The arbuscular mycorrhizal symbiosis influences sulfur starvation responses of Medicago truncatula


Author for correspondence:

Franziska Krajinski

Tel: +49 331 5678355



  • Arbuscular mycorrhizal (AM) symbiosis is a mutualistic interaction that occurs between the large majority of vascular plants and fungi of the phylum Glomeromycota. In addition to other nutrients, sulfur compounds are symbiotically transferred from AM fungus to host plants; however, the physiological importance of mycorrhizal-mediated sulfur for plant metabolism has not yet been determined.
  • We applied different sulfur and phosphate fertilization treatments to Medicago truncatula and investigated whether mycorrhizal colonization influences leaf metabolite composition and the expression of sulfur starvation-related genes.
  • The expression pattern of sulfur starvation-related genes indicated reduced sulfur starvation responses in mycorrhizal plants grown at 1 mM phosphate nutrition. Leaf metabolite concentrations clearly showed that phosphate stress has a greater impact than sulfur stress on plant metabolism, with no demand for sulfur at strong phosphate starvation. However, when phosphate nutrition is high enough, mycorrhizal colonization reduces sulfur stress responses, probably as a result of symbiotic sulfur uptake.
  • Mycorrhizal colonization is able to reduce sulfur starvation responses in M. truncatula when the plant's phosphate status is high enough that sulfur starvation is of physiological importance. This clearly shows the impact of mycorrhizal sulfur transfer on plant metabolism.


Nutrient availability is of paramount importance for the growth, performance, yield and reproductive success of plants, including crop plants, which are particularly dependent on a balanced nutrient supply (Marschner, 1995). At the molecular physiological level, several single nutrient ion studies have been performed (Amtmann & Armengaud, 2009), which have shown that nutrient starvation responses are widely interconnected (Astolfi et al., 2010; Watanabe et al., 2010). Sulfur (S) is usually taken up as sulfate by the plant roots, with any excess sulfate being stored in vacuoles. Minor amounts of volatile S compounds (e.g. H2S, SO2 and dimethyl sulfate (DMS)) can also be taken up and metabolized by the leaves (Durenkamp & De Kok, 2004). In addition to the water mobile pedospheric sulfate, organic matter contains substantial amounts of sulfuric compounds, for example, organic esters, which are released through the activity of soil bacteria, potentially providing substantial amounts of S for the soil (Schmalenberger et al., 2009).

Sulfur starvation leads to strong starvation phenotypes in plants, including reduced growth, yellowing of the leaves, and decreased seed number and viability. Recently, the impact of S starvation on the yield of crop plants has been receiving more attention as the reduced atmospheric input of S-containing compounds from industrial and other types of pollution is reducing the amount of S available to crop plants (Lehmann et al., 2008). This has led to a change from an excess supply of S to moderate or even starvation conditions in some areas.

The formation of an arbuscular mycorrhizal (AM) symbiosis leads to an improved nutritional status of the plant partner under low nutrient availability (Marschner & Dell, 1994; Clark & Zeto, 2000). Several studies showed the transfer of mineral nutrients such as phosphate and nitrate from the surrounding area through the fungal hyphae to the plant partner (Ames et al., 1983; Smith et al., 1994, 2003; Javot et al., 2007a; Leigh et al., 2009). The impact of AM symbiosis on phosphate uptake is quite well understood, and it has been demonstrated that plants possess a special, symbiotic phosphate uptake pathway (Javot et al., 2007b; Nagy et al., 2009). Pioneering work in the 1970s and 1980s also showed the uptake of sulfate by the mycorrhizal fungus and its translocation to the plant root and shoot (Rhodes & Gerdemann, 1978a,b; Hepper, 1984). This often resulted in increased growth, especially under S starvation conditions. A recent study using carrot (Daucus carota) root organ cultures also demonstrated the symbiotic transfer of organic S-containing compounds such as cysteine, methionine and glutathione (Allen & Shachar-Hill, 2009). The fact that S, as well as organic S-containing compounds, can be transported via the mycorrhizal pathway suggests that the AM symbiosis might also be beneficial for the plant's S nutrition. However, because S availability has not been a limiting factor in agriculture, the impact of mycorrhizal S transfer on plant metabolism is poorly understood. As leaves are the major sinks for plant S compounds and the main site of sulfate assimilation, the analysis of leaf metabolites was included in our study.

In previous decades, plant model systems such as Medicago truncatula were established for the analysis of the molecular background of the AM symbiosis. However, little information is available about the S assimilation pathway in M. truncatula. The influence of S nutrition on root nodule symbiosis has been studied in white cloverTrfolium repens (L.), and it was shown that the amount of S supplied to clover plants influenced the number of nodules and nitrogen (N)-fixing activity, similar to the starvation effects of phosphorus (P) and potassium (K) (Varin et al., 2010). Additionally, a symbiotic sulfate transporter, Sst1, of Lotus japonicus is necessary for nodule formation and essential for the symbiotic supply of S to the bacteria (Krusell et al., 2005). Little is known about the physiological impact of S transfer during the AM symbiosis. We assume that, under S starvation, the mycorrhizal S supply plays a significant role in plant S nutrition and thus also influences other cellular functions. Recent elemental analysis of amounts of S, N and P confirmed that mycorrhizal symbiosis is able to increase sulfate uptake and assimilation, and an analysis of the transcription level of predicted sulfate transporters showed a very complex picture of sulfate uptake pathways in M. truncatula (Casieri et al., 2012). However, S and P metabolisms are strongly connected and, in addition to N, the availability of P and S considerably influences plant growth (Watanabe et al., 2012). To investigate the correlation between phosphate nutrition, mycorrhizal colonization and S nutrition, we investigated the influence of mycorrhizal S uptake on plant S nutrition and growth, taking into account its link to phosphate availability. We determined the response of M. truncatula to S deprivation in the background of phosphate nutrition at the physiological and molecular levels. Another aim of the study was to investigate known sulfate starvation response genes, whose homologs were analyzed in M. truncatula, as well as the metabolite responses of S-containing metabolites and amino acids.

Materials and Methods

Growth, inoculation and fertilization of Medicago truncatula plants

Medicago truncatula Gaertn. cv Jemalong line A 17 seeds were incubated with concentrated sulfuric acid, followed by repeated washings with distilled water. Surface sterilization was carried out using 6% (v/v) sodium hypochlorite for 8 min. After rinsing with sterile water, seeds were germinated on water agar at 4°C for 48 h, followed by a 2-d incubation at room temperature. Three or four seedlings were placed in pots filled with quartz sand (0.8–1.2 mm). Half of the plants were inoculated with Rhizophagus irregularis (formerly Glomus intraradices) pure sand inoculum at a 10% (v/v) concentration. Plants were watered three times per week with half-strength Hoagland solution (Hoagland & Arnon, 1940) according to the different treatments (+P, 1 mM phosphate (KH2PO4); −P, 20 μM phosphate; +S, 1 mM sulfate (MgSO4); −S, no sulfate (substitution with 1 mM MgCl2)). All plants were grown in growth rooms under a 16-h light (photosynthetically active radiation 300 μmol photons m−2 s−1), 8-h dark (22°C), 60% relative humidity regime. In three independent experimental set-ups (one in spring and one in autumn 2010, and one in spring 2012), we harvested three pots with three to four plants per pot 5 wk after inoculation. The weight of roots and shoots was measured and leaf and root samples were frozen in liquid N. Part of the roots was used for staining with wheat germ agglutinin (WGA)-Alexa Fluor 488 (Invitrogen, Paisley, UK) to estimate the mycorrhizal parameters according to Trouvelot et al. (1986).

RNA extraction and qRT-PCR

RNA from root and shoot samples was extracted using the InviTrap Spin Plant RNA Mini Kit (Invitek, Berlin, Germany) and quantified in a spectrophotometer. DNase treatment was performed with the Turbo DNA Free Kit (Ambion) and 5 μg of RNA was reverse-transcribed using the Superscript III Reverse Transcriptase (Invitrogen).

qRT-PCR measurements were performed using SYBR green (Fermentas St. Leon-Rot, Germany) in the ABI PRISM 7900 HT sequence detection system (Applied Biosystems, Foster City, CA, USA). The cDNA samples were diluted 1 : 10, and 1 μl of diluted cDNA was used as a template in 10-μl PCR reactions containing 2.5 μM of each primer (for primer sequences, see Supporting Information Table S1). qRT-PCR was carried out in optical 384-well plates with the following PCR regime: 50°C for 2 min, 95°C for 10 min, 40 cycles of 95°C for 15 s and 60°C for 1 min. Melting curve analysis was carried out after each amplification to exclude unspecific amplifications from the analysis. Data analysis was performed using the sds2.2.1 software (Applied Biosystems). The threshold was set at 0.2. For normalization of the expression levels of the genes of interest, we used a housekeeping gene index derived from the genes for transcription elongation factor 1α (MtEf1α), glyceraldehyde-3-phosphate-dehydrogenase (MtGapDh) and MtPdf2 (Wulf et al., 2003; Kakar et al., 2008). For normalization, the housekeeping gene index cycle threshold (CT) value was subtracted from the CT value of the gene of interest, yielding a ΔCT value. The relative expression of the genes was calculated using the formula 2ΔCT. Three biological replicates in two technical replicates were measured for each treatment.

35S feeding experiments

Medicago truncatula seedlings were planted in a special pot with an inner compartment containing R. irregularis. The inner compartment contained mesh openings constructed using two meshes, which allowed fungal hyphae to pass through, but not M. truncatula roots. Half of the pots were inoculated with R. irregularis (pure sand inoculum), with all plants being fertilized three times per week with Hoagland solution. To facilitate mycorrhizal colonization, we used 20 μM PO4 for the first 4 wk and increased the concentration to 100 μM PO4 in the last week to enable a good uptake of SO4. After 5 wk of growth in a growth room, the pots were transferred to growth chambers and a 35S-sulfate solution was applied to the inner (fungal) compartment. In a control set-up, we also applied the solution directly to the root compartment. Leaves of the plants were harvested after 24 and 48 h and subjected to extraction with 0.1 M HCl, and the radioactivity was measured in a scintillation counter.

Leaf metabolite measurements

For ion measurements, 50 mg of frozen ground plant material was subjected to extraction in 500 μl of 0.1 mM HCl. Samples were centrifuged at 14 000 g, and the supernatant was filtered using an Ultrafree MC 5000 MC NMWL Filter Unit (Millipore, Schwalbach, Germany). After filtration, samples were diluted 20 times with deionized water and analyzed by high-performance anion-exchange chromatography with conductivity detection using a Dionex ICS-2000 system (Dionex, Idstein, Germany). Ions were eluted in a KOH gradient.

Quantitative analysis of thiols was performed by a combination of monobromobimane fluorescent labeling and HPLC (Newton et al., 1981). Fifty milligrams of frozen ground plant material was subjected to extraction in 500 μl of 0.1 M HCl with 10 mg of polyvinylpolypyrrolidone. Thiols were first reduced by mixing 120 μl of extract with 200 μl of 0.25 M 2-(cyclohexylamino)ethanesulfonic acid (CHES)-NaOH (pH 9.4) and 70 μl of 10 mM DTT. The reduced thiols were derivatized using 10 μl of 25 mM monobromobimane in acetonitrile. The labeling reaction was terminated by the addition of 220 μl of 100 mM methanesulfonic acid and the resulting solution was then subjected to HPLC analysis.

O-Acetylsulfate (OAS) and amino acids were extracted following a modified protocol from Krueger et al., (2009). Fifty milligrams of frozen ground plant material was subjected to extraction in 200 μl of 80% (v/v) and 200 μl of 50% (v/v) aqueous ethanol (buffered with 2.5 mM HEPES/KOH, pH 6.2) and 100 μl of 80% (v/v) aqueous ethanol. Ethanol/water extracts were subjected to HPLC analysis using a Hyperclone C18 ODS (octadecylsilyl) column (Phenomenex, Aschaffenburg, Germany). OAS and amino acids were measured by pre-column derivatization with O-phtalaldehyde in combination with fluorescence detection (Lindroth & Mopper, 1979; Kim et al., 1997).

Statistical analysis

Data were analyzed by analysis of variance (ANOVA), followed by Turkey's HSD test, to determine the significance of differences between treatments.

Hierarchical clustering analysis using Pearson correlation and principal component analysis were carried out using the MultiExperiment Viewer (MeV) (Saeed et al., 2003).


At high phosphate nutrition, mycorrhizal Medicago truncatula plants show fewer symptoms of S starvation

To obtain initial insights into the S starvation phenotype of Medicago truncatula, we cultivated plants using two different phosphate and sulfate concentrations in the nutrient solutions, which were applied to plants grown in sterile, washed quartz sand. We created severe phosphate starvation (−P) by adding 20 μM KH2PO4 as the only P source and high-phosphate treatments (+P) by using 1 mM KH2PO4. Previous experiments showed that M. truncatula plants grown at 1 mM phosphate are efficiently colonized by R. irregularis (Branscheid et al., 2010). S-depleted plants (−S) were fertilized without S, and S repletion (+S) was achieved by fertilization with 1 mM MgSO4. Three independent experiments were carried out in spring and fall 2010 and in spring 2012. As expected, S-starved plants developed symptoms of S starvation as described in other plants, such as stunted growth and yellowing of the leaves (Fig. 1a; Nikiforova et al., 2004; Lunde et al., 2008). However, the phosphate starvation phenotype was much more severe and resulted in a drastic growth depression (Fig. 1b). Under +P, S starvation resulted in decreased shoot fresh weights (Fig. 1b). Remarkably, the severely phosphate-starved (−P) plants did not show additional effects of S deprivation, indicating a greater importance of phosphate starvation for general plant metabolism.

Figure 1.

Sulfur (S) starvation phenotype in Medicago truncatula. (a) The visible S starvation phenotype in M. truncatula. Medicago truncatula plants were grown in 20-cm pots filled with quartz sand for 5 wk under S starvation or full nutrition conditions and fertilized with half-strength Hoagland solution containing 1 mM phosphate (+P). Full nutrition: the plant received full nutrition including 1 mM SO4. −S: symptoms of S starvation in M. truncatula. The plant was starved of S (0 mM SO4). S myc: the S starvation phenotype was reduced by mycorrhizal symbiosis. The plant was inoculated with Glomus intraradices and shows a healthier phenotype than the –S plant, which remained uninoculated and shows severe S starvation symptoms. (b) Shoot fresh weights of M. truncatula plants grown under different nutrition regimes. Medicago truncatula plants, inoculated (myc) or not (nm) with G. intraradices, were grown in quartz sand for 5 wk and fertilized three times per week with half-strength Hoagland solution containing two different phosphate (+P, 1 mM phosphate; −P, 20 μM phosphate) and sulfate (+S, 1 mM sulfate; −, no sulfate) concentrations. The results are average data for 36 plants under each condition grown at three different times (spring 2010, autumn 2010 and spring 2012) ± SD. Different letters within one graph represent statistically significant differences (< 0.05). (c) Relative expression of the M. truncatula phosphate transporter (MtPt4). The expression level of MtPt4 as a marker for mycorrhizal colonization was measured by qRT-PCR and normalized against MtPdf2. Data shown are average data for 36 plants for each condition grown at three different times (spring 2010, autumn 2010 and spring 2012) ± SD. Different letters within one graph represent statistically significant differences (P < 0.05).

To investigate the effect of an AM symbiosis on phosphate and S starvation, half of the plants were inoculated with R. irregularis. This led to a decrease in the severity of the S starvation phenotype (Fig. 1a).

At both phosphate concentrations applied, no effect of mycorrhizal colonization on shoot weights of S-starved plants was observed. MtPt4 transcript accumulation was significantly reduced in mycorrhizal roots of plants grown at +P compared with plants grown at strong phosphate starvation (−P), but transcript levels were not affected by S concentrations (Fig. 1c). Estimation of the mycorrhizal colonization frequency showed a > 95% frequency of root colonization in all plants grown at −P. The mycorrhizal colonization intensity of the colonized root areas and arbuscule abundance were clearly reduced in all plants grown under +P conditions (Fig. S1). Sulfur starvation showed no consistent influence on the mycorrhizal colonization parameter.

Next, we investigated the effect of S and phosphate starvation on the leaf protein concentration of mycorrhizal and nonmycorrhizal plants. Remarkably, S starvation led to a strongly decreased protein concentration in leaves under +P conditions (Fig. 2). Plants grown at −P/−S did not show decreased protein concentrations because the severe phosphate starvation was negatively affecting plant growth and fresh weights. These data underline the importance of S-containing amino acids for protein synthesis and therefore protein composition of the plant.

Figure 2.

Protein concentration of mycorrhizal and nonmycorrhizal M. truncatula plants under different phosphate and sulfur regimes. Total proteins were extracted from leaves of M. truncatula plants grown for 5 wk with (m) or without (nm) mycorrhizal colonization under −/+ phosphate (P) and +/− sulfur (S) fertilization. The protein contents were measured using a BioRad (München, Germany) protein assay based on the Bradford reagent. Bars represent the mean of four biological replicates ± SD for +P conditions (black bars) and two replicates for −P (gray bars). Different letters represent statistically significant differences (P < 0.05).

We observed a positive effect of mycorrhizal colonization on leaf protein concentration of plants grown at 1 mM phosphate. In summary, protein contents indicate reduced responses to S starvation in mycorrhizal plants if the plants were cultivated at +P. By contrast, no effect of mycorrhizal symbiosis on the S starvation phenotype was observed under strong phosphate starvation.

In vivo confirmation of mycorrhizal sulfur transfer

To confirm the transport of sulfate from R. irregularis to the plant, we grew M. truncatula plants in a two-compartment system. The plants were inoculated with R. irregularis and grown in pots filled with quartz sand. In the middle of each pot we placed a sand-filled compartment with an opening covered by a 150-μm (outside) and 20-μm (inside) nylon mesh, which prevents plant roots from entering the inner compartment. One and 2 d after application of 100 μCurie 35S-sulfate to the inner compartment, the radioactivity in the plant leaves was measured. The presence of 35S in the leaves of mycorrhizal plants (Fig. S2) clearly confirmed the uptake of sulfate by the fungal hyphae and the subsequent transfer of sulfate from the AM fungus to the host plant.

Transcriptional analysis of sulfur stress-related genes

We screened the current release of the M. truncatula genome V.3.5 for putative orthologs of genes involved in the S uptake and assimilation pathway. Similarity searches using sequences of genes involved in the S assimilation pathway of Arabidopsis thaliana led to the identification of highly similar sequences in M. truncatula (Table 1). As we observed a positive effect of mycorrhizal colonization on the S starvation phenotype at 1 mM phosphate fertilization, we investigated mRNA accumulation levels in leaves and roots of mycorrhizal and nonmycorrhizal plants grown under the two different S concentrations (+S and −S) and at 1 mM phosphate (Fig. 3). A marker gene of the S assimilation pathway, adenosine-5′-phosphosulfate (APS) reductase (APR), responded to sulfate starvation with increased expression levels in leaves of S-starved plants. Both APR transcripts were induced in roots of −S plants, with slightly decreased transcript levels in mycorrhizal roots in the same S treatment. Another known sulfate starvation responsive gene encoding a ChaC-like protein showed a clear accumulation of RNA in leaves of S-stressed plants, with a lower induction in roots of S-stressed plants.

Table 1. Putative Medicago truncatula orthologs of Arabidopsis thaliana genes involved in sulfur uptake and assimilation
AnnotationM. truncatula mt3.5 identifierClosest A. thaliana homolog ae-value aSequence identity (%)
  1. a

    A. thaliana AtGDB transcript data base (Blastn).

Adenosine-5′-phosphosulfate (APS) reductase 1Medtr3g009950.1At4g04610.17.00E-2980
APS reductase 2Medtr7g027550.1At4g21990.11.00E-2681
Serine acetyltransferaseMedtr5g095470.1At1g55920.11.00E-1487
ChaC-like family proteinMedtr1g030430.1At1g44790.11.00E-1382
O-acetyl-L-serine (OAS)lyaseMedtr5g006340.1At4g14880.24.00E-9279
Sulfate transporter (SulTr) 1.1Medtr 2g008470At4g08620.11.00E-1984
SulTr 1.2Medtr CU651589 3.1At1g22150.12.00E-0980
SulTr 1.3Medtr 5g061880At1g22150.12.00E-1879
SulTr 2.1bMedtr 3g087740At5g10180.14.00E-1383
SulTr 2.2Medtr 5g061860At5g10180.11.00E-0779
SulTr 3.1Medtr 1g071530At4g02700.12.00E-2784
SulTr 3.2Medtr 6g086170At3g15990.14.00E-1081
SulTr 4.1Medtr 7g022870At3g12520.13.00E-2679
SulTr2.1aMedtr 3g087730At5g62520.20.025
Figure 3.

Relative expression of sulfur (S) stress-related genes. The expression level of selected genes related to sulfur stress were measured by qRT-PCR and normalized against a reference gene index (MtEf1, MtGapDH and MtPdf2) (Wulf et al., 2003; Kakar et al., 2008). Gray bars show the expression in mycorrhizal plants, and white bars that in nonmycorrhizal plants. All plants were grown under phosphate fertilization (+P) conditions (0.5× Hoagland solution containing 1 mM phosphate, and 1 or 0 mM S) for 5 wk. Bars show the mean of three biological and two technical replicates ± SD. Different letters within a graph represent statistically significant differences (P < 0.05).

In parallel with genes involved in S assimilation, we investigated the expression of genes with similarities to plant sulfate transporters (Fig. 4). Similarity searches led to the identification of nine sequences in the M. truncatula genome with significant similarity at the nucleotide level to annotated sulfate transporter genes of A. thaliana (e-value threshold: 0.01), which belonged to the four main subclasses of plant sulfate transporters (Fig. S3). Transcripts of one of these genes (medtr3g087730), which belongs to group 2, were not detectable in roots or shoots of plants grown under our experimental conditions. Most of the putative sulfate transporter transcripts analyzed were induced in leaves and roots of S-starved plants. In contrast to A. thaliana, which possesses five members of group 3 sulfate transporters, we could only identify two in M. truncatula. One of them (medtr6g086170) shows close homology to Sst1 from Lotus japonicus, which was shown to be involved in the S supply to rhizobia in nodules, with the N-fixing capacity being clearly reduced in the corresponding mutants. Using our approach, we could detect only very low transcript levels with a slight induction under S starvation conditions in roots. The group 4 sulfate transporters are responsible for the export of sulfate from the vacuole to the cytoplasm. We could identify one member of this group (medtr7g022870), which shows up-regulation under starvation conditions in roots and shoots, indicating a similar function in M. truncatula. It is worth mentioning that we observed a down-regulation of four sulfate transporter genes (medtr2g008470, medtr5g061860, medtr1g071530 and medtr7g022870) in mycorrhizal roots under S starvation conditions compared with nonmycorrhizal roots under S starvation, but none of the analyzed putative transporter genes were significantly and specifically induced in mycorrhizal roots compared with nonmycorrhizal roots of plants fertilized with 1 mM phosphate.

Figure 4.

Relative expression of putative sulfur (S) transporter genes of Medicago truncatula. The expression level of putative sulfate transporter genes were measured by qRT-PCR and normalized against a reference gene index (MtEf1, MtGapDH and MtPdf2). Gray bars show the expression in mycorrhizal plants, and white bars that in nonmycorrhizal plants. All plants were grown under phosphate fertilization (+P) conditions (0.5× Hoagland solution containing 1 mM Pi, and 1 or 0 mM S) for 5 wk. Bars show the mean of three biological and two technical replicates ± SD. Different letters within a graph represent statistically significant differences (P < 0.05).

Leaf metabolite analysis

We analyzed metabolic changes in leaves to determine the metabolic response to sulfate and phosphate starvation and the influence of mycorrhizal colonization on both starvation responses. Single amino acids, as well as S-containing metabolites and ions, were analyzed in shoots of mycorrhizal and nonmycorrhizal plants grown under S starvation (−S) and S repletion (+S) as in plants fertilized with 1 mM phosphate (+P) and grown under strong phosphate starvation (−P). The content of selected metabolites and ions is shown in Fig. 5. Sulfate as well as cysteine and glutathione contents were highly correlated to each other and responded strongly to S fertilization.

Figure 5.

Metabolite concentrations in leaves of mycorrhizal (myc) and nonmycorrhizal (nm) Medicago truncatula plants grown under different phosphate (P) and sulfur (S) regimes. Leaves of M. truncatula plants grown for 5 wk with or without mycorrhizal colonization under +/− P and +/− S fertilization were treated with HCl to release soluble compounds. The S-containing compounds sulfate, cysteine, glutathione and O-acetylserine (OAS) as well as the phosphate content were measured by HPLC. Bars represent the mean of four biological replicates ± SD for +P conditions and two replicates for −P conditions. Different letters within a graph represent statistically significant differences (P < 0.05).

A principal component analysis (PCA) of metabolite and fresh weight data was carried out to determine the influence of S and phosphate nutrition and mycorrhizal colonization on leaf metabolite contents (Fig. 6). Principal component (PC) 1 described > 50% and PC 2 almost 30% of the variance, resulting in a clear separation of +P and −P conditions. Within the group of +P plants, S treatment rather than mycorrhizal colonization separated the data. Mycorrhizal colonization under +P/−S conditions shifted the data set between the extremes. Under −P conditions, mycorrhizal colonization instead of S nutrition produced minor effects on the whole experimental data set. P availability clearly dominated the plant response to nutrient availability rather than S availability; however, under conditions of 1 mM phosphate fertilization and S depletion (+P/−S), mycorrhizal colonization influenced the responses of the plants to +S conditions. The same was true when the data were analyzed as contents per plant (Fig. S4).

Figure 6.

Principal component analysis (PCA) score plot of the metabolite profile demonstrating changes in metabolite contents per unit fresh weight in Medicago truncatula under different nutritional conditions with (m) or without (nm) mycorrhizal colonization. The plots represent the contents of 27 metabolites and total soluble protein per unit fresh weight. PC, principal component.

Hierarchical clustering (Fig. 7, Supplemental Fig. S5) also clearly separated +P and –P conditions, indicating the dominant effect of this nutrient on plant metabolism in relation to S nutrition. When 1 mM phosphate was applied (+P), clustering differentiated between the presence and absence of S, while under P-deprived conditions (−P) the presence of mycorrhizal colonization overrode the availability of S. However, under −P conditions all metabolites were greatly reduced compared with +P conditions. The plant phosphate content was hardly influenced by the availability of sulfate, while phosphate starvation severely reduced the plant sulfate content, even in the presence of sulfate. Mycorrhizal colonization under −P conditions slightly alleviated the phosphate starvation-driven sulfate reduction. Phosphate availability likewise dominated nitrate accumulation, being greatly reduced when phosphate was absent. Also, S starvation reduced the plant nitrate content, with this effect being alleviated when plants were mycorrhizal-colonized at 1 mM phosphate. Root ion concentrations were generally very low, indicating efficient mobilization to the shoot (data not shown).

Figure 7.

Heat map representing metabolite changes per unit fresh weight in Medicago truncatula under different nutritional conditions with (myc) or without (nm) mycorrhizal colonization. Data were normalized to the average value under all conditions. The data with the log2 scale were subjected to hierarchical clustering analysis (HCA) using Pearson correlation to identify clusters of similar metabolic profiles. Cells with a high abundance are presented in red and those with a lower abundance in blue, with the scale above the heat map.

Primary metabolites of the S assimilation pathway like cysteine, methionine, glutathione, gamma-glutamylcysteine, sulfite, and also total protein content were correlated to the sulfate content in shoots (Fig 3), showing reductions under −S conditions which were slightly alleviated by mycorrhiza formation. Most of the data were mainly correlated to the ‘sulfate clade’, that is, sulfate availability in the plant: fresh weight, protein, nitrate, sulfite, thiosulfate, glu, asp, lys, thr, val, ala, phe, and tyr. The concentrations of OAS, a sulfate starvation marker metabolite and precursor of cysteine biosynthesis, were inversely correlated to sulfate, nitrate, and protein concentrations in tissues. The amino acids ile, leu, trp, gly, gln, and ser were, in contrast, correlated to phosphate availability directly, mainly as a consequence of not showing a sulfate availability-dependent reduction in content. The N-containing amino acids arg, asn, and his showed the opposite behavior to the sulfate clade, being greatly reduced when all nutrients were available and being increased under sulfate deprivation, with mycorrhizal colonization alleviating this effect, resembling the accumulation of OAS.


Most vascular plants are able to participate in a symbiotic interaction with AM fungi. The main benefit for both organisms in this relationship is an exchange of nutrients and carbon. In the past, the primary focus has been on mycorrhizal phosphate transfer, mainly because low phosphate concentrations in soils are often the main growth-limiting factor. Evidence for symbiotic uptake pathways for additional nutrients emerged from the recent identification of a number of other nutrient transporter genes, which are also specifically expressed or induced in mycorrhizal roots (Guether et al., 2009; Benedito et al., 2010; Gaude et al., 2012). Although a specific transporter involved in the symbiotic uptake of S compounds has not been identified to date, in vitro and in vivo experiments clearly demonstrated a symbiotic uptake pathway for sulfate (Rhodes & Gerdemann, 1978a,b; Hepper, 1984).

Indicative of sulfate starvation in plants are reduced thiol, protein and chlorophyll contents and, therefore, reduced biomass. There are also diverse pleiotropic effects on metabolism and gene expression (Nikiforova et al., 2006; Watanabe et al., 2010). This typical S starvation phenotype was also observed in Medicago truncatula. However, the response to S starvation was directly dependent on the plant's phosphate status. Under strong phosphate starvation, S fertilization treatments did not influence shoot fresh weights. This was probably related to the greater impact of phosphate nutrition on plant metabolism and the reduced demand for S in phosphate-starved plants (Hawkesford & De Kok, 2006). By contrast, when grown with 1 mM phosphate fertilization, plants developed severe symptoms of S starvation when no S was added. Mycorrhizal colonization had a significant impact on the S starvation phenotype when plants were grown under conditions of 1 mM phosphate fertilization.

The frequency of mycorrhizal colonization of plants grown under conditions of 1 mM phosphate fertilization was clearly reduced compared with severely phosphate-starved plants, which is consistent with the fact that the phosphate status of a plant systemically influences the frequency of mycorrhizal colonization (Breuillin et al., 2010; Balzergue et al., 2011). However, the reduced degree of mycorrhizal colonization was apparently sufficient to provide enough S via a symbiotic uptake pathway to reduce visible and molecular phosphate starvation responses. This may be related to the c. 2-fold lower concentrations of S compared with P in plant shoot dry matter (Marschner, 2005).

Although increased mycorrhizal colonization levels were found in phosphate-starved plants, a similar effect was not observed in response to changing S nutrition, with −S conditions not significantly enhancing the frequency of mycorrhizal root colonization.

The reduced starvation symptoms of mycorrhizal plants grown under S starvation were accompanied by a significant reduction in the transcription level of two APS reductase genes. APS reductase is involved in the reduction of sulfate, the step preceding S assimilation in plants. APS reductase transcription is strongly regulated by sulfate availability, with increased transcript levels during S starvation (Gutierrez-Marcos et al., 1996; Takahashi et al., 1997). The decrease in transcript levels of two APS reductase genes in leaves of mycorrhizal plants compared with nonmycorrhizal plants is presumably a consequence of increased S nutrition mediated by the AM symbiosis. As in A. thaliana and other plant species, sulfate deprivation leads to the induction of sulfate transporters in M. truncatula (Hawkesford & De Kok, 2006).

We identified nine putative sulfate transporter genes in M. truncatula, which correspond well to the known classifications in other species. In A. thaliana, transporters of the first group are transcriptionally regulated by the sulfate availability in the soil and induced by sulfate deficiency in plants (Rouached et al., 2009). The pattern of decreased transcript levels of four putative sulfate transporter genes in roots of mycorrhizal plants compared with roots of nonmycorrhizal plants probably reflects the reduced S starvation in mycorrhizal plants as a result of symbiotic uptake of S. We also found decreased transcript levels of four putative sulfate transporters in leaves of mycorrhizal plants, which presumably also reflects decreased S starvation of mycorrhizal plants. However, it is worth mentioning that we did not find a putative sulfate transporter with expression in mycorrhizal roots under conditions of 1 mM phosphate fertilization. This is consistent with the findings of a recent work analyzing M. truncatula S transporter gene regulation in mycorrhizal roots under strong phosphate starvation (Casieri et al., 2012). The mycorrhizal phosphate uptake pathway is mediated by mycorrhizal-inducible or mycorrhizal-specific phosphate transporters (Bucher, 2007; Javot et al., 2007a) and it is very likely that plants have an analogous, mycorrhizal-specific transporter for symbiotic S uptake. The currently available M. truncatula genome sequence covers c. 94% of expressed genes (Young et al., 2011), and it is possible that a mycorrhizal-specific sulfate transporter might be located in the part of the genome not yet available. Another possible explanation for the missing mycorrhizal sulfate transporter is that this putative protein is not similar to annotated plant sulfate transporters. Moreover, we cannot exclude the possibility that organic S compounds are also transferred via a mycorrhizal uptake pathway to the host plant, which would require different transporter systems.

It is obvious that reduced sulfate availability will eventually result in reduced levels of S-containing compounds. This is also true for M. truncatula. The concentrations of thiols such as glutathione and cysteine, as well as derived metabolites such as methionine and hence proteins in general, are reduced. It has been shown that sulfate starvation affects other pathways in a pleiotropic manner (Nikiforova et al., 2005) and is also linked to N metabolism (Kopriva et al., 2002), affecting in particular amino acids (Nikiforova et al., 2006), which is, as we have shown, also true for M. truncatula.

The dominance of phosphate starvation compared with S starvation was demonstrated by the more marked changes observed in the leaf metabolite concentrations of phosphate-starved plants. Additional S starvation caused only moderate changes in leaf metabolites, which presumably reflects reduced demand for S in phosphate-starved plants. Remarkably, in phosphate-starved plants, mycorrhizal colonization led to greater changes in leaf metabolite concentrations compared with additional S starvation. This underlines the increased phosphate nutrition of mycorrhizal plants and hence, as mentioned previously, the primary importance of phosphate as compared with S for plant metabolism.

A different effect of mycorrhizal colonization was observed in the leaf metabolite pattern of plants grown under conditions of 1 mM phosphate fertilization. Here, mycorrhizal colonization of S-starved plants led to a shift in the metabolite pattern toward S-replete plants, demonstrating the improved S nutrition of mycorrhizal plants and the importance of mycorrhizal S uptake for plant metabolism, when the plant's phosphate status is high enough for the plant to benefit from increased S nutrition.


This work was supported by the Max Planck Society.