation of this work, here we present the volatiles emitted by Hypoxylon griseobrunneum MUCL 53754 and Hypoxylon macrocarpum STMA 130423. These strains were selected, because both species released a characteristic and strong odour, as was already mentioned in the literature for H. macrocarpum [22,23], but the nature of the odoriferous compounds remained unknown. As will be shown, the bouquets of both species are composed mainly of highly substituted aromatic compounds whose structures were only securely identifiable by comparison to all the possible constitutional isomers with different ring substitution patterns.

Headspace analysis
The volatiles released by agar plate cultures of H. griseobrunneum and H. macrocarpum were collected using a CLSA [13]. After a collection time of one day the charcoal filter traps were removed and extracted with CH 2 Cl 2 , followed by GC-MS analysis of the obtained extracts. For both strains a large number of compounds from different compound classes including alcohols, ketones, esters, terpenes and pyrazines were identified. Besides the observed minor production of compounds from these classes aromatic compounds dominated, but the patterns were strain-specific.
The mass spectrum of the main compound 24 from H. griseobrunneum ( Figure 4A) showed several fragment ions in the low m/z region typical for an aromatic compound, while the frag-ment ion at m/z = 119 pointed to the loss of a methoxy group from the molecular ion ([M − 31] + ), suggesting the structure of a trimethylanisole for 24. Six constitutional isomers of this compound exist (Table 2). For four of these compounds the corresponding trimethylphenols were commercially available that were O-methylated with methyl iodide and K 2 CO 3 to yield compounds 24a, 24b, 24c and 24e. The other two isomers 2,3,4-trimethylanisole (24d) and 2,4,5-trimethylanisole (24) were obtained by ortho-methylation of 3,4-dimethylphenol (28) via a known procedure [33], followed by HPLC purification of the products 2,4,5-trimethylphenol (29a) and 2,3,4-trimethylphenol (29b) and subsequent O-methylation (Scheme 1). Comparison of the GC retention index of the natural product (I = 1225) to the retention indices of all six standards narrowed the possible structures down to those of 2,4,5-trimethylanisole (I = 1225) and 2,3,5-trimethylanisole (I = 1227), while all other isomers could be ruled out. The final structural assignment of 2,4,5-trimethylanisole for 24 was based on the better matching mass spectrum of this compound in comparison to the alternative of 24c. Compound 24 has not been reported from other natural sources before.
The identification of 24 was further supported by a feeding experiment with (methyl-2 H 3 )methionine. While the methylation pattern of the alternative structure 24c is difficult to understand via a polyketide biosynthesis mechanism, the formation of the assigned structure of 24 by a polyketide synthase (PKS) can be easily rationalised (Scheme 2). The acetate starter unit, bound to the acyl carrier protein (ACP) of an iterative fungal PKS, can be elongated with malonyl-SCoA (mal-SCoA) followed by C-methylation with S-adenosyl-L-methionine (SAM). Two    ylation into 24. In summary, this hypothetical biosynthetic mechanism includes three SAM-dependent methylation steps. A feeding experiment with (methyl-2 H 3 )methionine, the biosynthetic precursor of SAM, resulted in the incorporation of labelling into up to three methyl groups of 24, but not into the fourth methyl group ( Figure 5), which is in line with the biosynthetic model of Scheme 2. Note that because of an isotope effect the isotopomers of 24 can be separated by gas chromatography depending on their deuterium content [34,35], which makes the usage of (methyl-2 H 3 )methionine superior to the usage of 13 C-labelled methionine that would not have led to chromato-graphic separation of the isotopomers. In conjunction with the low incorporation rates obtained here, the results would have been difficult to interpret.
Another trace compound emitted by H. griseobrunneum showed a molecular ion at m/z = 136 and coeluted with exactly the same retention time as a second compound with a molecular ion at m/z = 138. In case of two coeluting compounds the individual compounds are often enriched in the right and left peak flanks, and their individual mass spectra can be extracted by careful background subtraction, but this was not the case here, so only the mass spectrum of the compound mixture was obtained ( Figure 4B). The analysis of the observed fragment ions suggested that the compound with the molecular ion at m/z = 136 may be one of the isomers of dimethylanisole, explaining the fragment ion at m/z = 105 by the loss of the methoxy group ([M − 31] + ), and in agreement with the 14 Da lower molecular ion in comparison to 24. All six isomers of dimethylanisole were commercially available and a comparison of retention indices together with a personal inspection of the mixed mass spectrum of Figure 4B and the mass spectrum of 3,4-dimethylanisole ( Figure 4C) unequivocally identified the natural product as 3,4-dimethylanisole (23, Table 3). Furthermore, the alternative structure of a trimethylphenol was ruled out, because all the isomers eluted later than 23 (Table 3). Interestingly, the elution order of the trimethylphenols is the same as for the corresponding trimethylanisoles with respect to their substitution patterns, and each trimethylphenol consistently elutes slightly later with an increase of the retention index by ca. 30-50 points than the trimethylanisole analogue ( Table 2 and Table 3), which is explainable by the significantly higher polarity of the phenols compared to the anisoles. Compound 23 was recently reported from Euphorbia golondrina [36], but was never observed as a fungal natural product so far.
Biosynthetically, the identified compound 23 can arise by a similar mechanism as discussed for 24, potentially as a minor product of the same PKS, only the C-methylation step in the second round of chain extension needs to be skipped  (Scheme 2). However, during the feeding experiment with (methyl-2 H 3 )methionine the formation of 23 was suppressed, possibly because the additional supply of methionine resulted in a higher efficiency of the programmed methylation steps towards 24.
The additional signals in the mixed mass spectrum ( Figure 4B) at m/z = 138, 123 and 95 that do not originate from 23 are present with similar relative proportions as in the mass spec-trum of veratrole (20, Figure 4D), and indeed a commercial standard of 20 revealed the same retention index of I = 1141 as the natural product, thus confirming the structure of veratrole for the second of the coeluting compounds. Its isomer 1,4dimethoxybenzene (22) and a trimethoxybenzene 21 were also detected. Comparison to all three commercially available isomers of trimethoxybenzene established the identity of 21 as 1,2,3-trimethoxybenzene (Table 4). 1,8-Dimethoxynaphthalene (27) was also found and has been reported previously from other Hypoxylon spp. [19,37]. The corresponding compound 1,8-dihydroxynaphthalene is a known precursor of fungal melanin pigments [38].
Two trace compounds exhibited the mass spectra shown in Figure 4E and Figure 4F that were similar to database spectra of 2,5-dimethyl-p-anisaldehyde (25) and methyl 2,5-dimethyl-panisate (26). The substitution pattern of these compounds is well explained by polyketide biosynthesis logic (Scheme 3). Starting from ACP-bound acetate, two non-reducing elongations with malonyl-SCoA, the first without and the second with C-methylation, followed by another elongation with reduction of the 3-oxo group and cyclisation yields the aromatic system of 25 and 26. Hydrolytic cleavage from the ACP and two methylations of the phenol and the carboxylic acid result in 26, while reductive cleavage and methylation of the phenol give 25. The aldehyde 25 was commercially available and matched the natural product in terms of mass spectrum and retention time.
Compound 25 was transformed into the corresponding methyl ester by treatment with iodine and potassium hydroxide in methanol [39]. The obtained material also showed identical behaviour in the GC-MS analysis to natural 26. Both compounds 25 and 26 are new natural products.
tion of another trace compound from H. macrocarpum as 3-methoxy-4-methylbenzaldehyde (41). The chlorinated compound 2,5-dichloro-1,3-dimethoxybenzene (46) was also rigorously identified by comparison to all possible regioisomers that we had synthesised in a previous study [18]. Interestingly, the sub-stitution pattern for the compound from H. macrocarpum is different to an isomer from the endophyte Geniculosporium sp. that was identified as 1,5-dichloro-2,3-dimethoxybenzene. Compound 46 has not been described as a natural product before. Another trace compound released by H. macrocarpum  exhibited a mass spectrum that pointed to the structure of a dimethoxybenzaldehyde ( Figure 8A). Comparison to all six commercially available isomers (Table 6) showed the identity of the natural product and 3,4-dimethoxybenzaldehyde (42).

Conclusion
Both investigated ascomycetes, Hypoxylon griseobrunneum and Hypoxylon macrocarpum, were found to emit complex mixtures of volatiles, mainly composed of aromatic compounds. As we have demonstrated, for unequivocal structural assignments based solely on GC-MS data it is important to compare the natural product to all possible constitutional isomers with different ring substitution patterns, because the mass spectra of these isomers are too similar to rely solely on MS data for compound identification. Therefore, also the retention index of the natural product must match the retention index of an authentic standard, and usually the retention indices of the isomeric aromatic compounds with different substitution patterns are sufficiently different for a confident structural assignment. Also biosynthetic considerations can help in the structure elucidation, because some aromatic substitution patterns are in line with a polyketide biosynthesis mechanism, while other substitution patterns may be difficult to understand. But such considerations should be made with care and should ideally be supported, e.g., by feeding experiments, as we have conducted in the present study. The main compounds of H. griseobrunneum were 2-methylbutan-1-ol, 1,8-cineol and 2,4,5-trimethylanisole, while H. macrocarpum released a completely different bouquet with the main compounds benzaldehyde, 2-phenylethanol, 4-methylsalicylaldehyde and 3,4-dimethoxytoluene. All these volatiles exhibit a characteristic smell and are likely main contributors to the odour produced by the fungi, but also some of the identified minor compounds may be important for the fungal fragrance. Notably, fungi of the genus Hypoxylon are interesting sources of new natural products, as exemplified by the identification of 2,4,5-trimethylanisole, 2,5-dimethyl-panisaldehyde and its corresponding methyl ester, and 2,5dichloro-1,3-dimethoxybenzene. Therefore, it will be of high interest to investigate the volatiles from further Hypoxylon species in the near future.

Experimental Strains and culture conditions
Hypoxylon griseobrunneum was obtained from a specimen collected in Martinique, Case Pilote, on a trail to Morne Venté on wood and bark of a dead dicotyledon branch in a mesophilic to xerophilic forest, on 25 August 2010 by Jacques Fournier [46]. A voucher specimen is deposited at the herbarium of the

Analysis of volatiles
The volatiles emitted by agar plate cultures of H. griseobrunneum and H. macrocarpum were collected through a closedloop stripping apparatus (CLSA) [13] for ca. 1 day at room temperature and under natural light-dark rhythm. The CLSA charcoal filter traps were extracted with CH 2 Cl 2 (50 μL, HPLC grade), followed by analysis of the extracts by GC-MS. Synthesis of 2,4,5-trimethylphenol (29a) and 2,3,4-trimethylphenol (29b) Diiodomethane (2.14 g, 8.0 mmol, 2 equiv) was dissolved in dry toluene (3 mL) under an argon atmosphere and the solution was cooled to 0 °C. To the vigorously stirred solution, Et 2 Zn in toluene (5.0 mL, 1.2 M, 6.0 mmol, 1.5 equiv) was added rapidly, followed immediately by the addition of 3,4dimethylphenol (500 mg, 4.0 mmol) in toluene (3 mL). The reaction mixture was stirred at 0 °C for 5 min and then under reflux for 1.5 h. The reaction mixture was cooled to 0 °C and then quenched with an aqueous solution of NaHCO 3 (10% w/w). The aqueous phase was extracted with diethyl ether for three times and the combined organic layers were dried over MgSO 4 . The solvent was removed under reduced pressure and the crude product was purified by column chromatography on silica gel (cyclohexane/ethyl acetate 5:1). The obtained product contained 29a and 29b as a mixture which was separated by HPLC (KNAUER Wissenschaftliche Geräte GmbH, Berlin, Azura; DAICEL Chiralpak IA column, 5 μm, 4.6 × 250 mm; hexane/2-propanol 95:5; retention times: 9.66 min (29b) and 10.89 min (29a)). The pure products were obtained as colourless liquids.

Synthesis of trimethylanisoles 24 and 24a-e
To a solution of the respective phenol derivative (23f-k, 15.0 mg, 0.11 mmol, 1 equiv) in dry DMF (2.2 mL), K 2 CO 3 (15.2 mg, 0.11 mmol, 1 equiv) was added and the mixture was stirred at room temperature for 30 min. Methyl iodide (31 mg, 0.22 mmol, 2 equiv) was added and the reaction mixture was stirred at room temperature overnight. The reaction was quenched by addition of water and the aqueous phase was extracted three times with EtOAc. The combined organic layers were dried over MgSO 4 and the solvent was removed under reduced pressure. The crude product was purified by column chromatography on silica gel (cyclohexane/ethyl acetate 20:1).