New Untargeted Metabolic Profiling Combining ... - ACS Publications

Jul 31, 2013 - UMR Biologie Végétale et Microbiologie Environnementale, CNRS, FR 13108 Saint-Paul-les-Durance, France. ⊥. Université d,Aix-Marsei...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/ac

New Untargeted Metabolic Profiling Combining Mass Spectrometry and Isotopic Labeling: Application on Aspergillus fumigatus Grown on Wheat Patricia M. Cano,†,‡ Emilien L. Jamin,†,‡ Souria Tadrist,†,‡ Pascal Bourdaud’hui,†,‡ Michel Péan,§,∥,⊥ Laurent Debrauwer,†,‡ Isabelle P. Oswald,†,‡ Marcel Delaforge,∇ and Olivier Puel*,†,‡ †

INRA, UMR 1331, Toxalim, Research Center in Food Toxicology, FR 31027 Toulouse, France Université de Toulouse, INP, 4-6 Allée de Monso, FR 31400 Toulouse, France § Groupe de Recherches Appliquées en Phytotechnologie, CEA, IBEB, Cadarache, FR 13108 Saint-Paul-les-Durance, France ∥ UMR Biologie Végétale et Microbiologie Environnementale, CNRS, FR 13108 Saint-Paul-les-Durance, France ⊥ Université d’Aix-Marseille, FR 13007 Marseille, France ∇ CNRS, URA 2096, CEA Saclay, FR 91191 Gif sur Yvette, France ‡

S Supporting Information *

ABSTRACT: Characterization of fungal secondary metabolomes has become a challenge due to the industrial applications of many of these molecules, and also due to the emergence of fungal threats to public health and natural ecosystems. Given that, the aim of the present study was to develop an untargeted method to analyze fungal secondary metabolomes by combining high-accuracy mass spectrometry and double isotopic labeling of fungal metabolomes. The strain NRRL 35693 of Aspergillus fumigatus, an important fungal pathogen, was grown on three wheat grain substrates: (1) naturally enriched grains (99% 12C), (2) grains enriched 96.8% with 13C, (3) grains enriched with 53.4% with 13C and 96.8% with 15N. Twenty-one secondary metabolites were unambiguously identified by high-performance liquid chromatography−high-resolution mass spectrometry (HPLC−HRMS) analysis. AntiBase 2012 was used to confirm the identity of these metabolites. Additionally, on the basis of tandem mass spectrometry (MSn) experiments, it was possible to identify for the first time the formula and the structure of fumigaclavine D, a new member of the fumigaclavines family. Post biosynthesis degradation of tryptoquivaline F by methanol was also identified during HPLC− HRMS analysis by the detection of a carbon atom of nonfungal origin. The interest of this method lies not only on the unambiguous determination of the exact chemical formulas of fungal secondary metabolites but also on the easy discrimination of nonfungal products. Validation of the method was thus successfully achieved in this study, and it can now be applied to other fungal metabolomes, offering great possibilities for the discovery of new drugs or toxins.

M

genotoxic substances of natural origin.10,11 In the past decade, exposure to mycotoxins has become a major concern worldwide in terms of human and animal health safety.3,12 Given their therapeutic and/or toxic effects, there is a growing interest in fungal metabolomics. Results of metabolomic studies are gathered in specific databases, such as AntiBase, whose latest version includes information on over 40 000 secondary metabolites from microorganisms and higher fungi.13 However, in 2008 over 97 000 fungal species were already described in the dictionary of fungi,14 and recent advances in the sequencing of fungal genomes do not correlate

icroorganisms are ubiquitous in our environment due to their great adaptability to different habitats. However, much less attention has been paid to fungi compared to bacteria and viruses, despite the fact that as much as 50% of the metabolites used by the pharmaceutical, medical, and agro− alimentary industries are of fungal origin.1,2 On top of that, fungi represent an increasing risk for food security as they are responsible for the loss of over 125 million tons of crops,3,4 and also a risk for biodiversity, since they have recently caused massive animal die-offs.5,6 As most microorganisms, fungi can produce a wide diversity of secondary metabolites, among which several compounds, known as mycotoxins, can induce deleterious effects directly on their natural hosts, but also indirectly through food and feed on humans and animals.7−9 Aflatoxins, for instance, stand among the most carcinogenic and © 2013 American Chemical Society

Received: June 18, 2013 Accepted: July 31, 2013 Published: July 31, 2013 8412

dx.doi.org/10.1021/ac401872f | Anal. Chem. 2013, 85, 8412−8420

Analytical Chemistry

Article

with an increase in the identification of new metabolites.15 It is thus presumable that the vast majority of fungal secondary metabolites still remain uncharacterized. Currently, one of the main technologies used for the untargeted analysis of metabolites is mass spectrometry (MS), which measures the mass-to-charge ratio of ions.16 Highresolution MS techniques provide precise structural information able to confirm the detection of already known compounds and also to obtain structural information on new metabolites.17,18 In order to overcome the drawbacks of directly injecting complex samples (mainly chemical isomers discrimination and matrix effects), separation techniques such as highperformance liquid chromatography (HPLC) are usually coupled to MS detectors. One of the main difficulties in the identification of metabolites lies in their small size, because they can be readily mistaken with contaminants from solvents, reagents, or other material used during sample preparation and analysis. Besides, a single metabolite can be detected as several different ions during in-source fragmentation, and this together with possible postbiosynthesis adduct formation gives rise to redundant signals which complicate metabolite identification.17 The method presented in this study makes use of both HPLC coupled to Orbitrap high-resolution MS (HPLC− HRMS) and uniform isotopic labeling of whole fungal metabolomes using carbon (13C) and nitrogen (15N). The integrated use of these two techniques enables specific detection and highly accurate mass measurement of fungal secondary metabolites. The double isotopic labeling allows formula ambiguities to be overcome by reporting the number of carbon and nitrogen atoms and also to discriminate adducts formed with solvents which can distort the chemical formula of biosynthesized ions. Isotopic labeling and, more particularly, stable isotopes have been widely used in metabolomics for pathway elucidation and in flux analysis but not so much for chemical formula determination.18−20 In previous studies, we used 13C-enriched wheat grains to label zearalenone and mycophenolic acid in order to analyze the hepatic metabolism of these two mycotoxins and also to use them as internal standards for their quantification in complex food matrixes.21,22 The feasibility of producing labeled mycotoxins by growing fungi on cereal grains was thus validated, but it was never used to elucidate the chemical formulas of new metabolites. In this study and for the first time, stable isotopic labeling and HPLC−HRMS were combined to analyze the whole secondary metabolome of Aspergillus fumigatus. This fungal species is one of the most important life-threatening humanpathogenic fungi. Virulence of this fungus is related to the production of multiple mycotoxins, such as the immunosuppressive gliotoxin and the necrosis-inducing trypacidin.23,24 For these reasons, A. fumigatus is currently one of the fungal species with the largest number of characterized secondary metabolites, which makes it a good validation model for this study. As a result, 21 metabolites produced by A. fumigatus were unambiguously identified, including fumigaclavine D that has never been described before.

Production of 13C- and 13C/15N-Labeled Wheat. Wheat uniform isotopic enrichments were performed on the variety Caphorn of Triticum aestivum as previously described.25 Two different enrichments were performed for this experiment: one nearly 100% 13C-enriched wheat (referred to as 13C wheat) and one nearly 50% 13C and nearly 100% 15N (referred to as 13 C/15N). Briefly, uniform enrichments of wheat with 15N and/ or 13C alone were conducted in hermetically sealed chambers (750 L) during a complete growth cycle, i.e., about 6 months, with labeled CO2 and labeled nitrate and ammonium salts as described in the Supporting Information. Final isotopic enrichments were measured with a Delta V Advantage isotope ratio mass spectrometer (Thermo Fisher Scientific). They were of 96.8% ± 0.15% for the 13C wheat, 53.4% ± 0.3% 13C and 96.8% ± 0.07% 15N for the 13C/15N wheat. Fungal Strain and Growth Conditions. Cultures of A. fumigatus (NRRL 35693, Northern Regional Research Laboratory, USDA-ARS, Peoria, IL, U.S.A.) grown on potato dextrose agar (PDA) for 7 days at 25 °C were used to inoculate the wheat grains that were previously soaked in water and autoclaved twice. Water activity (aw) of the grains was measured with an HC2-AW measurement device (Rotronik AG, Basserdorf, Switzerland), and sterile water was added, if necessary, to reach an aw of 0.98 before inoculation. Inoculation was performed with 250 μL of a spore suspension containing 2 × 105 conidia of A. fumigatus, obtained from a 7 day culture on PDA at 37 °C. A. fumigatus was allowed to grow for 15 days at 25 °C in the dark without shaking on polystyrene Petri dishes (Sarstedt, Marnay, France) containing 30 g of one of the three different substrates: (1) naturally enriched wheat grains (99% 12 C/1% 13C, referred to as 12C wheat) and grains of the abovementioned 13C wheat (2) and 13C/15N wheat (3). In order to discriminate between metabolites produced by A. fumigatus and wheat metabolites, 12C wheat grains were also incubated in the same conditions as the other samples but without fungi. High humidity levels were maintained throughout the experiment. Metabolite Extraction. After incubation, fungal mycelium had colonized the entire Petri dish and all the wheat grains were fully covered by mycelium. Metabolite extraction was performed in a fume hood using 200 mL of chloroform, as previously described,24 at room temperature in a 1 L glass bottle where the blend of the mycelia and the grains was inserted cut in small pieces. The bottle was placed on a shaking table IKA Labortechnik HS501 (IKA, Staufen, Germany) at 180 rpm for 36 h. The chloroform extract was then filtered and evaporated in two steps: first with a rotavapor R-210/215 (Buchi, Flawil, Switzerland) and then with a Zymark TurboVap-LV evaporator (McKinley Scientific, Sparta, NJ, U.S.A.). The resulting dry residue (metabolites) was dissolved in 400 μL of methanol and then filtered with 0.45 μm Whatman filters (Whatman, GE Healthcare, Kent, U.K.) before HPLC−HRMS analysis. Preparation of Metabolite Reference Compounds. Metabolite reference compounds have been collected over the years either from commercial sources, such as Sigma (SigmaAldrich, Saint-Quentin Fallavier, France), or as gifts from different research groups. All standards were dissolved in methanol for qualitative analysis by HPLC−HRMS as described below. Standard solutions were kept at −20 °C. Liquid Chromatography and Mass Spectrometry Analysis. Sample analysis was performed using HPLC coupled to an LTQ Orbitrap XL high-resolution mass spectrometer (Thermo Fisher Scientific, Les Ulis, France). Briefly, 10 μL of



MATERIALS AND METHODS Chemicals and Reagents. All solvents and chemicals were purchased from Fisher Scientific (Thermo Fisher Scientific, Illkirch, France). Solvents were HPLC grade except acetonitrile, which was MS grade. All other chemicals were analytical grade. 8413

dx.doi.org/10.1021/ac401872f | Anal. Chem. 2013, 85, 8412−8420

Analytical Chemistry

Article

methanol suspension of culture extracts was injected into a reversed-phase (150 mm × 2.0 mm) 5 μm Luna C18 column (Phenomenex, Torrance, CA, U.S.A.) operated at a flow rate of 0.2 mL/min. A gradient program was performed with 0.1% formic acid (phase A) and 100% acetonitrile (phase B) with the following elution gradient: 0 min 20% B, 30 min 50% B, from 35 to 45 min 90% B, from 50 to 60 20% B. HRMS acquisitions were achieved with electrospray ionization (ESI) in the positive and negative modes as follows: spray voltage +4.5 kV, capillary temperature 350 °C, sheath gas (N2) flow rate 40 au (arbitrary units), auxiliary gas (N2) flow rate 6 au in the positive mode, and spray voltage −3.7 kV, capillary temperature 350 °C, sheath gas (N2) flow rate 30 au, auxiliary gas (N2) flow rate 10 au in the negative mode. Full MS spectra were acquired at a resolution of 60 000 with a range of mass-to-charge ratio (m/z) set to 50−800. Data Analysis. Chromatograms and mass spectra interpretations were performed for the three differently labeled cultures with Xcalibur (Thermo Scientific). Then, ions from different substrates but the with same retention time (RT) and similar MS2 pattern, which thus corresponded to the same metabolite in the three substrates, were grouped to determine their chemical formula. The search criteria to determine chemical formulas were restricted to a mass error below 5 ppm, and C, H, O, N, S, P, Cl, K, Na, and Fe were the only accepted elements. Dereplication of adducts was considered as described by Nielsen et al.26 Following the criteria described above, a list of possible chemical formulas was generated for each of the three m/z corresponding to the same metabolite. Then, the three lists were cross-referenced using Mass Compare, an in-house-developed program, to identify the common formula in the three lists which thus corresponded to the exact chemical formula of the metabolite (downloadable at http://www.toulouse.inra.fr/axiomm). Finally, all the resulting chemical formulas were searched in the database AntiBase 2012.13 Only fungal products were considered for putative characterization. Putative annotation of metabolites was performed by interpretation of MS2 experiments, and when possible, identity was confirmed by comparison with the HPLC−MS2 analysis of a standard compound. MS3 experiments were performed when additional information was needed for structural determination.

Figure 1. Studying fungal secondary metabolites. Flowchart of the experimental procedure adopted in this study for fungal metabolomic profiling: a combination of isotope-labeled wheat grains with HPLC− HRMS. After biomass enrichment (1), the strain NRRL 35693 of A. fumigatus was grown for 15 days on the three different substrates (12C, 13 C, and 13C/15N) (2). After metabolite extraction, separation of the different metabolites was performed by HPLC and exact masses were obtained with high-resolution MS in both positive and negative ESI modes (3). Chemical formulas were determined by interpretation of high-resolution mass spectra and by the use of the Mass Compare software (4). The resulting formulas were matched against the database AntiBase 2012 to search for known metabolites, and when possible, comparison with published standards was performed (5).



RESULTS AND DISCUSSION In this article, an improved analytical method to specifically study fungal secondary metabolites is described. The workflow of this multi-isotopic metabolomics approach is schematized in Figure 1. First, 12C-, 13C-, and 13C/15N-labeled wheat grains were used to grow the fungal strain of interest (Figure 1, parts 1 and 2). An additional 12C wheat sample was incubated in the same conditions but without fungi, in order to discriminate wheat metabolites in the analysis. Since wheat grains were the only source of carbon and nitrogen and that fungi are heterotrophic organisms, all the produced secondary metabolites were either unlabeled (12C cultures), fully labeled (13C cultures), or doubly labeled (13C/15N cultures). After chloroform extraction, evaporation to dryness, and redissolving in methanol, extracts were analyzed by HPLC−HRMS in order to specifically detect and identify fungal metabolites (Figure 1, part 3). Wheat metabolites were excluded from the study by HPLC−HRMS analysis of the 12C wheat sample that was incubated without A. fumigatus. Chemical formulas were determined by interpretation of high-resolution mass spectra

and by the use of the software Mass Compare (Figure 1, part 4). Identification of metabolites was finally performed with the database AntiBase 2012 and comparison with HPLC−HRMS analysis of standards (Figure 1, part 5). In the present study, this analytical method was applied and validated with the wellknown metabolome of A. fumigatus.27 With this approach only compounds of biological origin were labeled and could thus be readily identified by two means: (1) Gaussian isotopic patterns in the 13C/15N wheat sample (Figure 2C), which were used to determine specific RT to search for corresponding masses in 12C and 13C wheat samples, and (2) mass shifts between m/z ratios of compounds of biological 8414

dx.doi.org/10.1021/ac401872f | Anal. Chem. 2013, 85, 8412−8420

Analytical Chemistry

Article

Figure 2. Determination of the exact formula of labeled metabolites. A. fumigatus was grown for 15 days on natural wheat grains, 99% 12C (A) as well as on grains labeled with stable isotopes: 97% of 13C (B); 53% of 13C and 96% of 15N (C). Left-hand panels represent mass spectra obtained after HPLC−HRMS analyses in positive ESI mode at 2.8 min. Right-hand panels represent MS2 spectra of the m/z ratio of interest in each substrate. The mass shift between the 12C sample (m/z 299.1759) and the 13C sample (m/z 317.2361) was used to determine the number of atoms of carbon (18), and the mass shift between the 12C sample and the 13C/15N sample (m/z 311.2032) was used to determine the number of atoms of nitrogen (2). Then, combination of the three formula lists from the differently enriched substrates with Mass Compare led to the unambiguous determination of the formula of fumigaclavine A (C18H22O2N2, 298.37948 g/mol).

origin displaying the same RT in 12C, 13C, and 13C/15N samples. These mass shifts corresponded to the number of labeled carbon and nitrogen atoms. Fungal origin was further validated by comparison with the 12C wheat sample without fungi (data not shown). Performance of the New Method for Untargeted Fungal Metabolic Profiling. Study of an Example: Formula

Determination of a Known Metabolite. As an example, Figure 2 depicts the determination of the exact elemental composition of fumigaclavine A. First, the molecule could be detected at RT 2.8 min in the 13C/15N sample, displaying a characteristic isotopic pattern with the major ion detected at m/z 311.2032 (Figure 2C). This compound was also detected at the same RT (2.8 min) but at different m/z ratios: 299.1759 in the 12C wheat 8415

dx.doi.org/10.1021/ac401872f | Anal. Chem. 2013, 85, 8412−8420

Analytical Chemistry

Article

Table 1. Detected Known Metabolites of A. fumigatus in 15 Day Old Cultures on Wheat Grains chemical family trypacidins

metabolite

formulaa

iona

RT (min)b

+

27.74

trypacidin

C18H16O7Na

[M + Na]

questin

C16H11O5

[M − H]−

27.82

monomethylsulochrin

C18H17O7

[M − H]−

28.07

anthraquinones and anthrones

emodin

C15H9O5

[M − H]−

36.20

pseurotins

pseurotin A

C22H25NO8Na

[M + Na]+

17.25

helvolic acid

helvolic acid

C33H43O8

[M − H]−

37.96

fumigaclavins

fumigaclavine A

C18H23N2O2

[M + H]+

2.8

fumigaclavine B

C16H21N2O

[M + H]+

2.5

festuclavin

C16H21N2

[M + H]+

4.17

fumitremorgin B

C27H32N3O4

[M − H2O + H]+

37.3

fumitremorgin C

C22H26N3O3

[M + H]+

23.47

verruculogen

C27H32N3O6

[M − H2O + H]+

37.7

C27H33N3O7Na

[M + Na]+

37.7

C27H32N3O7

[M − H]−

35.41

TR-2

C22H26N3O6

[M − H]−

6.86

cyclotryprostatin A

C22H24N3O5

[M − H]−

10.94

spirotryprostatin A

C22H26N3O4

[M + H]+

15.04

spirotryprostatin B

C21H22N3O3

[M + H]+

22.61

compound 1d

C22H24N3O6

[M − H]−

13.21

C22H25N3O6Na

[M + Na]+

13.21

fumitremorgins

fumiquinazolines

fumiquinazoline C

C24H22N5O4

[M + H]+

23.56

tryptoquivalines

tryptoquivaline F

C22H19N4O4

[M + H]+

20.12

fumigallins

fumagillin

C26H34O7Na

[M + Na]+

37.00

C m/zc

12

367.0785 0.7 ppm 283.0620 3 ppm 345.0989 3 ppm 269.0461 2.2 ppm 454.1469 0.7 ppm 567.2975 2.1 ppm 299.1759 1.8 ppm 257.1651 1.0 ppm 241.1699 0.1 ppm 462.2385 0.4 ppm 380.1968 0.1 ppm 494.2279 1.2 ppm 534.2202 1.6 ppm 510.2257 2.3 ppm 428.1840 3.0 ppm 410.1734 3.1 ppm 396.1900 4.4 ppm 364.1618 3.3 ppm 426.1684 3.2 ppm 450.1633 0.4 ppm 444.1664 0.4 ppm 403.1405 0.1 ppm 481.2193 0.8 ppm

13

C m/zc

385.1388 1 ppm 299.1159 3.6 ppm 363.1599 4.2 ppm 284.09650 2.2 ppm 476.2205 1.1 ppm 600.4079 1.4 ppm 317.2360 0.9 ppm 273.2191 2.1 ppm 257.2236 0.2 ppm 489.3285 1.6 ppm 402.2692 3.7 ppm 521.3183 1.5 ppm 561.3097 3.8 ppm 537.3164 2.4 ppm 450.2579 3 ppm 432.2471 3 ppm 418.2638 4.1 ppm 385.2344 4.2 ppm 448.2423 3.2 ppm 472.2377 0.9 ppm 468.2463 1.6 ppm 425.2135 0.8 ppm 507.3061 1.6 ppm

13

C/15N m/zc 376.1090 0.1 ppm 292.0922 3.1 ppm 355.1321 1.6 ppm 495.2653 1.2 ppm 467.1827 3.9 ppm 584.3544 1.8 ppm 311.2032 0.6 ppm 268.1889 0.6 ppm 252.1939 1.1 ppm 480.2793 1.7 ppm 392.2168 3.3 ppm 512.2681 3.6 ppm 551.2567 4.4 ppm 527.2636 2 ppm 443.2150 2.2 ppm 425.2043 2.0 ppm 409.2173 2.3 ppm 376.186 1.1 ppm 440.1960 2.2 ppm 464.1894 4.6 ppm 462.1945 2 ppm 419.1685 0.1 ppm 495.2653 1.2 ppm

a

Formula corresponds to the formula of the indicated ion form obtained with Mass Compare. bRetention time of the metabolite detected at which the three peaks associated to each molecule are detected. c12C m/z, 13C m/z, 13C/15N m/z indicate major m/z obtained for the indicated adduct form of each molecule in 12C wheat, 97% 13C wheat, and 53% 13C/96% 15N wheat samples, respectively. dWang et al., 2008.39

sample (Figure 2A) and 317.2361 in the 13C wheat sample (Figure 2B). Common elemental composition of the three m/z ratios was confirmed by comparison of MS2 spectra. The mass shift between the 12C wheat sample (99% 12C) (m/z 299, Figure 2A) and the 13C sample (96.8% 13C enrichment) (m/z 317, Figure 2B) was used to determine the number of atoms of carbon (18 for this compound). In addition, the mass shift between the 12C sample and the 13C/15N sample (m/z 311, Figure 2C) corresponded to half the number of carbon atoms (53.4% 13C enrichment) plus the number of nitrogen atoms

(96.8% 15N enrichment). Therefore, taking the nitrogen rule into account, the number of atoms of nitrogen could be deduced (two for this compound). Given the specified number of carbon and nitrogen atoms, and by comparison of the three lists of possible chemical formulas, it was possible to reach only one common chemical formula within matching errors below 5 ppm (1.757 ppm in 12C sample, 0.880 ppm in 13C sample, −0.635 ppm in 13C/15N sample, Table 1). This formula was C 18 H 23 O 2 N 2 , that of the protonated fumigaclavine A [C18H22O2N2 + H]+. 8416

dx.doi.org/10.1021/ac401872f | Anal. Chem. 2013, 85, 8412−8420

Analytical Chemistry

Article

MS2 spectra were also in agreement with the structure of this metabolite. Indeed, the combination of MS2 experiments performed in high resolution and the isotopic labeling allowed unambiguous attribution of the elemental composition of fragment ions. For example, the major fragment ion observed at m/z 239.1545 in the 12C sample corresponds to a loss of 60.0214 u, which was attributed to a loss of CH3CO2H (5.0 ppm) (Figure 2A). This attribution was confirmed by the detection of a loss of 62.0277 u in the 13C sample (13C2H4O2, 2.7 ppm) (Figure 2B) and the loss of 61.0247 (13C112C1H4O2, 3.9 ppm) in the 13C/15N sample (Figure 2C). In the same way, the other fragment ions could be attributed to consecutive losses of CH3CO2H + CH5N (m/z 208.1121 Figure 2A), CH3CO2H + C2H5N (m/z 196.1121 Figure 2A), CH3CO2H + C4H8 (m/z 183.0916 Figure 2A), and C6H13O2N (m/z 168.0806 Figure 2A). All these fragments are in agreement with the structure of fumigaclavine A. Development of an Analysis Software: Mass Compare. To facilitate the large interpretation of data, a program called Mass Compare was developed. This program allows for the determination of the chemical formula of a metabolite based on the lists of possible formulas for each of the three m/z corresponding to that metabolite (in 12C, 13C, and 13C/15N). These lists are generated by the software managing the HPLC− HRMS analyses, Xcalibur, which determines the putative element compositions from measured masses. Then, Mass Compare first determines the number of atoms of carbon (n) from the difference between 12C and 13C m/z ratios. Chemical formulas that do not contain n carbons are directly discarded. It then finds the common formulas in the 12C and 13C samples, and finally, it searches the resulting possible formulas in the 13 C/15N list, which results in a unique formula. An added feature of Mass Compare is the possibility to simulate isotopic clusters by specifying 13C and 15N enrichment percentages. This feature was used to determine whether fungi can discriminate between the different isotopes of carbon and nitrogen during biosynthesis of secondary metabolites. Experimental clusters were therefore compared to theoretical isotopic clusters, showing no differences between the two clusters (Supporting Information Figure S-1). This is in agreement with previous studies, using the same algorithm, concerning the production of mycophenolic acid by Penicillium brevicompactum22 Discrimination between Fungal Secondary Metabolites and Organic Adducts or Postbiosynthesis Degradation Products. Adduct formation during sample preparation or during ionization/desorption by electrospray is very common.26 In the present study, Na+ appeared to be the most frequent adduct. The presence of such adducts that do not contain any carbon or nitrogen atoms does not interfere with formula determination as described above. However, other adducts such as methanol (CH3OH), acetonitrile (CH3CN), formic acid (HCOOH), or ammonia (NH3) may also bind noncovalently to ions. These adducts result in the addition of unlabeled atoms of carbon or nitrogen, which changes the mass shifts between 12 C, 13C, and 13C/15N m/z ratios and thus hinders formula determination. Addition of unlabeled atoms of carbon or nitrogen may also occur by postbiosynthesis covalent transformations induced by solvents, such as hydrolysis and methylation. Supporting Information Figure S-2 shows an example of the presence of an unlabeled atom of carbon in the major ion of the 13C sample, which was revealed during formula determination. Accordingly, the formula was determined to be

C23H22O5N4 in the 12C wheat sample and 12C 13C22H22O5N4 in the 13C wheat sample. When diluted in acetonitrile instead of methanol no peak was detected at m/z 435.1666 in the 12C wheat sample (data not shown). This result suggested that methanol had induced the incorporation of the additional carbon atom. The presence of a methanol adduct was first excluded given the absence of a fragment ion that corresponded to the loss of a methanol adduct (Figure 3A). Interestingly, comparison and interpretation of MS2 experiments revealed similarities with the

Figure 3. Postbiosynthesis degradation of tryptoquivaline F. A. fumigatus was grown for 15 days on natural wheat grains (99% 12C), grains labeled 97% with 13C, and grains labeled 53% with 13C and 96% with 15N, before extraction of secondary metabolites. An MS2 experiment was performed to identify similarities and differences of fragmentation patterns between the methylated form of tryptoquivaline F (A) and tryptoquivaline F (B), which was detected by HPLC− HRMS analysis in positive ESI mode at 16.8 min at m/z 403.1398. On the basis of this comparison, a structure for the degraded tryptoquivaline F was postulated (C). This structure fits with a degradation of tryptoquivaline F (D) by methanol by (i) hydrolysis of one of the cyclic amide functions and (ii) methylation of the resulting carboxylic acid function. 8417

dx.doi.org/10.1021/ac401872f | Anal. Chem. 2013, 85, 8412−8420

Analytical Chemistry

Article

fragmentation pattern of tryptoquivaline F (m/z 403.1398, Figure 3B).24 Indeed, fragment ions observed at m/z 199, 171, and 147 were common to both MS2 spectra (Figure 3, parts A and B) and displayed the same possible raw formula. Additionally, MS3 experiments performed on these fragment ions indicated that each ion corresponded to the same substructure in both compounds (data not shown). On the basis of this, it could be hypothesized that the unknown metabolite corresponded to a covalent transformation of tryptoquivaline F by hydrolyzation or methylation. In addition, interpretation of MS3 experiments of the fragment ions that differed between the two metabolites (mainly 332, 257, and 239) enabled us to postulate a structure for the unknown metabolite (Figure 3C). Interestingly, this structure fits with a degradation of tryptoquivaline F (Figure 3D) by methanol by (i) hydrolysis of one of the cyclic amide functions and (ii) methylation of the resulting carboxylic acid function. This result highlights one of the main advantages of this method which was already mentioned before; that is, the discrimination of nonbiosynthesized molecules. Such discrimination is essential in metabolomic studies, especially when trying to identify new metabolites, but it is also usually a struggling point with most detection methods. This problem is completely overcome with the use of the double isotopic labeling and the measurement of exact masses by HPLC−HRMS. Application of the New Method for Untargeted Fungal Metabolomic Profiling. Results on the Known Metabolome of A. fumigatus. Table 1 summarizes all the known metabolites that could be identified in the metabolome of A. fumigatus, including their retention times, chemical formulas, adduct forms, and their exact masses in 12C wheat substrate as well as in 13C wheat and 13C/15N wheat. A total of 20 already known metabolites could be unambiguously detected, thus validating the detection performance of the methodology presented here. As indicated in Table 1, these metabolites belong to 9 out of the 21 different chemical families of metabolites produced by A. fumigatus.27 Gliotoxin, which is one of the most toxic mycotoxins produced by A. fumigatus, was not detected in our analyses. A previous study reported that gliotoxin tends to disappear from the substrate after 6 days of incubation,28 which can explain why this mycotoxin was not detected in the present study. Other families of metabolites such as fumigatins, sphingofungins, and siderochromes were not detected either, and this can be due to specific requirements for the production or detection of these metabolites. Also, the strain of A. fumigatus used in this study does not produce pyripyropenes. Identification of a New Metabolite Produced by A. fumigatus. In addition to these known metabolites, it was also possible to identify a new metabolite belonging to the family of the fumigaclavines, which we will further refer to as fumigaclavine D (Figure 4). This metabolite was detected at 4.2 min at m/z 315.1701 in the 12C wheat sample. Following the method presented in this study, the exact formula of this ion was determined to be [C18H22N2O3 + H]+. The MS2 spectrum of fumigaclavine D (Figure 4A) presented similarities with those of the fumigaclavines and particularly of fumigaclavine A (Figure 2A), which contains one less atom of oxygen than fumigaclavine D. MS3 experiments confirmed identical substructures of the fragment ions 208 and 196 from fumigaclavine A and fumigaclavine D. Thus, the position of the additional oxygen in fumigaclavine D was determined as necessarily depicted in Figure 4B. MS3 experiments of the

Figure 4. Structure elucidation of a new fumigaclavine produced by A. fumigatus. A. fumigatus was grown for 15 days on natural wheat grains (99% 12C), grains labeled 97% with 13C, and grains labeled 53% with 13 C and 96% with 15N, before extraction of secondary metabolites. The new fumigaclavine was detected at 4 min at m/z 315.1701. An MS2 experiment was performed (A) to identify similarities and differences of fragmentation pattern between this metabolite and fumigaclavine A, depicted in Figure 2. On the basis of this comparison, a structure for the new fumigaclavine was postulated (B), which contains one more hydroxyl group than fumigaclavine A (C).

fragment ion 239 resulted in different profiles in the two compounds, whereas a similar raw formula was determined as C16H19N2. This fragment ion was therefore attributed to the consecutive losses of CH3CO2• and •OH for fumigaclavine D and to the loss of CH3CO2H for fumigaclavine A (Figure 4, parts B and C, respectively). Given these results it was possible to elucidate the structure of this new member of the fumigaclavines family. No hit for this metabolite was found in AntiBase 2012 or in the literature; therefore, to the best of our knowledge it is the first time that fumigaclavine D has been identified. The importance of this finding comes from the known toxicity of fumigaclavines, which are part of the ergot alkaloids or clavine alkaloids family. Ergot alkaloids are able to interact with several monoamine receptors such as dopamine, serotonin, and α-adrenaline. The result of this interaction can have different beneficial or detrimental impacts on the cardiovascular, nervous, reproductive, and immune systems.29−31 Also, some of the clavine alkaloids have been shown to induce mutagenesis in mammalian cells.32 So far, three members of the fumigaclavine family have been described: fumigaclavine A, B, and C. Fumigaclavine C is the end-product in the biosynthetic pathway of these metabolites, whereas fumigaclavine A and B are the ultimate and penultimate intermediate metabolites, respectively.33 The structure of fumigaclavine A is the closest to that of the new fumigaclavine; therefore, they probably have a common 8418

dx.doi.org/10.1021/ac401872f | Anal. Chem. 2013, 85, 8412−8420

Analytical Chemistry

Article

(Auburn, Alabama, U.S.A.) for her help correcting the manuscript.

precursor. The ergot alkaloid biosynthesis pathway, by which these two metabolites are produced, has not been fully elucidated yet. However, 11 ergot alkaloid genes (eas) have been identified.34−36 Previous studies have shown that early steps of the biosynthetic pathway are well-conserved, whereas the latter steps are species-specific.37 Among these species specific genes, easK and easM encode for two cytochromes P450 (CYP) that are involved in the biotransformation of festuclavine into fumigaclavine B, then fumigaclavine A, and finally fumigaclavine C in A. fumigatus. It could thus be hypothesized that one of the two CYP performed a second nonspecific hydroxylation, which would lead to the production of the side-product fumigaclavine D. This kind of nonspecific CYP activity has already been observed in other biosynthetic pathways such as patulin biosynthesis.38



(1) Archer, D. B. Curr. Opin. Biotechnol. 2000, 11, 478−483. (2) Meyer, V. Biotechnol. Adv. 2008, 26, 177−185. (3) Council for Agricultural Science and Technology. Potential Economic Costs of Mycotoxins in United States; Task Force Report 138; CAST: Ames, IA, 2003; pp 136−142. (4) Fisher, M. C.; Henk, D. A.; Briggs, C. J.; Brownstein, J. S.; Madoff, L. C.; McCraw, S. L.; Gurr, S. J. Nature 2012, 484, 186−194. (5) Berger, L.; Speare, R.; Daszak, P.; Green, D. E.; Cunningham, A. A.; Goggin, C. L.; Slocombe, R.; Ragan, M. A.; Hyatt, A. D.; McDonald, K. R.; Hines, H. B.; Lips, K. R.; Marantelli, G.; Parkes, H. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 9031−9036. (6) Frick, W. F.; Pollock, J. F.; Hicks, A. C.; Langwig, K. E.; Reynolds, D. S.; Turner, G. G.; Butchkoski, C. M.; Kunz, T. H. Science 2010, 329, 679−682. (7) Bennett, J. W.; Klich, M. Clin. Microbiol. Rev. 2003, 16, 497−516. (8) Streit, E.; Schatzmayr, G.; Tassis, P.; Tzika, E.; Marin, D.; Taranu, I.; Tabuc, C.; Nicolau, A.; Aprodu, I.; Puel, O.; Oswald, I. P. Toxins 2012, 4, 788−809. (9) Grenier, B.; Applegate, T. J. Toxins 2013, 5, 396−430. (10) Wild, C. P.; Gong, Y. Y. Carcinogenesis 2010, 31, 71−82. (11) Kensler, T. W.; Roebuck, B. D.; Wogan, G. N.; Groopman, J. D. Toxicol. Sci. 2011, 120 (Suppl. 1), S28−S48. (12) Oswald, I. P.; Marin, D. E.; Bouhet, S.; Pinton, P.; Taranu, I.; Accensi, F. Food Addit. Contam. 2005, 22, 354−360. (13) Laatsch, H. AntiBase 2012, The Natural Compound Identifier; Wiley-VCH: Weinheim, Germany, 2012; http://www.wiley-vch.de/ stmdata/antibase.php. (14) Kirk, P. M.; Cannon, P. F.; Minter, D. W.; Stalpers, J. A. Ainsworth & Bisby’s Dictionary of the Fungi, 10th ed.; CAB International: Wallingford, U.K., 2008. (15) Fox, E. M.; Howlett, B. J. Curr. Opin. Microbiol. 2008, 11, 481− 487. (16) Schuhmacher, R.; Krska, R.; Weckwerth, W.; Goodacre, R. Anal. Bioanal. Chem. 2013, 405, 5003−5004. (17) Dettmer, K.; Aronov, P. A.; Hammock, B. D. Mass Spectrom. Rev. 2007, 26, 51−78. (18) Klein, S.; Heinzle, E. WIREs Syst. Biol. Med. 2012, 4, 261−272. (19) Bueschl, C.; Krska, R.; Kluger, B.; Schuhmacher, R. Anal. Bioanal. Chem. 2013, 405, 27−33. (20) Kluger, B.; Bueschl, C.; Lemmens, M.; Berthiller, F.; Häubl, G.; Jaunecker, G.; Adam, G.; Krska, R.; Schuhmacher, R. Anal. Bioanal. Chem. 2012, 1−6. (21) Yen, S.; Pean, M.; Puel, O.; Loiseau, N.; André, F.; Delaforge, M. In Mycotoxins and Phycotoxins. Advances in Determination, Toxicology and Exposure Management; Niapau, H., Trujillo, S., van Egmond, H. P., Park, D. L., Eds.; Wageningen Academic Publishers: Wageningen, The Netherlands, 2006; pp 185−193. (22) Bravin, F.; Duca, R. C.; Loiseau, N.; Pean, M.; Puel, O.; Delaforge, M. World Mycotoxin J. 2008, 1, 275−281. (23) Kwon-Chung, K. J.; Sugui, J. A. Med. Mycol. 2009, 47, S97− S103. (24) Gauthier, T.; Wang, X.; Sifuentes Dos Santos, J.; Fysikopoulos, A.; Tadrist, S.; Canlet, C.; Artigot, M. P.; Loiseau, N.; Oswald, I. P.; Puel, O. PLoS ONE 2012, 7, e29906. (25) Péan, M.; Boiry, S.; Ferrandi, J.-C.; Gibiat, F.; Puel, O.; Delaforge, M. J. Labelled Compd. Radiopharm. 2007, 50, 569−570. (26) Nielsen, K. F.; Månsson, M.; Rank, C.; Frisvad, J. C.; Larsen, T. O. J. Nat. Prod. 2011, 74, 2338−2348. (27) Frisvad, J. C.; Rank, C.; Nielsen, K. F.; Larsen, T. O. Med. Mycol. 2009, 47, S53−S71. (28) Boudra, H.; Morgavi, D. P. Anim. Feed Sci. Technol. 2005, 120, 113−123. (29) Panaccione, D. G.; Coyle, C. M. Appl. Environ. Microbiol. 2005, 71, 3106−3111.



CONCLUSIONS As a conclusion, we successfully validated a new untargeted method for fungal metabolomic profiling on A. fumigatus. The combination of high-accuracy mass spectrometry and double isotopic labeling efficiently enabled the specific detection and the unambiguous determination of chemical formulas of known metabolites produced by this species. One of the main advantages of the method presented here is the discrimination of nonbiological products, which is usually a source of interference in these kinds of studies. Related to that, it was possible to identify a product that resulted from the degradation of tryptoquivaline F by methanol. Additionally, a new member of the fumigaclavine family was characterized for the first time, bringing new perspectives on the biosynthetic pathway of this important family of metabolites. To the best of our knowledge it is the first time that such a strategy has been used to characterize a fungal metabolome. As a conclusion, method robustness has been validated and it can now be applied to unknown compounds from other fungal metabolomes. It offers great possibilities for the discovery of new drugs and toxins in different research domains such as pharmacology, toxicology, and phytopathology.



ASSOCIATED CONTENT

S Supporting Information *

Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.



REFERENCES

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Fax: 33 (0) 5 61 28 51 45. Notes

The authors declare no competing financial interests.



ACKNOWLEDGMENTS Patricia Cano was supported by doctoral fellowships cofunded by the Agence Nationale de Sécurité Sanitaire de l’Alimentation, de l’Environnement et du Travail (ANSES) and the Institut National de la Recherche Agronomique (INRA). This work was supported by Grants (ES-2007-063 and EST-2011/ 1/087) from ANSES. We thank P. M. Scott, S. Bony, H. Osada, H. Fujimoto, J. Dorner, and C. Avendano for A. fumigatus chemical standards. HPLC−HRMS analyses were performed within the MetaToul-AXIOM analytical platform. We also thank Marisa Younanian from the University of Auburn 8419

dx.doi.org/10.1021/ac401872f | Anal. Chem. 2013, 85, 8412−8420

Analytical Chemistry

Article

(30) Schardl, C. L.; Panaccione, D. G.; Tudzynski, P. In The Alkaloids. Chemistry and Biology; Cordell, G. A., Ed.; Academic Press: San Diego, CA, 2006; Vol. 63, pp 45−86. (31) Pertz, H. Planta Med. 1996, 62, 387−392. (32) Glatt, H.; Pertz, H.; Kasper, R.; Eich, E. Anticancer Drugs 1992, 3, 609−614. (33) O’Hanlon, K. A.; Gallagher, L.; Schrettl, M.; Jochl, C.; Kavanagh, K.; Larsen, T. O.; Doyle, S. Appl. Environ. Microbiol. 2012, 78, 3166−3176. (34) Lorenz, N.; Haarmann, T.; Pazoutova, S.; Jung, M.; Tudzynski, P. Phytochemistry 2009, 70, 1822−1832. (35) Wallwey, C.; Li, S. M. Nat. Prod. Rep. 2011, 28, 496−510. (36) Ryan, K. L.; Moore, C. T.; Panaccione, D. G. Toxins 2013, 5, 445−455. (37) Coyle, C. M.; Panaccione, D. G. Appl. Environ. Microbiol. 2005, 71, 3112−3118. (38) Artigot, M. P.; Loiseau, N.; Laffitte, J.; Mas-Reguieg, L.; Tadrist, S.; Oswald, I. P.; Puel, O. Microbiology 2009, 155, 1738−1747. (39) Wang, F.; Fang, Y.; Zhu, T.; Zhang, M.; Lin, A.; Gu, Q.; Zhu, W. Tetrahedron 2008, 64, 7986−7991.

8420

dx.doi.org/10.1021/ac401872f | Anal. Chem. 2013, 85, 8412−8420