Bacteria associated with truffle-fruiting bodies contribute to truffle aroma
Summary
Truffles, symbiotic fungi renown for the captivating aroma of their fruiting bodies, are colonized by a complex bacterial community of unknown function. We characterized the bacterial community of the white truffle Tuber borchii and tested the involvement of its microbiome in the production of sulphur-containing volatiles. We found that sulphur-containing volatiles such as thiophene derivatives, characteristic of T. borchii fruiting bodies, resulted from the biotransformation of non-volatile precursor(s) into volatile compounds by bacteria. The bacterial community of T. borchii was dominated by α- and β-Proteobacteria. Interestingly, all bacteria phyla/classes tested in this study were able to produce thiophene volatiles from T. borchii fruiting body extract, irrespective of their isolation source (truffle or other sources). This indicates that the ability to produce thiophene volatiles might be widespread among bacteria and possibly linked to primary metabolism. Treatment of fruiting bodies with antibacterial agents fully suppressed the production of thiophene volatiles while fungicides had no inhibitory effect. This suggests that during the sexual stage of truffles, thiophene volatiles are exclusively synthesized by bacteria and not by the truffle. At this stage, the origin of thiophenes precursor in T. borchii remains elusive and the involvement of yeasts or other bacteria cannot be excluded.
Introduction
Truffles are symbiotic fungi that develop underground in association with plant roots, forming ectomycorrhizas (Mello et al., 2006). Ectomycorrhizal associations, dominant in boreal and temperate forests, are of high ecological relevance since they improve plant nutrition and health (Read, 1991; Buscot et al., 2000; Martin et al., 2001). About 180 truffle species (Tuber spp) associate with angiosperm and gymnosperms and naturally occur in Europe, North-America and Asia (Bonito et al., 2010).
The genome sequencing of the Périgord black truffle Tuber melanosporum expanded the status of truffles from a food delicacy to a scientific model valuable for the study of complex symbiotic interactions (Martin et al., 2010). Truffles ectomycorrhizas and fruiting bodies harbour a diverse microbial community including bacteria, yeasts and filamentous fungi (Barbieri et al., 2005; 2007; Buzzini et al., 2005; Pacioni et al., 2007). Of these microbes, only bacteria have been extensively studied in truffles. Complex bacterial communities have been reported to establish in different truffle species: T. aestivum (Gryndler et al., 2013), T. magnatum (Barbieri et al., 2007), T. melanosporum (Antony-Babu et al., 2013) and T. borchii (Sbrana et al., 2000; Barbieri et al., 2005). Bacteria colonize both the external (peridium) and internal part (gleba) of truffles and seem to be selected from the soil communities during the early stage of truffle formation (Antony-Babu et al., 2013). Although bacterial communities differ depending on truffle species analysed, a core microbiome composed of α-Proteobacteria from the family of Bradyrhizobiaceae seems common to all species studied so far (Barbieri et al., 2005; 2007; Antony-Babu et al., 2013). Factors responsible for the selection of these bacteria remain mysterious. However, these bacteria could have a role in the development, growth and nutrition of truffle-fruiting bodies (Sbrana et al., 2000; 2002; Barbieri et al., 2007; 2010; Antony-Babu et al., 2013; Pavić et al., 2013).
Truffle-fruiting bodies emit intense aromas (Splivallo and Maier, 2011; Splivallo et al., 2011). Sulphur-containing volatiles have a central role in truffle aroma because they serve as attractants to mammals and contribute to truffle aroma sensed by humans (Talou et al., 1990; Culleré et al., 2010; Splivallo and Maier, 2011; Splivallo et al., 2011). The origin of sulphur-containing volatiles in truffles is unclear as they might be derived from the truffle fungus itself but also from the microbial community inhabiting truffle-fruiting bodies (Buzzini et al., 2005; Splivallo and Maier, 2011; Splivallo et al., 2011). Analysis of the Black truffle's genome T. melanosporum, however, suggested that Black truffle may produce its volatiles without the involvement of bacteria (Martin et al., 2010; Maxmen, 2010). Whether other truffle species can produce their volatiles by themselves or require the participation of associated bacteria remain unknown.
Here, we investigated the role of truffle-associated bacteria in the formation of the aroma of a white truffle species, Tuber borchii. T. borchii is a cultivated truffle species naturally occurring in Europe that has been recently introduced in New Zealand (Bonito et al., 2010). T. borchii has been for long used in laboratories as a model organism to study truffles because of the relative faster growth of its mycelium compared with other truffle species. T. borchii-fruiting bodies emit some volatiles [thiophene derivatives such as 3-methylthiophene, hereafter referred to as (1), and 3-methyl-4,5-dihydrothiophene, referred to as (2)], which are species-specific and might be partially responsible for their characteristic aroma (Bellesia et al., 2001; Mauriello et al., 2004; Zeppa et al., 2004; Splivallo et al., 2007). The concentrations of thiophene volatiles in T. borchii have been reported to occur solely in fully mature (71–100% maturity) truffles (Zeppa et al., 2004) and to increase upon storage at room temperature (Bellesia et al., 2001), questioning the role of truffle-associated bacteria in the production of the aroma. To test this hypothesis, we analysed the evolution of the composition of the bacterial communities along storage as well as the production of thiophene derivatives. Then we tested the ability of bacterial isolates from T. borchii to produce thiophene volatiles. Overall, our results demonstrated that thiophene volatiles characteristic of T. borchii-fruiting bodies were produced by the microbiome inhabiting truffle-fruiting bodies.
Results
Composition of T. borchii bacterial communities evolves during storage
As a first step, we characterized the composition of bacterial communities associated to T. borchii during storage. Six T. borchii truffles were subsampled after 0, 2, 4 and 6 days of storage at room temperature. The bacterial communities were characterized and quantified by fluorescent in situ hybridization (FISH) using the eubacterial universal probe EUB338 as well as probes specifically targeting Firmicutes (LGC354A), Actinobacteria (HGC69a), Bacteroidetes (CF319), α- (ALF1b), β- (BET42a), γ-Proteobacteria (GAM42a) and Enterobacteriales (EntB and EntD). Because bacterial distribution within fruiting bodies can be highly non-homogenous (Antony-Babu et al., 2013), quantification of the bacterial communities was done by homogenizing gleba samples through grinding. T. borchii bacterial community composition was comparable with the one previously described in T. borchii (Barbieri et al., 2007), with an overall majority of Proteobacteria (Fig. 1) among which α- and β-Proteobacteria were dominant. γ-Proteobacteria, including Enterobacteriales, and bacteria belonging to Bacteroidetes were also detected but at a much lower level. No Firmicutes or Actinobacteria could be detected at day 0 (Fig. 1). The overall bacterial community evolved upon storage with a shift of community composition with time. The quantity of bacteria first tended to increase at 2 days and was then reduced at day 6. A similar reduction in community size over time was observed in the α- and the β-Proteobacteria. The Bacteroidetes, the Enterobacteriales, the Actinobacteria and the γ-Proteobacteria populations remained quite low and stable during storage. Last, the Firmicutes population size slightly increased during the last days of storage.
Change in the bacterial population inside T. borchii fruiting bodies as a function of storage time. The panels show bacterial population dynamics (±SE) of all bacteria [Eubacteria expressed as bacterial cell count per milligram of truffle-fruiting body (dry weight)] and specific classes/groups (expressed as percentage of the total Eubacteria). The community was dominated by α-and β-Proteobacteria. No statistical difference was detected among storage days within Eubacteria or bacterial classes/groups (n = 6 truffle-fruiting bodies, for all ANOVA: 0.03 < F < 1.62; 0.22 < P < 0.99). Enterobacteriales data from FISH probe EntD (probe EntB gave comparable results – not shown).
Additionally to address bacterial cell distribution within undisrupted truffle tissues, thin sections of T. borchii were hybridized with FISH probes against α- and β-Proteobacteria, the two main bacterial groups identified in ground samples. α- and β-Proteobacteria could be observed in both the peridium and the gleba by confocal microscopy imaging (Fig. 2). Dense colonies of α- and β-Proteobacteria were observed in the peridium. However, this colonization was patchy with vast area without visible bacterial cells (Fig. 2A and B). Bacterial cells were present in between fungal cells but not inside the fungal cells in both gleba and peridium. Colonies of α-Proteobacteria did not appear to contain bacteria from other phyla in the peridium as illustrated by the complete overlay of Eubacteria and alpha probes (Fig. 2A). In contrast, β-Proteobacteria were found in mixed population with other bacteria (Fig. 2B) as demonstrated by the absence of overlay between eubacterial probe (FITC, green) and β-Proteobacteria (cy3, red). In the gleba, α-Proteobacteria were found as isolated cells as well as dense colonies (Fig. 2C) while β-Proteobacteria were more evenly distributed (Fig. 2D).
Localization of α- and β-Proteobacteria in thin sections of T. borchii peridium and gleba.
A,B. Hybridization of 30 μm sections of T. borchii of peridium.
C,D. Hybridization of gleba (C,D) with the universal eubacteria Eub338 mix probe coupled to FITC (A, B, C) or cy3 (D) and with α-Proteobacteria specific probe coupled to cy3 (A, C) or β-Proteobacteria specific probe BET42A coupled to cy3 (B) or FITC (D) as observed by confocal microscopy.
White bars represent 10 μm. asc., ascii; b, bacterial cluster; myc., fungal cell; sp., spore. Each picture is representative of observations.
Bacterial community composition differ between peridium and gleba
Antony-Babu and colleagues (2013) recently demonstrated that bacterial communities from Tuber melanosporum strongly differed between the peridium and the gleba. To determine if a similar pattern was also present in T. borchii, we compared by FISH bacterial community composition in gleba and peridium samples at day 0. Only the population size of Bacteroidetes significantly differed between gleba and peridium. The population size of the latter phylum was five times higher in the peridium compared with the gleba (gleba: 137 ± 122; peridium: 697 ± 101; unit: cell count/mg dry weight; P-value: 0.026, Mann–Whitney U-test). No statistical difference between peridium and gleba was observed for the Firmicutes (gleba: 0 ± 0; peridium: 10 ± 10), Actinobacteria (gleba: 0 ± 0; peridium: 45 ± 45), α-Proteobacteria (gleba: 1370 ± 499; peridium: 947 ± 365), β- Proteobacteria (gleba: 1356 ± 453; peridium: 1336 ± 483), γ-Proteobacteria (gleba: 115 ± 82; peridium: 137 ± 43) and Enterobacteriales (gleba: 240 ± 210; peridium: 6 ± 3).
Concentration of thiophene volatiles in T. borchii gleba correlates with the bacterial abundance
Concentrations of thiophene volatiles have been reported to increase with the storage of truffle-fruiting bodies at room temperature (Bellesia et al., 2001), suggesting a potential role of bacteria in the production of these volatile compounds. We hypothesized that the concentrations of thiophene volatiles correlated with bacterial community composition and abundance within truffle-fruiting bodies. To test this hypothesis, thiophenes produced from the same gleba samples characterized by FISH were quantified by solid-phase microextraction–gas chromatography/mass spectrometry (SPME-GC/MS). Thiophene volatiles (1) and (2) were emitted all along the time course. The total density of Eubacteria were significantly correlated with the levels of thiophene volatile (2) only [volatile (1): Pearson R2 = 0.008, P = 0.686; volatile (2): R2 = 0.220, P = 0.021). A significant correlation for volatiles (1) and/or (2) was also observed for the dominant α-Proteobacteria and a minor group representing Bacteroidetes (for α-Proteobacteria, volatile (1): R2 = 0.171, P = 0.045; volatile (2): R2 = 0.172, P = 0.044; for Bacteroidetes, volatile (1): R2 = 0.002, P = 0.821; volatile (2): R2 = 0.303, P = 0.005).
Bacteria isolated from fruiting bodies, but not T. borchii mycelia, are able to generate thiophene volatiles from T. borchii-fruiting bodies
The correlation between the evolution of bacterial communities during storage and the concentration of thiophene volatiles led us to hypothesize that some bacteria could be involved in the production of these compounds. To test this hypothesis, we tested the ability of bacteria isolated from T. borchii-fruiting bodies to produce thiophene compounds. Bacteria were isolated from soil adhering to the peridium and from the gleba of T. borchii-fruiting bodies on tryptic soy agar (TSA 3%) plates. Based on colony PCR and 16S rRNA sequences, the isolated strains belonged to β- and γ-Proteobacteria, Bacteroidetes, Firmicutes and Actinobacteria (Table 1). We were not able to isolate strains from the α-Proteobacteria phylum despite their high abundance in the fruiting body. Since those bacteria could also be involved in T. borchii aroma, we tested two Rhizobiales strains obtained from the German Collection of Microorganisms and Cell Cultures (Table 1).
| Isolate reference in the present manuscript | Isolated from | Gleba or peridium/adhering soil | GenBank Accession Number 16S rRNA | Bacterial phylum/class or order | Genus |
|---|---|---|---|---|---|
| A | T. borchii | Gleba | KC618433 | Actinobacteria | Microbacterium |
| B | T. borchii | Gleba | KC618434 | Bacteroidetes/Flavobacteria | Chryseobacterium |
| C | T. borchii | Peridium/adhering soil | KC618435 | Bacteroidetes/Sphingobacteria | Sphingobacterium |
| D | T. borchii | Gleba | KC618436 | Bacteroidetes/ Sphingobacteria | Sphingobacterium |
| E | T. borchii | Peridium/adhering soil | KC618437 | Firmicutes/Bacillales | Brochothrix |
| F | T. borchii | Peridium/adhering soil | KC618438 | Firmicutes/Bacillales | Brochothrix |
| G | T. borchii | Peridium/adhering soil | KC618439 | β-proteobacteria/Burkholderiales | Comamonas |
| H | T. borchii | Gleba | KC618440 | β-proteobacteria/Burkholderiales | Comamonas |
| I | T. borchii | Peridium/adhering soil | KC618441 | γ-proteobacteria/Enterobacteriales | Serratia |
| J (Serratia plymuthica IC14b) | Soil (melon plantation) | – | – | γ-proteobacteria/Enterobacteriales | Serratia |
| K (Serratia marcescens MG1b) | Cucumber (fruit) | – | – | γ-proteobacteria/Enterobacteriales | Serratia |
| L (Serratia proteamaculans B5ab) | Salmon | – | – | γ-proteobacteria/Enterobacteriales | Serratia |
| M | T. borchii | Peridium/adhering soil | KC618442 | γ-proteobacteria/Pseudomonadales | Pseudomonas |
| N | T. borchii | Gleba | KC618443 | γ-proteobacteria/Pseudomonadales | Pseudomonas |
| O | T. borchii | Peridium/adhering soil | KC618444 | γ-proteobacteria/Xanthomonadales | Stenotrophomonas |
| P | T. borchii | Gleba | KC618445 | γ-proteobacteria/Xanthomonadales | Stenotrophomonas |
| Q (Ensifer adhaerens strain LFG19a; DSMZ ref. 18131a) | Not determined | – | – | α-proteobacteria/Rhizobiales | Ensifer |
| R (Bosea thiooxidans strain BI-42; DSMZ ref. 9653a) | Agricultural Soil | – | AJ250796 | α-proteobacteria/Rhizobiales | Bosea |
- a. Available from the German Collection of Microorganisms and Cell Cultures (www.dsmz.de). b. Isolates of Serratia tested in an earlier report for their plant-growth-promoting properties (Blom et al., 2011) (kindly provided by Dr. Laure Weisskopf, University of Zürich, Switzerland).
A bioassay approach was used to test the ability of single bacterial strains to transform ethyl acetate extracts of truffle-fruiting bodies into thiophene volatiles (Supporting Information Fig. S2). Surprisingly, thiophene volatiles (1) and (2) were produced by all bacterial isolates but only in the presence of fruiting body extract (Fig. 3A and B). The amount of thiophene compounds produced varied between bacterial strains, Proteobacteria and more especially α-, β-Proteobacteria and Enterobacteriales among the γ-Proteobacteria being the most efficient producers (Fig. 3B). Surprisingly thiophene volatiles (1) and (2) were not only produced by bacteria isolated from T. borchii-fruiting bodies, but also by bacterial isolated from other sources (i.e. soil, plants, Table 1). The quantities of thiophene volatiles (1) and (2) released in our bioassays were similar to ∼10 times lower to those detected from T. borchii-fruiting bodies (Fig. 3C). Killed bacteria by autoclaving did not produce thiophene volatiles indicating that active metabolism was required (Fig. 3A).
Ability of bacteria and truffle mycelium to produce thiophene volatiles.
A. Chromatograms [total ion current (TIC) and extracted m/z 97] illustrating that bacteria but not truffle mycelium have the ability to transform truffle extract into two volatiles characteristic of T. borchii-fruiting bodies; volatile (1) is 3-methylthiophene and volatile (2) is 3-methyl-4,5(2H)thiophene. Chromatograms are shown with normalized intensities.
B. Levels (±SE) of thiophene volatiles (1) and (2) (log scale normalized to growth) and corresponding growth (±SE) (OD600) of different bacterial classes/groups (n = 3) (isolates are coded from A to R – See Table 1). Different letters denote values that differ significantly from each other (Fischer LSD post-hoc test P < 0.05) – ANOVA for 3-methylthiophene: F = 15.94, P = 0.000000; ANOVA for 3-methyl-4,5(2H)thiophene: F = 6.26, P < 0.001].
C. Quantities of thiophenes (1) and (2) released per SPME vial by T. borchii gleba (n = 6) and in the bioassays with bacteria (n = 3 per strain) described in Supporting Infomation Fig. S1. To make the values comparable between bioassays and gleba, thiophene quantities (extracted masses m/z 97 and m/z100) have been normalized to the TIC of the chromatograms.
To test whether the fungus was also able to produce thiophene volatiles by itself, volatile profiles of in vitro pure culture of T. borchii strain 43Bo isolated in 1996 were analysed. No thiophene volatiles were produced by T. borchii mycelial cultures in vitro (Fig. 3A). To make sure that the inability of T. borchii mycelial culture to produce thiophene volatiles was not a consequence of prolonged storage in pure culture of strain 43Bo, mycelium was freshly isolated from a fruiting body producing thiophene volatiles. The newly isolated strain IG2 was not able to produce any thiophene derivatives, confirming the previous results. Altogether these data indicate that bacteria but not truffle mycelium are able to biotransform the non-volatile precursors of T. borchii gleba into volatile thiophenes.
The precursor of thiophene volatiles is specific of the sexual stage of T. borchii and of no other truffle species
In a second step, we tested whether bacteria could produce thiophene volatiles when grown on T. borchii mycelium. Most surprisingly, even the most efficient thiophene-producer under our assay's conditions, an Enterobacteriales of the Serratia genus (Strain I, Table 1), failed to produce the thiophene volatiles typical of T. borchii-fruiting bodies when grown on sterilized T. borchii mycelium (strain 43Bo) that had been pre-grown in pure culture. Live mycelial cultures of T. borchii strain 43Bo also failed to produce thiophene volatiles on malt extract with or without addition of L-methionine, which had been suggested to be the precursor of thiophene volatiles in T. borchii (Zeppa et al., 2004). Methionine, however, did induce in live mycelial cultures (strain 43Bo) the synthesis of numerous sulphur (non-thiophene) volatiles such as dimethyl sulphide and dimethyl disulphide (data not shown). These results suggest that mycelium of T. borchii does not produce the precursors of thiophenes in vitro. Furthermore, the Serratia isolate I (Table 1) produced the thiophene volatiles (1) and (2) on T. borchii gleba but not when grown on the gleba of other truffle species, including the white truffle T. magnatum or the black truffles T. aestivum, T. mesentericum and T. brumale.
Overall, these results indicate that thiophene production is linked to the sexual stage (fruiting body) of T. borchii. They also demonstrate that the aroma precursor of thiophene volatiles is/are unique to T. borchii.
Antibacterial but not antifungal agents inhibit the production of thiophene volatiles in T. borchii-fruiting bodies
To further investigate which organisms are responsible for the production of specific truffle-fruiting body volatiles, we used antimicrobial agents targeting either fungi or bacteria. A homogenate of T. borchii-fruiting bodies was treated with aqueous solutions containing either the broad-spectrum antibacterial agent streptomycin, or one of the antifungal agents clotrimazole or amphotericin B, or pure water as a control. The concentrations of volatiles (1) and (2) were measured at the beginning of the experiment and after 48 h (Fig. 4). Production of thiophene volatiles (1) and (2) was only suppressed by the bactericide streptomycin (Fig. 4). The concentration of thiophenes (1) and (2) in samples treated with the fungicides (48 h, Fig. 4) increased to the same level as in the water control (Fig. 4), indicating that fungal metabolism was not necessary for the production of thiophene volatiles. Furthermore, the concentration of thiophene volatiles (1) and (2) induced by the different treatments correlated with the bacterial population density in the gleba samples (cultivable fraction of bacteria – Fig. 4).
Effect of antimicrobial treatments on truffle volatiles. Mean levels (±SD – log scale) of thiophene volatiles (1) and (2) and fraction of cultivable bacteria in fruiting body homogenate of T. borchii upon treatment with:
A. H20/control or aqueous solutions of the antifungal agents clotrimazole
B. Amphotericin B.
C. Antibacterial agent streptomycin
n = 9 replicates per treatment. Compared with the initial concentrations of thiophene volatiles released at the beginning of the experiment (H20/Control, 0h), synthesis of thiophene volatiles was only blocked by the antibacterial agent streptomycin (C). The increase in thiophene concentrations observed in the samples treated with the antifungal agents (A, B) for 48 h were equivalent to the H2O/Control 48 h, demonstrating that truffle fungi are not involved in the production of thiophene volatiles. Different letters indicate significant differences among treatments (P < 0.05, ANOVA and Fischer LSD post-hoc test).
Bacteria fed with methionine produce a volatile related in structure to the thiophenes characteristic of T. borchii-fruiting bodies
Interaction of fungi with bacteria might activate fungal biosynthetic pathways that are otherwise suppressed in pure mycelial cultures (Scherlach and Hertweck, 2009; Schroeckh et al., 2009). We therefore co-cultured our Serratia bacterial isolate I (Table 1) with mycelia of white and black truffles, with or without an excess of L-methionine. Regardless of L-methionine addition, neither volatile (1) and (2) was detected. Surprisingly, another thiophene volatile, 2-methyltetrahydrothiophen-3-one [designated (3)], was detected in the presence of excess methionine not only from co-cultures of white or black truffle mycelia with Serratia but also from the single bacterial cultures without mycelium (Fig. 5). Cultures of truffle mycelia supplemented with extra methionine failed to produce volatile (3) when Serratia was absent (Fig. 5), confirming our hypothesis that the formation of thiophene derivatives by truffle-fruiting bodies depends on bacteria generating sulfur-containing heterocycles from linear precursor(s) such as L-methionine.
Methionine induces thiophene derivative 2-methyltetrahydrothiophen-3-one in Serratia.
A. Volatile (3) was not detected from T. borchii fruiting bodies.
B. It was detected whenever the bacterium was present.
C. It was not detected when the bacterium was absent.
Chromatograms (extracted mass m/z 116) are shown with normalized intensities.
Discussion
Some fungi are hotspot for bacteria (Warmink et al., 2009). Truffle-fruiting bodies do not derogate to this observation and are intensively colonized by bacteria that can reach densities up to 108 cells per gram of truffle (Barbieri et al., 2005; 2007; Antony-Babu et al., 2013). Similar to what has been previously described (Barbieri et al., 2005), samples of T. borchii analysed here were intensely colonized by bacteria. In line with previous reports about T. borchii (Barbieri et al., 2005), T. magnatum (Barbieri et al., 2007) and T. melanosporum (Antony-Babu et al., 2013), the bacterial community was dominated by α-Proteobacteria. These similarities between studies and between truffles species suggest that truffle-fruiting bodies, regardless the species, provide a specific habitat for some bacteria. This core microbiome of truffle-fruiting bodies might be supplemented with additional species depending on the fungal species, the maturation stage or the environment as suggested by the divergence observed for some classes. For example the abundance of β-Proteobacteria (Fig. 1) was much higher here compared with earlier reports on white truffles T. borchii (Barbieri et al., 2005) and T. magnatum (Barbieri et al., 2007).
α-Proteobacteria are very versatile in adapting to diverse environments ranging from the oceans (volcanoes on the ocean floor, surface waters) to soil (Ettema and Andersson, 2009). They also have an intriguing ability to interact with plants in either pathogenic or non-pathogenic mutualist/commensal relations (i.e. Rhizobium, Azospirillum). In fungi, the α-Proteobacteria phylum is dominant not only in truffles (Barbieri et al., 2005; 2007; Antony-Babu et al., 2013), but also in the ectomycorrhizal fungus Laccaria bicolor (Bertaux et al., 2005), suggesting the existence of some adaptative or competitive mechanism over other bacterial phyla. What creates the competitive advantage for α-Proteobacteria in truffles is unknown, but theoretically it might be driven by an increased ability to compete for specific resources (niche effect) or to degrade/produce toxic secondary metabolites (Hibbing et al., 2010). A study on the genome of 92 α-Proteobacteria revealed major genetic differences between those whose lifestyles were associated to plants (symbiotic and non-symbiotic) compared with those that were not plant-associated (Pini et al., 2011). By similarity, genetic adaptation might also exist in α-Proteobacteria associated to fungi and might possibly result from co-evolution between bacteria and their fungal hosts. In the specific case of the Périgord truffle T. melanosporum, the abundance of α-Proteobacteria was shown to increase in the truffle gleba over the period from a collection season (Antony-Babu et al., 2013). Taken together with the information that nitrogen fixation actively takes place in truffle-fruiting bodies of T. magnatum (Barbieri et al., 2010), and that some members of the α-Proteobacteria are well known for their ability to fix nitrogen, this suggests that the dominance of the latter bacterial phylum in truffles might be driven by special nutritional requirements. Another possibility might be that α-Proteobacteria might be better at degrading antibacterial agents (maybe the precursor of thiophene derivatives) than other bacterial phyla. Both theories, only speculative at this stage, will need to be tested experimentally.
In addition, our work suggests the existence of a structuration of the bacterial communities within the fruiting bodies of T. borchii. As observed in T. melanosporum (Antony-Babu et al., 2013), the composition of bacterial communities in the peridium diverged from the one of the gleba with an enrichment of bacteria belonging to the Bacteroidetes phylum in the peridium of both truffle species. Interestingly, our data also suggest that bacterial taxa could behave differently within the truffle with α-Proteobacteria tending to form ‘pure’ colonies, while β-Proteobacteria being present in mixed communities. These patterns suggest the existence of regulated network of interactions between bacteria in the truffle-fruiting bodies. Mechanisms regulating these networks and their consequence on the community function remain to be explored.
If the presence of bacteria in truffle-fruiting bodies is known since decades, the nature of the interactions occurring between bacterial communities and truffle fungi remain unclear. Do bacteria only consume truffle nutrients or do mutualistic interactions occur between the fungal host and the bacterial community? One of our aims was to investigate if thiophene derivatives, volatiles characteristic of T. borchii's aroma, were actually derived from the truffle fungus itself or from the microbiome inhabiting truffle-fruiting bodies. By following an approach similar to Koch's postulate in microbiology, we demonstrated that the bacterial community inhabiting T. borchii-fruiting bodies was responsible for the production of thiophene derivatives (1) and (2) characteristic of the fruiting body aroma of the latter species. Such intimate interactions between bacteria and fungi have been documented earlier, for example in the case of the rice fungal pathogen Rhizopus which toxins are derived from endobacteria of the Burkholderia genus (Partida-Martinez and Hertweck, 2005). Unlike in the Rhizopus/Burkholderia case, the interaction in truffles is not limited to a single bacterial genus, and indeed all bacteria tested in this study were able to produce thiophene volatiles from T. borchii-fruiting body extract, irrespective of whether they had been isolated from truffles or from other sources (i.e. plant, soil). This indicates that the ability to produce thiophene volatiles is widespread among bacteria and might possibly be linked to primary metabolism. This also raises the question of the ecological role of thiophene volatiles in nature.
Methionine was suggested to be the precursor of thiophene volatiles in T. borchii (Zeppa et al., 2004). Methionine did induce in our assays with Serratia volatile (3), a thiophene of related structure to volatiles (1) and (2), but not the latter two volatiles. There is no data in literature on the pathways or the genes leading to the synthesis of thiophene volatiles (1) and (2). However, data exist on the synthesis of volatile (3), both in yeast (Howell et al., 2005) and in bacteria (Nawrath et al., 2010). In the yeast Saccharomyces cerevisiae, volatile (3) might be derived from the cyclization and reorganization of 4-mercapto-4-methylpentan-2-one (4-MMP), itself released from the cleavage from a cystein moiety (Cys-4-MMP). Volatile (3) is furthermore strongly induced by the deletion of YAL012W, a putative carbon-sulfur lyase, suggesting an enzymatic biosynthesis route in yeast (Howell et al., 2005). A blast search of YAL012W against 3271 expressed sequence tags of T. borchii (Lazzari et al., 2007) did not reveal any match in the white truffle; this suggests that the latter gene might either not be expressed or might be absent in T. borchii. In the bacterium Chitinophaga Fx7914, volatile (3) is derived from two primary metabolites, homocysteine and pyruvate (Nawrath et al., 2010). By analogy to the intermediates and pathways of the latter two examples, thiophene volatiles (1) and (2) might be derived in truffles from primary metabolites synthesized or transformed by truffles, but also by yeasts (Buzzini et al., 2005) and bacteria.s
The production of volatiles (1) and (2) by bacteria from T. borchii gleba (or gleba extract) but not from other truffle species supports the inference that T. borchii is involved in the synthesis of at least one intermediate leading to thiophene volatiles (1) and (2). Additionally, the upregulation in sulfur metabolism in the fruiting bodies of T. borchii (Zeppa et al., 2010) and T. melanosporum (Martin et al., 2010; Splivallo and Maier, 2011; Splivallo et al., 2011) provides indirect evidence that truffles are actively involved in the production of sulfur metabolites during their sexual stage. Indeed the upregulation in homocysteine synthase (tbhos) and putative sulfate transporter (tbsul1) in mature T. borchii-fruiting bodies suggests that these genes might be indirectly involved in the synthesis of volatiles (Zeppa et al., 2010).
Overall, this strongly suggests that thiophene volatiles are derived from bacteria and further suggests that the non-volatile precursors of thiophene compounds might be produced by the intimate interactions of truffles, yeasts and bacteria. In contrast, other non-thiophene volatiles might be completely synthesized by the fungus itself. The complex blend of 20 to 50 molecules creating the aroma of T. borchii would thus originate from the mixed activity of the fungus and its microbiome.
Our results demonstrate that bacteria are central players in the production of thiophene volatiles (1) and (2) in T. borchii. The formation of other volatiles or the contribution of bacteria to aroma formation in other truffle species remains to be investigated. Similarly, the full biosynthetic pathway leading to thiophene volatiles and the exact contribution of yeasts and truffle remains to be elucidated. Our results open new horizons in the biotechnological production of fungal aromas suggesting that microbes might play a central role in the production of key fungal odorants (Splivallo and Maier, 2011). They also raise new ecological and mechanistic questions about the multitrophic interactions among truffles, their plant symbionts, mammals, insects and the bacterial community associated with truffle-fruiting bodies. By participating in the elaboration of sulphur volatile compounds that attract mammals, bacteria could indirectly participate in the dissemination of truffle spores and thus play a key role in the life cycle of the fungus.
Experimental procedures
Truffle-fruiting bodies
Truffle-fruiting bodies of T. borchii, T. magnatum, T. aestivum, T. mesentericum and T. brumale were purchased (Supporting Information Table S1) and were identified based on spore morphology (Ceruti et al., 2003).
Truffle mycelial cultures
Mycelia of T. borchii isolated more than 15 years ago (strain 1Bo = ATCC 96540 and 43Bo) (Bonuso et al., 2009) as well as a strain (strain IG2 – GenBank Accession KF414978) freshly isolated from a T. borchii-fruiting body which produced volatiles (1) and (2)) and T. melanosporum (strain TN1) were grown as described earlier (Splivallo et al., 2012). For the time series and feeding experiment with methionine, mycelial cultures and culture-negative controls (no mycelium) in malt extract broth were homogenized with a blender and transferred to SPME vials (aliquots of 4.5 ml per vial). Vials were supplemented with 0.5 ml of an aqueous solution containing either L-methionine (final concentration in SPME vials 5 mM) or water (control). Samples were either incubated in the dark or under 16 h photoperiods (to reflect natural conditions in the soil) at 23°C. Volatiles were measured by SPME-GC/MS after 1, 2, 3, and 5 days of incubation as described hereafter.
Selection of bacterial strains
Bacteria were isolated from the surface (peridium and adhering soil) and the inner part (gleba) of six fresh mature T. borchii-fruiting bodies (purchased in 2008 from TartufLangue, Cuneo, Italy). Unwashed truffles were shortly vortexed in a sterile 0.85% NaCl solution, and a ∼300 mg piece of gleba was subsequently excised and homogenized in 0.85% NaCl. Bacteria were isolated from the NaCl solutions by dilution plating on TSA 3% (Barbieri et al., 2005). A total of 29 strains were selected based on their colony morphology and colour (not shown). The full 16S rRNA was amplified by colony PCR using the universal primers ‘UP-Forward’ and ‘UP-Reverse’ (Barbieri et al., 2000; 2005). Sequencing was performed from both ends by Macrogen Europe (Amsterdam, the Netherlands).
The ability of bacteria to produce thiophene volatiles was tested on the following isolates: (i) A sub-selection of 13 bacterial isolates from our 29 strains was done reflecting the typical bacterial diversity in T. borchii (Barbieri et al., 2005) and included two strains of β- Proteobacteria, five strains of γ-Proteobacteria, three strains of the Bacteroidetes, two strains of the Firmicutes and one strain of the Actinobacteria (GenBank accession numbers are listed in Table 1). (ii) Five bacterial strains isolated from sources different from truffle-fruiting bodies (i.e. plant, soil) were also tested for their ability to produce thiophene volatiles. They included two strains of α-Proteobacteria and three strains of Serratia (Enterobacteriales) (Table 1). All bacterial isolates were stored at −70°C in 1:1 glycerol (85%) : tryptic soy broth (TSB). Before bioassays were performed, bacterial glycerol stocks were plated on TSA (3%) to check for purity.
Analysis of aroma profiles by SPME-GC/MS
Aroma of truffle gleba (300 mg), bacterial and mycelial cultures were profiled by SPME-GC/MS as described earlier (Splivallo et al., 2012). The occurrence of thiophene volatiles 3-methylthiophene (1) and 3-methyl-4,5-dihydrothiophene (2) was confirmed in T. borchii samples collected independently in Piedmont (n = 20 samples collected in 2006, 2008 – See Supporting Information Table S1). The identity of all volatiles listed in this study was confirmed through their MS fragmentation patterns, Kovats indices and when available with synthetic standards (i.e. 3-methylthiophene and see also Supporting Information Table S2).
Bioassays with fruiting body ethyl acetate extract
To test the ability of bacteria isolated from T. borchii-fruiting bodies and of the other bacterial isolates from soil and plants (Table 1), as well as the ability of truffle mycelium to generate thiophene volatiles, 490 g T. borchii-fruiting bodies were freeze dried, homogenized in a mortar and extracted overnight with 2 × 400 ml hexane followed by 2 × 400 ml ethyl acetate. The ethyl acetate fractions (800 ml) were pooled, concentrated till dryness in a vacuum evaporator at 50°C and resuspended in 450 ml malt extract broth (1%, pH 7.0). A negative control containing only dried ethyl acetate was prepared in the same way. The malt extract solution was sterile filtered (0.22 μm) and transferred (900 μl) to sterile SPME vials. SPME vials were either inoculated with (i) 100 μl of actively growing bacterial cultures (single strains in TSB 30 g l−1), with (ii) TSB only (negative control for the bacterial cultures) or with (iii) a pellet of 100 μl of Serratia Isolate I previously killed by autoclaving (another negative control for the bacterial cultures to make sure that active bacterial growth was required to produce thiophene volatiles) – see Supporting Information Fig. S1 for experimental design – or with (iv) T. borchii mycelium strain 43Bo or IG2, or with (v) malt extract broth (control for the mycelial cultures described in Truffle mycelial cultures). SPME vials containing bacterial cultures (and the malt extract control) were incubated under gentle vertical shaking at 15°C, reflecting spring soil temperatures typical of Italian regions where T. borchii truffles are collected. Furthermore, emission of thiophene volatiles and bacterial growth were determined after 25 h of incubation, which corresponds to the late exponential growth phase for our Serratia isolate I strain and the maximum emission of thiophene volatiles (1) and (2). SPME vials containing homogenized truffle mycelium were incubated for 25 h in the dark at 23°C (optimal growth condition for T. borchii) before generating volatile fingerprints.
Determination of bacterial growth
Bacterial growth was measured at 600 nm (OD600) after blank subtraction (100 μl of bacterial culture or malt extract broth per well) in an Epoch 96-well microtiter plate spectrophotometer (BioTeck Instruments, VT, USA).
Bioassays with dried fruiting body and mycelium homogenates
Truffle-fruiting bodies (1–3 g per species) of T. borchii, T. magnatum, T. aestivum (syn. T. uncinatum), T. mesentericum, T. brumale (Supporting Information Table S1) or mycelial cultures of T. borchii (strains 1Bo and 43Bo) (Bonuso et al., 2009), and T. melanosporum (strain TN1) and negative culture controls (prepared as described in Truffle mycelial cultures), were freeze dried and each sample was homogenized separately to a fine powder in a mortar. The homogenates were sterilized with 10 ml CH2Cl2 (overnight incubation), after which CH2Cl2 was removed from the homogenates in a two-step process (vacuum evaporation at 50°C for 6 h followed by heating to 80°C for 3 h). A quantity of 50 mg of sterile fruiting body homogenate or mycelium homogenate was transferred to a sterilized 20 ml SPME vial, which was then inoculated with either 100 μl bacterial glycerol stock (1:1 bacterium in TSB : 85% glycerol), or with 100 μl 1:1 TSB : 85% glycerol as control or with truffle mycelium following the same scheme described in Bioassays with fruiting body ethyl acetate extract. The detailed scheme of the growth conditions and volatile sampling is described in Supporting Information Fig. S2.
Bioassays with mixed cultures truffle mycelium/bacteria
A quantity of 2.0 g of truffle mycelial cultures (T. borchii strains 1Bo and 43Bo, and T. melanosporum strain TN1) and culture-negative controls (without mycelium), prepared as described earlier (Splivallo et al., 2012), were transferred to 20 ml SPME vials and inoculated with 100 μl of Serratia sp. isolate I (identified as Serratia sp. based on 16S rRNA sequencing) glycerol stock (1:1 bacterium grown in TSB:85% glycerol) or with 100 μl 1:1 TSB : 85% glycerol as control. Samples in SPME vials were either incubated as such or supplemented with excess L-methionine (3 mg per SPME vials supplied in 100 μl H2O) in order to check if excess methionine induced the production of thiophene volatiles. Samples were incubated for 48 h at 23°C in the dark before generating the aroma fingerprinting by SPME-GC/MS.
Treatment of fruiting bodies with antibacterial and antifungal agents
Two T. borchii truffles (∼40 g in total) were homogenized in a mortar until obtaining a paste which was then transferred to 2.0 ml Eppendorf tubes containing each 300 mg truffle homogenate. Sterile water was added to the samples: 1.0 ml per tube of either pure water (control) or water containing the broad-spectrum antibacterial agent streptomycin (400 μg ml−1) or one of the antifungal agents clotrimazole (400 μg ml−1) or amphotericin B (250 μg ml−1). Tubes were closed and incubated at 20°C on a rotary shaker (160 r.p.m.). After 48 h incubation, samples were centrifuged (2 min at 12 000 g), the aqueous phase was removed from the gleba homogenate, and the latter was then transferred to 20 ml SPME vials to quantify thiophene volatiles (1) and (2). Additionally, volatile fingerprinting was performed on 300 mg gleba homogenate at the beginning of the experiment (0 h) to measure the initial quantities of thiophene volatiles (1) and (2).
Fluorescence in situ hybridization (FISH)
Six fruiting bodies of T. borchii were washed with tap water, dried and cut using a sterile scalpel to generate four subsamples of 500 mg each per truffle-fruiting body. Samples were stored in 2.0 ml Eppendorf tubes and allowed to age at room temperature for 0 (samples processed immediately), 2, 4 and 6 days. Each subsample was divided in two parts, one 300 mg sample for volatile fingerprinting by SPME-GC/MS and one 200 mg sample which was fixed for subsequent FISH analysis. Fixation was performed in 2.0 ml Eppendorf tubes by incubating samples overnight at 4°C in 1200 μl of a 2.25% paraformaldehyde solution in phosphate saline buffer (PBS) – PBS = 8 g of NaCl, 0.2 g of KCl, 1.44 g of Na2HPO4, 0.24 g of KH2PO4 in 100 ml distilled H20, pH adjusted to 7.4. Samples were washed with 3 × 1.0 ml ice-cold PBS after fixation and stored in 1:1 EtOH96% : PBS at −20°C until FISH analysis. FISH was performed on fruiting body homogenates (gleba samples at days 0, 2, 4 and 6 and peridium samples at day 0) or on thin sections of gleba and peridium samples (30 μm) as described for T. magnatum (Barbieri et al., 2007) with the probes listed in Supporting Information Table S3. Each sample was co-hybridized with the universal probe mix Eub338 coupled to FITC to quantify bacterial population and a phylum specific probe coupled to cy3 (Supporting Information Table S3). Signals representing all bacterial cells and specific phyla were quantified visually on homogenized samples after image acquisition under an epifluorescence microscope (BX41, Olympus). Each homogenized sample was mounted and hybridized independently two times and three images were acquired for each replicate (6 images in total). The mean bacterial cell number was normalized to the sample biomass and expressed as [cell count/mg dry weight of gleba]. Fluorescent signals representing Firmicutes, α-, β-, γ-Proteobacteria, Bacteroidetes and Enterobacteriales (Supporting Information Table S3) were counted directly under the fluorescence microscope and expressed as the percentage of Eubacteria (mean of n ≥ 3 observation from the same sample). Values expressed as %Eubacteria were then transformed to [cell count / mg dry weight of gleba] after multiplication by the total bacterial cell count in the sample.
Thin sections of T. borchii-fruiting bodies (30 μm) were hybridized with the probes specific for α- and β-Proteobacteria listed in Supporting Information Table S3 and observed with a Radiance 2100 Rainbow confocal microscope (Bio-Rad).
Data processing
All bioassays in SPME vials performed to check the capacity of bacteria or truffle mycelia to produce thiophene volatiles comprised at least three independent replicates. Treatment of T. borchii fruiting body with antimicrobial agents was performed on four replicates per treatment. Six T. borchii-fruiting bodies were used for FISH (subsamples at 0, 2, 4 and 6 days of aging).
Volatiles fingerprints were processed as described earlier (Splivallo et al., 2012). Specific m/z typical of thiophene volatiles were furthermore extracted from the total ion chromatograms for data visualization and statistics (m/z 97 occurs both in compounds (1) and (2), while m/z 100 occurs in (2) only and m/z 106 in 2-methyltetrahydrothiophen-3-one (3) only – MS fragmentation patterns are shown in Supporting Information Fig. S3.
Acknowledgements
We would like to thank the following people for their support and for material provided: Prof. Ivo Feussner and Dr. Cornelia Herfurt for access to the GC/MS and Dr. Laure Weisskopf for providing the Serratia strains MG1, B5a and IC14 (Supporting Information Table S1); Prof. Alessandra Zambonelli for T. borchii mycelia strains 1Bo and 43Bo, and Dr. Gérard Chevalier for T. melanosporum mycelium strain TN1. Financial support was provided by the Deutsche Forschungsgemeinschaft – DFG (http://www.dfg.de, DFG Grants SPL1191/2-1 and SPL1191/4-1 to RS). AD lab is supported by the Laboratory of Excellence ARBRE (ANR-11-LABX-0002-01).
Conflict of interest statement
RS declares that a patent has been filed regarding the production of truffle aroma using truffle-associated microbes (Splivallo and Maier, 2011). The other author(s) declare that they have no competing interests.
