Distinct gene expression and secondary metabolite profiles in suppressor of prosystemin-mediated responses2 (spr2) tomato mutants having impaired mycorrhizal colonization

Plant growth and AMF inoculation

Seeds of wild-type (WT) tomato (Solanum lycopersicum L. cv. Castlemart) and spr2 mutant plants were surface-sterilized by soaking in a 70% ethanol solution for 60 s, with a 20% household bleach solution (5% w/v sodium hypochlorite) for 5 min, and then rinsed three times with sterile water. All seeds were germinated in a sterile soil mixture constituted by equal parts of sand and loam which was autoclaved six times. One-week-old seedlings were removed and transplanted to 1.3 L pots (one plant per pot, ten plants per genotype) containing the same sterilized soil mixture. At the time of transplanting, ten plants per genotype were inoculated with 3 g of a soil-based (1:1 sand-loam) inoculum containing ca. 100 AMF spores per g. In the first (E1) of a series of three experiments, performed between April and May 2016, the plant’s inoculum was Rhizophagus irregularis (Biofertilizante, INIFAP, México), whereas in two subsequent ones (E2 and E3, performed between early April-late May and late May-early July 2019, respectively), a consortium of six AMF species, that is, Glomus fasciculatus, G. constrictum, G. tortuosum, G. geosporum, Gigaspora margarita, and Acaulospora scrobicurata (MycoRacine_VA, MycoBiosfera, México), was used. Similar number of control plants were supplied with 3 g of sterilized soil mixture only. The inoculated and control plants were watered 3 times per week and fertilized once a week with a Long Asthon solution in which the P content was reduced to 7 µM (75% lower than that in the full strength solution), until harvest. They were kept in a growth room with a 16 h/ 8 h light/ dark photoperiod at 27 °C (light) and 23 °C (dark) with an illumination of approximately 250 µmol m⁻² s⁻¹. Roots were harvested at different time points: at 50 (E1), 45 (E2) or 32 (E3) days post-inoculation (dpi). The root system was split lengthways at harvest: one half was stained to evaluate mycorrhizal colonization, whereas the remaining root and leaf tissues were frozen, ground in liquid nitrogen, stored at −80 °C until required for further analysis.

Estimation of AMF root colonization

To evaluate mycorrhizal colonization, root fragments of control and mycorrhized plants (120 per genotype) were stained with trypan blue (Phillips & Hayman, 1970) and observed with a light microscope. AMF colonization was determined in the three independent experiments using two methods: E1 was analyzed as described by Trouvelot, Kough & Gianinazzi-Pearson (1986) using MYCOCALC software (www2.dijon.inra.fr/mychintec/Mycocalc-prg/download.html), while E2 and E3 were analyzed using the magnified intersections method described by McGonigle et al. (1990). Three colonization parameters were analyzed in E1: frequency of mycorrhization (F%), representing the percentage of root segments showing internal colonization; intensity of mycorrhization (M%), the average percentage of colonization of root segments, and arbuscule abundance (A%), the percentage of arbuscules in the whole root system. In E2 and E3, AMF colonization levels were determined by quantifying the proportion of root length segments containing arbuscules, vesicles and/or hyphae, each representing the arbuscular (A), vesicular (V) and hyphal colonization (H) levels, respectively.

Additionally, physiological parameters such as plant height and changes in chlorophyll content were measured during the duration of the E2 experiment. In addition, maximal photochemical efficiency (Fv/Fm) and the photosynthetic potential Index (PI_abs) were measured as indirect indicators of the quantum efficiency of photosystem II. Chlorophyll content was measured using a CCM-200 plus Chlorophyll Content Meter (Opti-Sciences Inc., Hudson, NH, USA). Fv/Fm and PI_abs were determined from data recorded with a portable Pocket PEA chlorophyll fluorometer (Hansatech Instruments Ltd.; Norfolk, UK). All photosynthesis-related measurements were consistently done at noon.

Determination of jasmonic acid (JA), salicylic acid (SA) and SA conjugates

Jasmonic acid levels were determined by GC–MS using the methodology described by Muller & Brodschelm (1994). Free SA and SA conjugates were also analyzed by GC–MS, as described by Malamy, Henning & Klessig (1992).

Targeted metabolite profiling and direct fatty acid (FA) analysis by gas chromatography–mass spectrometry

The GC–MS targeted metabolite analysis of leaf and root samples was performed according to Eloh et al. (2016). In situ FA analysis was done as described by Park & Goind (1994), with slight modifications. These consisted of increasing the 10 min-each methanolysis and methylation steps to 60 and 30 min, respectively. Raw GC–MS data are available from Zenodo (https://zenodo.org/), DOI 10.5281/zenodo.3560965.

Sample preparation for untargeted metabolomic analyses

Frozen tomato roots of control and mycorrhizal plants, respectively, were lyophilized and finely ground in a Mixer Mill MM 400 (Retsch GmbH, Haan, Germany) for 12 s at 30 Hz. Subsequently, 25 mg of plant powder was extracted with 1 mL of an aqueous methanol-formic acid solution (75% v/v and 0.15% v/v, respectively). The mixture was sonicated for 15 min in a water bath at maximum frequency and centrifuged at 10,000 × g for 10 min at 4 °C. The supernatants were filtered through a 0.22 µm syringe filter prior to analysis by mass spectrometry (see below). All samples were freshly prepared in triplicate.

Metabolic fingerprinting of tomato root and leaf extracts by mass spectrometry

For non-targeted metabolite profiling, tomato root methanolic extracts were analyzed by direct liquid introduction electrospray ionization/mass spectrometry (DLI–ESI–MS) as described before (Montero-Vargas et al., 2013). Measurements were performed using an ion trap mass spectrometer LCQ-Fleet (Thermo Scientific, Waltham, MA, USA) in positive mode at a flow rate of 10 µL min⁻¹. Mass spectra were acquired in continuous mode in a range of 50–1300 m/z. The scan time was set to 300 ms with three microscans repeated ten times. The instrument settings were: 3.9 kV source voltage, 35 V capillary voltage, 330 °C capillary temperature, 80 V tube lens voltage, 15 arbitrary units (AU) of nitrogen sheath gas and 20 AU of auxiliary gas. Additionally, samples were analyzed to detect specific ions corresponding to α-tomatine, its biosynthetic precursors and catabolic products (Montero-Vargas et al., 2018).

Raw data processing and data analysis

Raw mass spectrometry data were converted to .mzML format with MSConvert (Chambers et al., 2012). The mass spectrum data were processed in R (http://www.rproject.org) using the package MALDIquant version 1.15 (Gibb & Strimmer, 2012) programed to execute the following tasks: .mzML data import, summary of all scans of each spectrum, smoothing by a Savitzky-Golay filter, peak alignment and peak selection detection. Raw data are available from Zenodo (https://zenodo.org/), DOI: 10.5281/zenodo.3560965. Finally, a comparison matrix (m/z values for columns and file names for rows) with the intensity of peaks were exported in csv format for statistical analysis, applying binning with a bin width of 1 m/z and intensity-based normalization. A total of 367, from more than 1,000 ion signals, were used after initial data cleaning steps.

Unsupervised techniques were employed to explore the effect of mycorrhizal on the metabolic fingerprints of the genotypes employed. These consisted of a hierarchical clustering (HCA) and a principal component analysis (PCA), both implemented in ClustVis (Metsalu & Vilo, 2015). For multiple comparison of the means a Tukey’s Honest Significant Difference (HSD) was applied with a confidence interval of 95%. When the normality of the data was not achieved, the Kruskal–Wallis test was applied, considering p values ≤ 0.05 as significant.

For putative metabolite identification, a homebuilt metabolite database for tomato was created based on previous reports (Moco et al., 2006; Itkin et al., 2011; Caprioli et al., 2015) and the SolCyc database (http://solcyc.solgenomics.net/). Subsequently, automatic matching of the m/z list was performed employing the software SpiderMass (Winkler, 2015).

Extraction of total RNA, cDNA preparation and qPCR analysis

Total RNA was extracted from 100 to 500 mg of frozen root tissues with the Trizol reagent (Invitrogen, Carlsbad, CA, USA), according to the manufacturer’s instructions, with modifications. These consisted of the addition of a salt solution (sodium citrate 0.8 M + 1.2 M NaCl) during precipitation in a 1: 1 v/v ratio with isopropanol and further purification with LiCl (8 M) for 1 h at 4 °C. All RNA samples were analyzed by formaldehyde agarose gel electrophoresis and visual inspection of the ribosomal RNA bands upon ethidium bromide staining. Total RNA samples (4 μg) were reverse-transcribed to generate the first-strand cDNA using an oligo dT20 primer and 200 units of SuperScript II reverse transcriptase (Invitrogen, Carlsbad, CA, USA). The cDNA employed for the qRT-PCR assays was initially prepared from 4 μg total RNA. It was then diluted 5-fold in sterile deionized-distilled (dd) water prior to qRT-PCR. Amplifications were performed using SYBR Green detection chemistry and run in triplicate in 96-well reaction plates with the CFX96 Touch Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA). Reactions were prepared in a total volume of 20 μl containing: 2 μl of template, 2 μl of each amplification primer (2 μM), 8 μl of iQ SYBR Green supermix (Bio-Rad, Hercules, CA, USA) and 6 μl of sterile dd water. Quantitative real-time PCR was performed in triplicate for each sample using the primers listed in the Table S1. Primers were designed for each gene, based on cDNA sequences derived from the tomato genome (Sol Genomics Network; Mueller et al., 2005). Primer design was performed using DNA calculator software (Sigma-Aldrich, St. Louis, MO, USA). The following protocol was followed for all qRT-PCR runs: 15 min at 95 °C to activate the JumpStart Taq Polymerase (Sigma-Aldrich, St. Louis, MO, USA), followed by 40 cycles of denaturation at 95 °C for 15 s and annealing at 60 °C for 1 min. Slow amplifications requiring an excess of 32 cycles were not considered for analysis. The specificity of the amplicons was verified by melting curve analysis after 40 cycles. Baseline and threshold cycles (Ct) were automatically determined using CFX Manager Software version 2.1. PCR efficiencies for all genes tested were greater than 95%. The effect of genotype (WT vs. spr2) and treatment (mycorrhizal vs. non-mycorrhizal) on the expression of a battery of selected genes was calculated using the 2^−ΔCt method (Livak & Schmittgen, 2001). Transcript abundance data were normalized against the average transcript abundance of two reference genes: TIP41 and SAND (Expósito-Rodríguez et al., 2008). Values reported are those of 3 independent mycorrhization experiments each one of which was analyzed using mRNA extracted from a single pooled sample prepared by combining the roots of all 10 plants used per experiment. Gene expression data represent the mean ± SE of three technical replicates per combined root pool per independent experiment. qPCR raw data are available from Zenodo (https://zenodo.org/), DOI 10.5281/zenodo.3560410.

Statistical analysis of mycorrhizal colonization experiments

Data were analyzed by one-way ANOVA to determine whether or not the means of the different treatments tested were equal. A multiple comparison procedure with the Tukey’s test was performed to find significant differences between means. All tests were conducted using the Minitab 15 statistical software package (Minitab Inc., State College, PA, USA). Differences at p < 0.05 were considered as statistically significant.

Mycorrhizal colonization is compromised in spr2 mutants

Three independent experiments were performed. AMF colonization parameters, irrespective of the AMF used as inoculum and method employed to evaluate colonization, were significantly higher in roots of WT plants compared to spr2 mutants. The AMF colonization parameters of two independent experiments sampled at 32 days after inoculation (dpi) (E3; Fig. 1A) and 45 dpi (E2; Fig. 1B), were determined using the magnified intersections method (McGonigle et al., 1990). They were very similar to those of a previous experiment, sampled at 50 dpi (E1), in which AMF colonization was determined using the MYCOCALC software (Fig. S1). Apart from the gene expression analysis, all subsequent data was obtained from leaf and root tissues derived from the E2 and E3 experiments.

Similar to a previous characterization of the spr2 tomato mutant (Li et al., 2003), no significant differences in the the growth rate and morphology of spr2 and WT plants were observed during the experiments herewith described. However, measurement of physiological parameters during the course of E2, indicated that mycorrhizal colonization promoted plant growth, determined as plant height, in WT plants. No growth promotion was recorded in spr2 plants (Table S2). Irrespective of the genotype, photosynthetic parameters were not altered by mycorrhizal colonization. However, contrary to reported data (Zare-Maivan, Khanpour-Ardestani & Ghanati, 2017), AMF colonization had a contrasting effect on chlorophyll content; the significant increase detected in leaves of mycorrhizal spr2 plants was opposite to the tendency toward lower chlorophyll content observed in equivalent WT leaves (Table S2).

Mycorrhizal-induced changes in jasmonic (JA) and salicylic acid (SA) content in leaf and roots

JA and SA levels were measured in leaves and roots of plants sampled at 45 dpi. At this sampling point, significantly increased JA contents were detected in both leaves and roots of mycorrhizal WT plants. The effect was more pronounced in leaves. No increase in JA content was produced in equivalent organs of mycorrhizal spr2 plants (Figs. 2A and 2B). Mycorrhizal colonization had no effect on SA levels in roots and leaves of WT. SA content in leaves of spr2 plants was lower than WT plants and was not modified by AMF colonization, while a significant increase in SA levels was detected in mycorrhizal spr2 roots (Figs. 2C and 2D). No SA-glycosides were detected.

Gene expression profiles

In WT and spr2 roots, AMF colonization led to several changes in the expression levels of a selected set of genes, which were analyzed at 32, 45 and 50 dpi (Table 1). In general, changes in gene expression were more intense and frequent in roots analyzed at 32 dpi and tended to decrease as the mycorrhizal colonization period extended from 45 to 50 dpi. The gene expression profiles obtained were the following:

Targeted metabolic profile of tomato roots and leaves

The results obtained from a GC–MS analysis designed to detect selected metabolites, predominantly associated with primary metabolism in roots and leaves of control and mycorrhizal tomato plants analyzed at 45 dpi is shown in Tables 2 and 3. Different tendencies in metabolite variation patterns were detected in roots (Table 2). A comparison of the roots of non-mycorrhizal WT and spr2 plants, showed that metabolite abundance was predominantly higher in roots of spr2 plants, since 10 of 14 metabolites affected in the mutant were significantly increased in spr2 roots. Relevant exceptions were glucose, fructose and sucrose. AMF colonization had a positive effect on root metabolite accumulation, considering that in both WT and spr2 mycorrhizal roots, 10 of 12 and 13 significantly impacted metabolites, respectively, were increased in response to AMF colonization. Isocitric was significantly reduced in mycorrhizal roots of both genotypes, while proline ratio was significantly increased in mycorrhizal WT roots but reduced in equivalent spr2 roots. Those that increased their abundance in mycorrhizal roots of both WT and spr2 roots were PO₄³⁻, diethylene glycol, succinic, propanoic, and fumaric organic acids, furanone and L-threonic acid. Differences were minimal between mycorrhizal WT and spr2 roots. Only the abundance of DL-malic acid and L-threonic acid was found to be significantly different between them, both being higher in AMF-colonized spr2 roots.

Conversely, the abundance of 16/31 metabolites was significantly affected in leaves of spr2 mutant plants compared WT plants (Table 3). Nine of these were decreased and 7 were increased. Relevant differences were the reduction of 2-butenoic acid, D-glucuronic acid, 1-cyclohexene-3, 4, 5-1-carboxylic acid, L-threonic acid, PO₄³⁻ and sucrose, while serine, oxoproline and maleic acid increased. Fructose and glucose levels were also higher in spr2 leaves. Compared to non-mycorrhizal controls, AMF colonization modified the accumulation of 10 foliar metabolites in spr2 mutant plants, predominantly in a negative way, since only 3 metabolites (i.e., maleic acid, fructose and D-glucuronic acid) showed significantly increased abundance. A comparison of leaf metabolite abundance between mycorrhizal WT and spr2 plants followed this tendency, since only 5 of the 18 metabolites whose abundance significantly differed in response to AMF colonization in these plants, were found to increase in mycorrhizal spr2 leaves. Apart from fructose, glucose, maleic acid and oxoproline, already mentioned above, glycerol was also among the metabolites that significantly increased their abundance in mycorrhizal spr2 leaves.

Fatty acids (FAs) in roots and leaves

The C16:2 Δ^{7E, 10E}, hexadecatrienoic acid and parinaric FAs were not detected in roots of both genotypes, whereas compared to leaves of spr2 plants (see below), C18:3 levels were only reduced 2.1-fold in spr2 roots, while C18:2 levels were only ca. 1.3 higher (Table 4). In WT roots, AMF colonization increased the content of palmitic and elaidic FAs, while iso-methyl C16:0 (15 Me), eicosanoic (20:0), behenic (C22:0) and lignoceric (C24:0) FAs were significantly reduced. Oleic acid was the only FA whose abundance was increased in response AMF colonization in spr2 roots. A comparison between mycorrhizal WT and spr2 roots showed that FA content was, in general, significantly higher in WT roots, since 10/11 FAs, excepting linoleic acid, were significantly higher in mycorrhizal WT roots.

In leaves of WT plants, AMF colonization positively affected the amount of palmitic, the iso-methyl branched C16:0 (14 Me) FA, C16:1 (Δ^7E), linoleic, parinaric (C18:4 Δ^{9Z, 11E, 13Z,15E}), stearic and eicosanoic (C20:0) FAs, but decreased the content of cis-7, 10, 13 hexadecatrienoic acid (C16:3 Δ^{7E, 10E, 13E}) and linolenic acid (Table 5). FA composition in spr2 plants, most predominantly in leaves, accurately reflected the lost functionality of FATTY ACID DESATURASE7, as their levels of polyunsaturated FAs (i.e., C16:3, linolenic and parinaric acids), compared to WT leaves, were considerably reduced (e.g., an approximately 20-fold reduction in linolenic acid) or remained undetected (e.g., 16:3 and parinaric FAs), whereas the content of C16:2 Δ^{7E, 10E} and linoleic FAswas between 6- and 12-fold higher. However, only the content of stearic (increased) and palmitelaidic (decreased) FAs was significantly modified by AMF colonization in leaves of spr2 plants (Table 5). A comparison between WT and spr2 mycorrhizal leaves indicated that most FAs detected, except C16:2 Δ^{7E, 10E} and linoleic acid, were significantly more abundant in the former.

Metabolic fingerprinting of tomato roots

Direct liquid introduction electrospray mass spectrometry (DLI–ESI–MS) fingerprints were generated to examine the effect of mycorrhizal colonization on the global metabolic profile in roots of WT and spr2 mutant plants. PCA analysis of data from 367 significant metabolites identified in roots sampled at 32 and 45 dpi showed that at 32 dpi, the three PCs, representing 39.2% of the variance permitted the separation of root metabolomes by both treatment and genotype factors (Figs. 3A and 3B). Similar results were obtained with roots sampled from the 45 dpi experiment, where the separation of root metabolomes by treatment and genotype was possible with the three PCs representing 48.8% of the variance (Figs. 3C and 3D). A clear discrimination between genotype (WT, spr2), treatment (± AMF) and duration of treatment (32 vs. 45 dpi) was also revealed by the formation of separate clusters in the heat-map generated after HCA using the 100 most intense ions (Fig. 4).

Mass fingerprints were determined for each individual experiment. The results, shown in Tables S3 and S4, represent the putative metabolite ions whose significant change in abundance was associated with higher/lower AMF colonization levels at 32 and 45 dpi. They reinforced the above findings showing that the effect of AMF colonization on root primary and secondary metabolism was dependent on factors such as genotype and colonization time. They also agreed partially with the concept that AMF colonization efficiency involves changes in the content of several categories of lipids (see above), including lysophospholipids, in addition to phytosteroids (see below), carotenes, phenolic compounds, polyamines, auxins, cytokinins, amino acids and other nitrogen-containing compounds (Akiyama & Hayashi, 2006; Drissner et al., 2007; Floss et al., 2008b; Floss & Walter, 2009; Whiteside, Garcia & Treseder, 2012; Jiménez-Bremont et al., 2014; Rivero et al., 2015; Bedini et al., 2018; Sánchez-Bel et al., 2018; Liao et al., 2018).

A targeted assay focusing on the analysis of the tomatine biosynthetic pathway (Montero-Vargas et al., 2018) was performed. Differences were observed between genotypes and between non-colonized and colonized plants. They were also influenced by the duration of colonization. In general, the content of α-tomatine, its biosynthetic precursors and related catabolites (Fig. 5; Tables 6 and 7), was significantly lower in spr2 mutant roots. This effect was more pronounced in roots sampled at 45 dpi. The influence of mycorrhizal colonization was also time-dependent in WT and spr2 roots. At 32 dpi, significantly affected tomatine-related metabolites and total SGAs had a lower abundance in colonized spr2 roots, while this pattern was reversed at 45 dpi (Fig. 5; Tables 6 and 7). Thus, α-tomatine content, which was significantly reduced in spr2 roots in response to AMF colonization at 32 dpi, became enhanced at 45 dpi. In mycorrhizal WT roots, α-tomatine and most other SGAs tested remained significantly higher than controls only at 32 dpi (Figs. 5A and 5B; Tables 6 and 7).

Discernable biochemical and transcriptional changes occur in spr2 tomato mutants compromised in mycorrhizal colonization in tomato

Tomato spr2 mutant plants consistently showed reduced AMF colonization (Fig. 1; Fig. S1) and consequently were unable to benefit from the growth promotion effect usually produced by the AM symbiosis, as observed in mycorrhizal WT plants (Table S2). Low mycorrhization efficiency in spr2 plants was also in agreement with prior studies reporting deficient mycorrhization in roots of this mutant plant (Tejeda-Sartorius, Martínez de la Vega & Délano-Frier, 2008; Song et al., 2015). They differed, however, from data obtained from the JA-deficient defenseless-1 tomato plant mutants (Howe et al., 1996) which was found to have increased AMF colonization than WT and prosystemin overexpressing transgenic plants (Formenti & Rasmann, 2019). Lower colonization efficiency in spr2 roots could have partly resulted from their incapacity to significantly increase JA content above control levels in response to AMF colonization (Fig. 2). Since spr2 mutants are impaired in the orchestration of the JA burst required for the activation of wound- and systemin-related defense responses (Li et al., 2003), it may be argued that AMF promote along-term maintenance of constitutively higher JA levels as a strategy to ensure efficient root colonization. This is in accordance with different scales of AMF-associated JA accumulation previously reported in various plant models, including tomato (Hause et al., 2007; Hause & Schaarschmidt, 2009; López-Ráez et al., 2010; Sánchez-Bel et al., 2016). A plausible use of this strategy could be to orchestrate a JA-related suppression of immunity responses triggered by microbe-associated molecular patterns in order to enhance a positive plant–microorganism interaction, as previously observed in Arabidopsis roots (Jacobs et al., 2011; Lakshmanan et al., 2012). Incidentally, increased JA levels were also in agreement with the proposed priming of JA-dependent defenses in mycorrhizal plants (Cameron et al., 2013; Sánchez-Bel et al., 2016), a trait that was shown to be abolished in spr2 mutant plants (Song et al., 2015). These scenarios complement other JA-related phenomena proposed to support mycorrhizal colonization (Hause et al., 2007; Tejeda-Sartorius, Martínez de la Vega & Délano-Frier, 2008).

An additional condition that could have negatively affected AMF colonization in spr2 plants was the significant SA accumulation observed in mycorrhizal spr2 roots (Fig. 2). In contrast to a previous report linking SA accumulation with a disrupted ω−3 FATTY ACID DESATURASE7 function in this mutant (Avila et al., 2012), SA accumulation above WT levels in spr2 roots occurred exclusively in response to AMF colonization. Nevertheless, the accumulation of SA in spr2 roots having deficient AMF colonization was in agreement with copious evidence, in various plant models, linking SA with a negative effect on the mycorrhization process at various stages (Fernández et al., 2014; Liao et al., 2018). This outcome is considered to be a consequence of the primary role played by SA in plant defense against biotrophic microorganisms (Pieterse et al., 2009; Gutjahr & Paszkowski, 2009).

Arbuscular mycorrhizal fungi colonization coincided with the strong induction of the LePT4 mycorrhizal marker (Table 1). Interestingly, the expression levels of this mycorrhizal colonization marker was found to correlate with AMF colonization efficiency only in experiment E1 (Fig. S1), which was the longest (i.e., 50 dpi) and was performed using a single AMF species. Conversely, in E2 and E3, LePT4 expression levels were inversely correlated with AMF colonization efficiency (Fig. 1; Table 1). The latter combination was in contradiction with findings that established a specific association between this particular phosphate transporter and a functional AMF mycorrhizal symbiosis in Lotus japonicus, Medicago truncatula, wheat, and rice (Xu et al., 2007; Gutjahr & Parniske, 2013; Wang et al., 2017; Zhang et al., 2019). The discrepancy between the results obtained in E1 with those of E2 and E3 suggest that the use of an AMF consortium vs. a single AMF species and, perhaps, the duration of the colonization period could have been contributing factors to the difference observed. Supporting evidence for this possibility is the observation that mycorrhizal phosphate uptake varies among different AMF species. Also, the AMF conditions established in E2 and E3 were more reminiscent of the dynamics of phosphate uptake in the field, which is assumed to be the combined result of the diverse AMF types that contribute separately to phosphate uptake in response to specific environmental conditions, perhaps similar to those encountered in spr2 roots (Kobae, 2019). Other possible explanations for this apparent disparity may be the likelihood that LePT4 could not be essential for the establishment of the mycorrhizal symbiosis in tomato, considering that the expression of other P transporter genes (i.e., LePT3 and LePT5) was also found to be induced by AMF colonization (Nagy et al., 2005). In addition, mycorrhizal-specific LjPT4 and MtPT4 phosphate transporters were also regulated by early root responses to phosphate levels in non-mycorrhizal roots (Volpe et al., 2016). Another possibility is that mycorrhizal-specific P transporters might have supplementary functions, as supported by the defensive role against pathogenic fungi or abiotic stress recently assigned to mycorrhizal-specific P transporters in wheat (Zhang et al., 2019) and tomato (Volpe et al., 2018). On the other hand, the ca. 3-fold reduction in LeTP4 expression levels that occurred as the colonization period extended from 32 to 45 days, and the complete reversal its expression pattern in mycorrhizal WT and spr2 roots at 50 dpi (Table 1), was in accordance with the expression of mycorrhizal-specific P transporters in wheat roots, which were found to be induced at different stages of the AMF symbiosis (Zhang et al., 2019).

Contrasting AMF colonization in WT and spr2 roots coincided with the possibility that mycorrhization efficiency was enhanced by the repression of GAI. Although this proposal disagrees with the proposed regulatory role played by DELLA repressors of GA signaling during the establishment of AM symbiosis (Yu et al., 2014; McGuiness, Reid & Foo, 2019) it is, nevertheless, consistent with other studies evoking a more dynamic role for GAs during the colonization process. Thus, GA levels are believed to fluctuate widely in order to maintain a balance suitable for both for DELLA stability and proper AMF colonization (Foo et al., 2013; Martín-Rodríguez et al., 2015, 2016).

BR-related genes were strongly repressed in mycorrhizal WT roots, while they were generally up-regulated in mycorrhizal spr2 roots. This was also in contradiction with the overall positive influence that BRs had in various plant models, including tomato (Bitterlich et al., 2014a, 2014b; Foo et al., 2016; Tofighi et al., 2017). However, most of this information was obtained using BR-deficient mutants under experimental conditions that may have influenced other plant hormones (e.g., ethylene; Jiroutova, Oklestkova & Strnad, 2018), known to affect mycorrhizal colonization. Another possible drawback was that the effect on AMF colonization was revealed only after BR levels were severely reduced (Foo et al., 2013). This scenario reflects the paucity of information regarding the precise role played by BRs during the AMF colonization process (Bedini et al., 2018; McGuiness, Reid & Foo, 2019).

ABA-related gene expression patterns indicated that the positive influence exerted by the mycorrhizal colonization on ABA biosynthesis and signaling (Herrera-Medina et al., 2007; Martín-Rodríguez et al., 2016), was not affected in the spr2 mutant. On the other hand, ET gene expression patterns concurred with ample evidence proposing that ET generally has an inhibitory effect on the AMF colonization (Foo et al., 2013, 2016; Pozo et al., 2015; Gomez Monteiro Fracetto, Pereira Peres & Rodriguez Lambais, 2017). Thus, the majority of the ET biosynthetic and signaling genes analyzed were induced in mycorrhizal spr2 roots, predominantly at 32 dpi.

Similar to ABA, the positive participation assigned to the 9-LOX oxylipin pathway in the AMF colonization process (López-Ráez et al., 2010; León-Morcillo et al., 2016) was not affected in the spr2 mutants. A comparable argument could be used regarding the 13-LOX pathway, where key regulatory genes of the metabolism and regulation of jasmonates such as LOXD, AOS1, JMT and JAZ2, previously found to be induced in mycorrhizal tomato plants (López-Ráez et al., 2010), had similar expression patterns in both WT and spr2 mycorrhizal roots. On the other hand, some WR gene expression profiles were contrary to expectancy. Extensively repressed AROGP expression in mycorrhizal WT roots was incompatible with the conceived need to partially degrade complex carbohydrates in the extracellular matrix to allow fungal spread and periarbuscular matrix formation (Liu et al., 2003). Also, the induction of the LHA1 gene in spr2 roots coupled to its repression in equivalent WT roots, at 32 dpi, appeared to be contrary to the presumed role of H⁺-ATPases as activators of secondary transport systems at the symbiotic interfaces in tomato (Rosewarne et al., 2007; Liu et al., 2016) and as mediators phosphate transport and plant growth in M. truncatula (Krajinski et al., 2014). However, this function is apparently performed in tomato by LHA2, a distinct H⁺-ATPase isoform. Thus, LHA2 transcripts were found to accumulate in mycorrhizal tomato roots, contrary to LHA1 whose expression was repressed by AMF colonization (Ferrol et al., 2002). These workers hypothesized that the selective down-regulation of LHA1 could reflect a precise role for this transporter during phosphate uptake in tomato, namely in epidermal cells under non-mycorrhizal conditions. They further argued that this resembled the downregulation of two phosphate transporter genes in mycorrhizal M. truncatula roots.

The broad repression of RBOH1 in mycorrhizal WT roots was contrary to the proposed NADPH oxidase-mediated increase of ROS in M. truncatula required to facilitate root cortex colonization by AMF arbuscules (Belmondo et al., 2016a, 2016b). The induction of PS in spr2 mycorrhizal roots at 32 dpi was counter to reports showing that mycorrhizal roots of PS overexpressing plants yielded significantly higher A% levels than WT roots (Tejeda-Sartorius, Martínez de la Vega & Délano-Frier, 2008) and that exogenous systemin promoted pre-symbiotic and early AMF colonization phases in tomato (De la Noval-Pons et al., 2017a, 2017b). A possible argument to explain these apparent contradictions could be that a suppression of certain JA-dependent WR genes is needed to limit JA-related defense responses in order to allow the proper establishment of the mycorrhizal symbiosis (Gutjahr et al., 2015; Martin et al., 2016). Conversely, the strong repression the SCP WR gene in mycorrhizal WT roots could have positively affected AMF colonization efficiency by altering the abundance and activity of secreted proteases, including SCPs that are known to activate defense responses via proteolytic degradation processes (Kohler et al., 2015). Apoplastic SCPs are also known to control the generation of peptide signals critical for proper fungal development within the root (Rech, Heidt & Requena, 2013). However, the induction of the JA-inducible PINII WR marker gene in both WT and spr2 mycorrhizal roots at 32 dpi was enigmatic.

A ca. 2-fold higher expression level of the CCD7 gene in mycorrhizal spr2 roots, compared to equivalent WT roots was observed. Irrespective of this difference, the behavior observed was consistent with most experimental evidence indicating that constant CCD7 gene activation usually observed during AMF colonization is symptomatic of this enzyme’s involvement, together with CCD1, in the production of AMF-induced C13/C14 apocarotenoids. These include α-inolglucoside, the cyclohexenone blumenol and mycorradicin, considered to be signature regulatory metabolites of the AMF symbiosis at late colonization stages (López-Ráez et al., 2015; Hou et al., 2016; Lanfranco et al., 2018; Fiorilli et al., 2019). The contrasting CCD1b expression levels observed in WT and spr2 mycorrhizal roots further suggests that apocarotenoid synthesis via CCD1b was differentially regulated in this mutant plant. This finding reinforced the consensus that further research is required to define how apocarotenoid flux is regulated in plants and how these distinctive chemicals, originating from the same metabolic pathway, interact with other phytohormones to regulate the mycorrhization process (Fiorilli et al., 2019).

Arbuscular mycorrhizal fungi colonization had no effect on the expression of PAL5, while it led to the repression of the PAL3 gene in mycorrhizal WT roots, contrary to the induction of both PAL3 and PAL4 genes in mycorrhizal spr2 roots at 32 dpi. These results also disagreed with consensual findings indicating the positive role played by PAL enzymes in the AMF colonization process (Morandi, 1996), via their crucial function as mediators of the biosynthesis of secondary metabolites that stimulate AMF root colonization (Mandal, Chakraborty & Dey, 2010; Steinkellner et al., 2007). However, the induced expression of the AMF-colonization responsive FLS gene, required for flavonol synthesis (Scervino et al., 2005), was unaffected by the spr2 mutation. This contrasted with the BEAT gene, whose expression was ca. 10-fold higher in mycorrhizal WT roots sampled at 50 dpi. This gene could have positively impacted the mycorrhizal process via its negative influence on methyl salicylate synthesis (Bera, Mukherjee & Mitra, 2017). Likewise, the ca. 4-fold higher DXS-2 expression levels detected in mycorrhizal WT roots, was in agreement with the positive role played by this gene in the AMF symbiosis via its key involvement in the MVA isoprenoid biosynthetic pathway leading, among others, to the generation of biosynthetic precursors of the above-mentioned C13 and C14 apocarotenoids (Walter, Floss & Strack, 2010; Floss et al., 2008a; Liu et al., 2003; Kuhn, Küster & Requena, 2010).

CAS1 catalyzes the cyclization of 2, 3-oxidosqualene, a precursor of cycloartenol, a key metabolic intermediary in sterol biosynthesis, whereas GAME1 codes for a galactosyltransferase involved in the synthesis of steroidal α-tomatine (Itkin et al., 2011). Both these genes were induced in spr2 mycorrhizal roots at 32 dpi, while CAS1 was widely repressed by AMF colonization in WT roots. A similar expression pattern was produced by the FPS1, a gene required for the synthesis of farnesyl diphosphate, an early biosynthetic precursor of sterols and triterpenoids (Abe, Rohmer & Prestwich, 1993). These results disagreed, however, with the higher accumulation of tomatine, its biosynthetic precursors and catabolic products in mycorrhizal WT roots, particularly at 32 dpi (Fig. 5; Tables 6 and 7). They were also incompatible with data showing that downregulation of GAME1 resulted in an almost 50% reduction in α-tomatine levels in tomato leaves (Itkin et al., 2011). However, GAME1 silencing was observed to lead to the upregulation of various genes known to be involved in pathogen defense. It also caused severe morphological alterations due to changes in membrane sterol levels. This scenario suggests that α-tomatine and related compounds may contribute to maintain AMF colonization in tomato by limiting certain plant defense-responses. They may also act as a stabilizing factor during AMF colonization through their contribution to the maintenance of plant cell viability and, possibly, to the enrichment of membrane micro-domains that modify the fluidity and dynamics of the plasma membrane in order to favor symbiotic plant interactions (Babiychuk et al., 2008; Von Sivers et al., 2019). The above arguments suggest that tomatine and related compounds could be important regulators of AMF symbiosis in tomato. Moreover, tomatine metabolite data obtained in this study coincided with previous reports indicating that tomatine biosynthesis was negatively affected in spr2 mutant plants (Montero-Vargas et al., 2018) and with findings showing that resistance to necrotrophic fungal and oomycete diseases in spr2 mycorrhizal tomato plants, was significantly lower than comparable WT plants due to their reliance on the JA-regulated induced systemic response (Song et al., 2015). It also raised the possibility that impaired α-tomatine accumulation in mycorrhizal spr2 roots could have been partly responsible for their lower AMF colonization efficiency.

Another contributing negative factor could have been the induction of SA-related genes, which occurred almost exclusively at 32 dpi in spr2 mycorrhizal roots (Table 1). Augmented expression of these genes coincided with the significant accumulation of root SA levels and with the induction of WRKY60, coding for a tomato TF, closely related to the Arabidopsis WRKY70. This TF is known to regulate the SA-dependent SAR active against biotrophic fungal pathogens (Bai et al., 2018), and to inhibit JA-ET-responsive defense gene expression (Li et al., 2019). Reduced colonization efficiency also agreed with significantly increased expression levels of the SAMT gene, coding an enzyme catalyzing the formation of MeSA, considered to be an important component of the long-distance circuitry needed to establish plant SAR (Park et al., 2007; Shah & Zeier, 2013). The only discrepancy was the induced expression in mycorrhizal spr2 roots of the SSI1 gene, coding for a stearoyl acyl carrier protein FAD that converts stearic acid to oleic acid, a fatty acid known to inhibit SA signaling (Kachroo et al., 2008). Perhaps its induction reflected the proposed ability of AMF to modulate SA-related defense signaling. In any case, the reduced mycorrhizal efficiency in spr2 roots having induced levels of SA-related genes corroborated the above mentioned notion that SA acts as an inhibitor of the mycorrhizal symbiosis (Blilou, Ocampo & García-Garrido, 1999; Herrera-Medina et al., 2003; Bedini et al., 2018).

Differential fatty acid composition and accumulation patterns in leaves and roots could contribute to the contrasting AMF colonization levels observed in WT and spr2 plants

A significantly lower content of mostly all FAs except linoleic acid, was detected in mycorrhizal spr2 leaves and roots. This key biochemical modification could have contributed to the lower mycorrhization efficiency that characterizes spr2 plants. The increased levels of palmitic acid and a tendency towards higher oleic in mycorrhizal WT roots acid were relevant in the context of the AMF symbiosis. First, augmented palmitic acid in WT mycorrhizal leaves and roots agreed with recent evidence showing that the AM symbiosis increases the lipid flux and redirects it to generate 16:0 β-monoacylglycerols, which are later transferred from the plant to the periarbuscular apoplast (MacLean, Bravo & Harrison, 2017). Additionally, higher oleic acid in mycorrhizal WT roots, may have favored mycorrhizal colonization by inhibiting SA signaling (Kachroo et al., 2004).

Lower FA content in spr2 mycorrhizal leaves and roots suggested that the C supply to the mycorrhizal fungi, recently demonstrated to be composed by both lipids and sugars (Jiang et al., 2017; Luginbuehl & Oldroyd, 2017; Roth & Paszkowski, 2017) was probably an additional factor responsible for lower colonization efficiency in spr2 roots. It remains to be defined if the absence of polyunsaturated 16:3 and 18:3 FAs in spr2 leaves and roots could have also affected the establishment of AMF by altering the fluidity and/or transport capacity of their plant cell membranes, including the periarbuscular membrane. This change could have indirectly affected P transport across membranes, or even spr2 plant fitness, considering that low polyunsaturated FA composition in fad2 Arabidopsis mutants was shown to reduce the mobility of membrane lipids and to concomitantly impair the Na⁺/H⁺ pump function and the proton translocating activity of ATPases (Zhang et al., 2012). Also relevant was the relation to data generated by an untargeted metabolomic analysis in tomato that revealed that α-linolenic acid derivatives were positively affected by the mycorrhizal symbiosis (Rivero et al., 2015). Finally, FA data were in accordance with a report showing that, in addition to flavonoids, linolenic acid, and linoleic acid were key factors responsible of regulating AM colonization in litchi trees (Shu et al., 2016).

Differential metabolite accumulation in roots coincided with differential AMF colonization in WT and spr2 plants

A supervised principal component analysis of signals derived from an untargeted metabolomic analysis of root extracts emphasized the variability observed between mycorrhizal roots of WT and spr2 plants (Figs. 3 and 4; Tables S3 and S4). Additionally, a GC–MS targeted metabolic analysis revealed that AMF colonization had a much more pronounced effect on metabolite abundance in leaves than in roots of plants sampled at 45 dpi. The relevant increase in sucrose content, coupled to lowered glucose and fructose levels, in WT mycorrhizal leaves, was in agreement with their increased need to redirect sucrose from leaves to roots to support their higher colonization rates compared to spr2 plants (Roth & Paszkowski, 2017). A similar argument could explain the significantly higher level of PO₄³⁻ in leaves of mycorrhizal WT leaves. On the other hand, significantly higher content of DL-malic acid and L-threonic acid in spr2 roots deficient in mycorrhizal colonization could have mirrored a metabolic condition not amenable to AMF colonization (Bedini et al., 2018). With respect to the SA-dependent implementation of the AMF-inhibiting SAR described above, it is germane to add that SA is also believed to involve a repression of fermentation, higher cytosolic oxidative potential and other conditions known to favor AMF colonization (Bedini et al., 2018).