Bacteria of the genus Rhodopseudomonas (Bradyrhizobiaceae): obligate symbionts in mycelial cultures of the black truffles Tuber melanosporum and Tuber brumale

Background This work aimed at characterizing 12 isolates of the genus Tuber including Tuber melanosporum (11 isolates) and Tuber brumale (one isolate). This was done using internal transcribed spacer (ITS) sequences, confirming their origin. Results Analysis of their mating type revealed that both MAT1-1 and MAT1-2 exist within these isolates (with 3 and 8 of each, respectively). We observed that each of these cultures was consistently associated with one bacterium that was intimately linked to fungal growth. These bacterial associates failed to grow in the absence of fungus. We extracted DNA from bacterial colonies in the margin of mycelium and sequenced a nearly complete 16S rDNA gene and a partial ITS fragment. We found they all belonged to the genus Rhodopseudomonas, fitting within different phylogenetic clusters. No relationships were evidenced between bacterial and fungal strains or mating types. Rhodopseudomonas being a sister genus to Bradyrhizobium, we tested the nodulation ability of these bacteria on a promiscuously nodulating legume (Acacia mangium), without success. We failed to identify any nifH genes among these isolates, using two different sets of primers. Conclusions While the mechanisms of interaction between Tuber and Rhodopseudomonas remain to be elucidated, their interdependency for in vitro growth seems a novel feature of this fungus. Electronic supplementary material The online version of this article (doi:10.1186/s40064-016-2756-6) contains supplementary material, which is available to authorized users.


Background
Truffles, hypogeous ascomycetes belonging to the genus Tuber, include ectomycorrhizal species of major socioeconomic interest. Some species, such as Tuber melanosporum, Tuber magnatum and Tuber aestivum, are edible and have great market value. Production of truffles depends on tree saplings (of species belonging to genera such as Quercus, Corylus and Tilia) appropriately inoculated with fungal inoculants, produced within traditional or industrial nurseries. Various forms of inocula may be used in these nurseries, ranging from soil under productive trees, through crushed fresh, deep frozen or dried fruit bodies (whole or as debris), to mycelial cultures. Several authors have shown that it is possible to synthesize mycorrhizas with Tuber mycelial cultures (Chevalier and Frochot 1997;Sisti et al. 1998). However, Tuber species are generally difficult to isolate and cultivate in laboratory conditions, species like Tuber borchii (Barbieri et al. 2005), Tuber rufum, Tuber uncinatum and Le Roux et al. SpringerPlus (2016) 5:1085 Tuber macrosporum (Iotti et al. 2002) being among the easiest.
At least for two species (T. magnatum and T. melanosporum) the long-standing question of whether Tuber species are homo-or heterothallic was recently solved with the identification of two mating type loci carrying either MAT1-1-1 or MAT1-2-1 genes (Rubini et al. 2011). MAT1-1-1 and MAT1-2-1 encode a protein with an alpha domain, and a high-mobility DNA binding protein (HMG), respectively. In these species, sexual reproduction, which is necessary for fructification, only occurs between two different mating types. In fructification, each spore is of only one mating type, and the proportion of spores of each mating type is about 50 %. The gleba is from only one mating type, identical to that of all the ectomycorrhiza surrounding the fructification (Rubini et al. 2011). Mating type identification is thus a major challenge on the way to mastering Tuber fructification in the soil. Strain Mel28 of T. melanosporum, whose genome has been fully sequenced (Martin et al. 2010), is of the MAT1-2-1 type. Having mycelial cultures representative of different mating types would be of great interest both for lab experiments and plant tests.
Bacteria are known to be ubiquitously associated with ectomycorrhization (for the concept of mycorrhiza helper bacteria, see the review by Frey-Klett et al. 2007) and neither truffle ectomycorrhiza nor their fruit bodies (Table 1) are exceptions. Recently, Mello et al. (2013) showed that "brûlés"-burnt areas around productive trees-of T. melanosporum markedly affected soil bacterial communities. These bacteria may have various effects on Tuber mycelium growth, including inhibition and promotion. For example, one strain of Staphylococcus aureus has been shown to produce volatile organic compounds potentially involved in T. borchii mycelial growth inhibition (Barbieri et al. 2005). Unexpected procaryotic functional activities like nitrogen fixation have been evidenced in ascocarps of T. magnatum by Barbieri et al. (2010). Bacteria have been shown to participate in truffle aroma elaboration through the production of thiophene volatiles (Splivallo et al. 2015). In a recent paper, Benucci and Bonito (2016) observed by 454 pyrosequencing the dominance of the genus Bradyrhizobium within Tuber ascocarps of various geographic origins, but not in other truffle genera like Kalapuya, Terfezia or Leucangium.
However, information is still lacking on the characteristics of mycelial cultures of T. melanosporum, including their associated bacteria. The aim of this study is to identify the bacterial strains associated with T. melanosporum and T. brumale in culture, as it could help to control mycelial isolation, to mass produce Tuber inoculum and to generate truffle productive saplings.

Isolation and culture of Tuber mycelium
Mycelia were originally isolated from ethanol-sterilized fruit bodies of T. melanosporum and T. brumale (Table 2), by axenically placing a piece of gleba on solid Maltea Moser medium, according to Chevalier (1972). All isolates were routinely subcultured on 2 % Cristomalt ® (Difal, Seysses, France) agar medium (modified from Chevalier (1972), Maltea Moser being replaced by Cristomalt), at 25 °C in the dark. When necessary the antibiotics chloramphenicol, tetracycline, gentamicin and Table 1 Diversity of bacterial genera characterized either directly ("ascocarps") or after isolation from ascocarps ("isolates") or mycelial cultures of different Tuber species in recent publications streptomycin were individually added to the medium at concentrations routinely used in the lab for ectomycorrhizal mycelium cultivation, i.e., 50, 10, 10, and 80 mg l −1 , respectively (Bâ et al. 2011).
Bacterial strain cultivation assays were attempted on yeast mannitol agar (YMA) medium (Vincent 1970), classically used for cultivating Bradyrhizobium and Rhodopseudomonas in the lab. Bacterial strains were named by placing B before the number of the fungal strain with which they were associated (e.g., BMel18 for the bacteria associated with the fungal strain Mel18).

Microscopic observations
Changes in mycelial and bacterial growth were followed in Petri dishes examined with a Nikon AZ100 microscope.

Molecular characterization
Total fungal DNA was extracted using REDExtract-N-Amp polymerase chain reaction (PCR) kits (Sigma-Aldrich, St. Louis, MO) according to the manufacturer's instructions. Mycelium was confirmed as Tuber by analysis of internal transcribed spacer (ITS) sequences using the highly conserved fungal rRNA gene primers ITS1F (Gardes and Bruns 1993) and ITS4 (White et al. 1990) for PCR. Each PCR reaction (25 µl) contained 2 µl of template DNA, 1× Reaction Buffer (1.5 mM MgCl 2 ), 200 µM of each dNTP, 0.5 µM of each primer, 2× bovine serum albumin and 1 U of GoTaq ® DNA polymerase (Promega Corporation, Madison, Wi). The PCR thermal protocol consisted of an initial 5 min denaturation step at 95 °C, 35 amplification cycles of 95 °C for 30 s, 52 °C for 1 min, 72 °C for 1 min, and a final extension step of 72 °C for 10 min. After agarose gel electrophoresis, gel bands of the expected size were excised and PCR products were purified using Illustra GFX PCR DNA and Gel Band Purification Kit (GE Healthcare, UK). DNA was sequenced (Genoscreen, France) with the same primer ITS1F as used for PCR. Complementarily, Tuber mating types were determined by PCR with each of the pairs of primers dedicated to MAT1-1 and MAT1-2 according to Rubini et al. (2011).
Bacteria were characterized according to their nearly complete 16S rDNA sequences and their partial 16S-23S rRNA ITS. A loopful of bacterial cells, taken from the margin of the mycelial colony, was suspended in 20 µl of sterile water and cell debris removed by centrifugation at 13,000 rpm for 1 min at room temperature; 2 µl of the supernatant was used as a template for PCR.
For nifH genes, two pairs of primers were tested, nifHF/ nifHI (Laguerre et al. 2001) and polF/polR (Poly et al. 2001). They were tested on four randomly chosen bacteria: BMel18, BMel28, BmelC89 and BBTR3. The PCR mix and thermal conditions were as described earlier except for the annealing conditions: 57 °C for 1 min with nifHF/ nifHI and 55 °C for 30 s with polF/polR. A positive control was performed using Bradyrhizobium diazoefficiens Type strain USDA 110 originally isolated from soybean nodule in Florida, in 1957 (Delamuta et al. 2013).

Nucleotide sequence analyses
Fungal ITS, bacterial 16S rRNA and ITS 16S-23S rRNA sequences were corrected using the sequence viewer program 4 Peaks (http://4peaks.en.softonic.com/mac, accessed 11 March 2016). The fungal ITS, bacterial 16S rRNA and ITS sequences were deposited in GenBank and their accession numbers are presented in Tables 3 and 4. For bacterial characterization, multiple alignment and phylogenetic tree construction were performed using the multiplatform program SeaView version 4 (Gouy et al. 2010). This interface drives the Clustal Omega program for multiple sequence alignments and includes the BioNJ distance-based tree reconstruction method and the maximum likelihood (ML) based phylogeny program PhyML.

Results
All isolates of T. melanosporum and T. brumale examined in the present study exhibited slow growth. They required 2-3 weeks to initiate a new, visible mycelium crown around the plug. The growth rate appeared to be slightly accelerated when regularly subcultured. Slow growth is a general feature of the genus Tuber, depending on the media, which are generally based on malt or potato dextrose agar. On these solid media, growth generally takes 4-8 weeks to stabilize at its maximum level. In our conditions, the average radial mycelial growth was estimated as 1 cm in 3 weeks. All the Tuber cultures included bacterial associates. The addition of chloramphenicol or tetracycline to the Cristomalt agar medium had little effect on bacterial (and mycelial) growth. Others antibiotics such as gentamicin or streptomycin totally blocked both mycelial and bacterial development. These bacterial isolates generally grew very poorly on YMA media, once isolated from Tuber mycelia, and did not survive repeated subculturing, limiting the possibilities  of enzymatic or antibiotic resistance characterization. Under the microscope, mycelium generally appears as the first medium colonizer, outgrowing from the plug, the bacterial associate proliferating around growing hyphae and ensheathing them with a slight delay (Fig. 1a, b). As presented in Table 3, fungal ITS sequencing confirmed the original taxonomic identity of each of the cultivated Tuber isolates. The mating types of the different mycelial cultures (Table 3) are presented in Additional file 1, three and eight of them being of the MAT1-1 and MAT1-2 type, respectively. The mating type being generally determined by PCR response to each pair of primers, not followed by sequencing, we tested these primers on some of the bacterial DNA extracts (BMelBal3, BMelC89, BMel2VDA3 and BMel3VDA4): none of them allowed us to obtain an amplicon. The PCR positive controls performed on fungal DNA representative of the two mating types (T. melanosporum strains MelC89 and Mel2VDA3 for MAT1-1 and MAT1-2, respectively) were all positive, with a band of the expected size (421 and 550 bp for MAT1-1 and MAT1-2, respectively; not illustrated).
The homologies of the nearly full-length 16S rRNA sequences of 11 bacterial isolates and one partial 16S rRNA sequence for BMel14 are presented in Table 4. All these sequences were close to Rhodopseudomonas spp. (Table 4) with an identity percentage of 99-100 %. The phylogenetic tree (Fig. 2) allows us to identify a first cluster of five strains closely related to Rhodopseudomonas sp. strain N-I-2, an endophytic bacteria isolated from Prunus avium (Quambush et al. 2014). The second  Barbieri et al. 2005Barbieri et al. , 2007 were also included in this phylogeny. They were selected as representative of the different 16S clusters obtained by these authors. They appear to be close to Bradyrhizobium strains, B. elkanii for T. borchii, and in a separate cluster for T. magnatum.
Within the Bradyrhizobiaceae, all these associates of white truffle ascomata were quite distinct from our black truffle mycelial associates (Fig. 2). The mean length of the partial 16S-23S rRNA ITS sequence was 770 bp. BLASTn analysis (Table 4) and phylogenetic tree reconstruction (Fig. 3) confirmed that all mycelial bacteria clustered within the Rhodopseudomonas clade, which also included the strain Bradyrhizobium sp. CCBAU 85059.
The trials that we carried out to test symbiotic characteristics such as nitrogen fixation and nodulation on our Tuber associated strains, remained unsuccessful: both pairs of nifH primers that we tested repeatedly failed to give a PCR product related to a nifH gene. The positive control with B. diazoefficiens type strain gave a band of the expected size with both pairs of primers. None of the bacterial strains nodulated the promiscuous legume A. mangium 3 months after inoculation. In these routinely used culture conditions, nodulation is known to usually occur within 2-3 weeks after inoculation (Perrineau et al. 2011).

Discussion
As reported by Iotti et al. (2002), first isolation from inner ascocarp tissue is generally not too difficult as compared with subsequent subculturing, as many cultures do not survive that step, a phenomenon that had already been mentioned by Chevalier (1972). Both these papers also reported lag phases of different duration according to the species, with no relationships, regarding these lag phases, between the first outgrowth and the subsequent subculturing. They also reported that optimal culture media for isolating and subculturing could be different: casein hydrolysate and Maltea Moser, respectively (Chevalier 1972); or modified Woody Plant Medium and Malt or Potato Dextrose Agars, respectively (Iotti et al. 2002). Among important factors that may influence growth are pH, temperature and dietary elements (Michaels 1982, https://ir.library.oregonstate.edu/xmlui/handle/1957/ 9224, accessed 11 March 2016, most optimal temperatures being around 20 °C, and pH over 7. In such conditions, maximum growth was reached after 7 weeks for T. melanosporum, with marked intraspecific variations. This species was one of the slowest among the six tested Based on ribosomal sequence analyses, bacterial associates were all found to belong to Rhodopseudomonas, a genus of alpha proteobacteria closely related to Bradyrhizobium (Giraud and Fleischman 2004). Among Rhodopseudomonas species, R. palustris is a photoautotrophic bacterium, taxonomically close to some photoheterotrophic Bradyrhizobium species, that efficiently nodulates stems and roots of the legume Aeschynomene. At this time, no strains of R. palustris are described as nodulating legumes, but they quite commonly harbor nifH genes (Cantera et al. 2004). The low discriminatory power of 16S rRNA has long been recognized within the Bradyrhizobiaceae , necessitating a complementary characterization using other targets, such as 16S-23S ITS. Both these targets allowed us to confirm the close relationships between Rhodopseudomonas and Bradyrhizobium. Moreover, when we reblasted the ITS sequences of Bradyrhizobium sp. CCBAU 85059, isolated, as strain CCBAU 85080, from Astragalus tatsienensis nodules in Tibet (Hou et al. 2009), the closest identified strains belonged to the genus Rhodopseudomonas. This is probably a result of the fact that these authors only considered the genus Bradyrhizobium in their phylogenetic analyses. Concerning strain R_45974 of T. robiniae, isolated from root nodules of Robinia pseudoacacia, it was previously described as Rhodopseudomonas sp. (De Meyer et al. 2011) on the basis of a 16S rRNA gene phylogeny. More recently, it has been re-assigned to this new genus in Bradyrhizobiaceae after complementary characterization comprising physiological and biochemical tests, and sequencing of housekeeping genes (De Meyer et al. 2012).
More generally regarding molecular characterizations, no particular clustering of the bacterial sequences was detected in regard to our black truffle ascocarp species or geographical origin, mycelial strain or mating type: there is not, at this stage, any evidence of specificity between a given Tuber mycelium and its associated Rhodopseudomonas. However, the fact that, despite a relatively heterogeneous geographic origin of the ascocarps, all the mycelia-associated strains fall within the same genus Rhodopseudomonas is consistent with a non-random  On the fungal side, according to Chevalier (1972) elimination of the mycelial bacteria frequently leads to the loss of the corresponding T. melanosporum culture. This author reported that one of these bacteria was attributable to the genus Arthrobacter, with the identification tools available at that time. More recently, Barbieri et al. (2000Barbieri et al. ( , 2002 showed that T. borchii mycelial cultures were associated with unculturable bacteria of the Cytophaga-Flexibacter-Bacteroides phylum. Remarkably, some of these bacteria were detected as viable within the hyphae. While the presence of other gram negative bacteria in mycelial cultures has already been reported, few or none of these associates were precisely identified. Such studies to characterize the dependency of these Tuber-bacteria associations are made extremely difficult owing to (1) the slow growth of Tuber mycelium and (2) the non-cultivability of associated Rhodopseudomonas and thus difficulties in characterizing and eliminating bacteria from mycelial cultures.
Our Tuber isolates originating from ascocarp inner tissue, it seems likely that in vitro mycelia-accompanying bacteria could also be of ascocarpic origin. Although, after several decades, the original ascocarpic material is no longer available to check the identity of myceliumassociated strains to original ascocarpic bacteria, we know from the literature that truffle ascocarps have been shown to host a number of different microbes including yeasts (Buzzini et al. 2005), fungi (Pacioni et al. 2007) and bacteria (Table 1). Some bacterial genera have repeatedly been reported as colonizing ascocarpic tissue of various Tuber species (Table 1). They belonged to different lineages of proteobacteria, most of them being within alpha or gamma proteobacteria. Among the dominant genera, Bradyrhizobium and Pseudomonas were almost universally reported whatever the Tuber species under consideration and be it after isolation or directly from total ascocarp DNA. It has to be noted that bacterial communities varied according to the degree of maturation of the ascocarp as shown with T. melanosporum (Antony-Babu et al. 2013) and T. magnatum (Barbieri et al. 2007). Bradyrhizobium is over-represented within bacterial communities directly characterized from the ascocarp, as is the case for Pseudomonas among bacterial isolates. In T. magnatum ascocarp, Barbieri et al. (2010) reported significant amounts of nitrogen fixation (nitrogenase activity estimated by acetylene reduction), together with the presence of nifH genes in ascocarps at different degrees of maturation. The phylogenetic positions of bradyrhizobia associated with ascocarps of T. borchii (Barbieri et al. 2005) and T. magnatum (Barbieri et al. 2007) showed that none of them clustered with our T. melanosporum or T. brumale mycelial bacteria. As Antony-Babu et al. (2013) remind us, Bradyrhizobium is "consistently found at all stages of the maturation process and in different truffle species". These authors suggested that the selection of Bradyrhizobiaceae results from deterministic events allowing these bacterial taxa to tolerate and colonize the particular environment of the ascocarpic tissues (with sulfur-containing molecules and aromatic volatile compounds).
During the isolation steps, two main pitfalls have to be avoided: the duration of hyphal outgrowth from the Tuber explant in competition with other ascocarp inhabiting microbes, and failure to grow after the first subculture, a particular fate reported by Giomaro et al. (2005). These authors argue that this could be due to the non-acclimation of the ascocarpic mycelium to the saprophytic stage of in vitro growth. A positive role of the Tuber associated bacterial strains as helpers in this progressive adaptation to the new lifestyle cannot be excluded. However, in the absence of pure, bacteria-free cultures of Tuber mycelium, all the confrontation trials (co-cultivation of Rhodopseudomonas-associated mycelium and freshly isolated Rhodopseudomonas, on various solid media) we attempted (data not shown) were inconclusive in terms of fungal growth response (radial mycelial growth).
In this study, we observed that whatever the origin of the ascocarp, none of the 12 mycelial cultures was devoid of Rhodopseudomonas associates. However, these associates are genetically diversified in several clusters, for both 16S and ITS rDNA, without evidencing any relationship between strain clustering and criteria such as geographical origin of ascocarps, age of mycelial culture since its isolation (27-15 years), isolation operator, mycelium mating type or Tuber species. In a separate experiment, conducted in 2013 to isolate bacterial associates from fresh T. borchii and T. melanosporum ascocarps, we never obtained Rhodopseudomonas strains among 123 and 126 bacterial isolates from each ascocarp, respectively (data not shown), which seems in accordance with non-cultivability of these Tuber associates. Remarkably, Bradyrhizobiaceae are often well represented among ascocarpic DNA, whatever the Tuber species, but absent from isolates in the different studies listed in Table 1, the only exception being one isolate from T. magnatum that grouped with bacteria of the genus Bosea, another member of the Bradyrhizobiaceae (Barbieri et al. 2007). Attempts to separate mycelium from bacteria (on several selected antibiotics) remained unsuccessful. These Rhodopseudomonas associates appear to be consistently essential to mycelial life and development. In the same way, repeated subculturing of isolated Rhodopseudomonas induced a rapid decline and loss of isolates. There appears to be a reciprocal dependency for longlasting in vitro growth of both associates.

Conclusions
In the production of Tuber-inoculated plantlets for truffle producers, the use of mycelial cultures is possible but limited by both the availability of fungal cultures and by the mass production of fungal inoculants. This work shows the constant occurrence of Rhodopseudomonas associates in the production of Tuber mycelium. The availability of both mating types among these mycelial cultures is also of major interest as the occurrence of two compatible mating types is essential to Tuber fructification. The marketing of Tuber-associated plants, estimated to involve over 500,000 plants per year, could benefit from re-considering the use of mycelial inoculants (instead of applying, at a rate of at least 1 g per plant, crushed truffle fructifications whose market price is about 1000 euros per kilogram) based on the combination of both mating types and their associated Rhodopseudomonas. Such practices would allow a better mastering of inoculant quality and consistency, and possibly later on improve and accelerate field fructification of black truffles.