Lotus tenuis plants used in this study were from a commercial variety, tolerant to Argentinean saline-alkaline soils. L. corniculatus plants were from a wild population, recovered from area known as Devesa de El Saler (Latitude 39°20′41″ N; Longitude 00°19′12″ W, Valencia, Spain). Taxonomical determination of the recovered wild L. corniculatus population was carried out according to morphological traits of vegetative and reproductive organs, as described in Valdés .
Seeds of both populations were scarified and sown according to Escaray et al. . Four days after rootlet emergence, seedlings were transplanted to 500 cm3 pots containing Mollisols type soil (USDA classification) and cultivated in a pollinator-free greenhouse under the following regime: 14 to 15 hours of photoperiod with a photosynthetic flux density (PPFD) from 600 to 900 μmol m-2 s-1 and average temperature of 24°C (Argentinean conditions).
Flowers of L. tenuis plants to be cross-pollinated were emasculated and crossed with pollen recovered from L. corniculatus using standard techniques. A reciprocal cross using L. tenuis as pollen donor and emasculated L. corniculatus plants as maternal parent was also performed. Seeds harvested from manually cross-pollinated L. tenuis or L. corniculatus plants were sown as reported above. Resulting seedlings were transplanted into 300 cm3 pots containing perlite/sand (1/1 v/v) and cultivated in a growth chamber under the following regime: 16 hours of white light (PPFD of 450 μmol m-2 s-1) at 24°C followed by 8 hours in the dark at 20°C and relative humidity ranging from 55 to 65%. All plants were irrigated with Hoagland 0.5 × nutritional solution for 30 days. From an F1 population consisting of 50 L. tenuis x L. corniculatus plants, four plants representative of one cross (named LH1, LH2, LH3 and LH4) were selected for the further analyses. To this end, each individual hybrid and parents were propagated from cuttings . Kariological analysis was carried out according to Escaray et al. .
Finally, plants of the F1 hybrid population were randomly intercrossed by manually pollinating the emasculated flowers. An F2 population comprising 200 plants was constituted by sowing seeds recovered and cultivated in a pollinator-free greenhouse, under the conditions described above. Leaves from these plants were harvested for PA staining.
Morphological and nutritional evaluation of hybrid plants
Clones of hybrid plants were grown simultaneously with clones of parental plants in 2000 cm3 pots containing Natracuol typical soil (USDA classification) in a greenhouse under the Argentinian condition described above. Total biomass production was measured over four harvests by growing plants for four months in spring and summer (2010) following a completely randomized experimental design, with ten repetitions per plant. Each month the shoots were cut 3 cm above soil level; the collected material was dried at 37°C and total dry weight determined. Just before the first and second cutting, plants underwent morphological determinations. Stem length and the number of stems per plant were measured and leaf morphology analyzed. Total leaf area, leaf length/width ratio and the number of trichomes/cm2 per leaf were determined for the four fully expanded leaves on the main stem. Leaves were scanned and analyzed by Image-Pro Plus 4.5 (Media Cybernetic, Sylver Spring, USA) to measure their area. Trichome number on the adaxial side of leaves was counted under an Optical Microscope (Nikon SMZ800, Nikon, Tokyo) at 10× magnification.
The evaluation of total protein (TP) content and in vitro dry matter digestibility (IVDMD) were carried out using dried shoots from L. corniculatus, L. tenuis and four hybrid plants. Determination of TP was made following the Kjeldahl method . The IVDMD analysis was performed according to Vogel et al.  on 0.5 g of dried material using a nylon bag with standard porosity incubated in a Daisy II (Ankom) incubator for 48 hours in the presence of the rumen liquor obtained from fistulated cows fed with alfalfa.
Anthocyanin and chlorophyll determination
To quantify pigment concentrations, clones of the four selected hybrids and of both parents were cultivated for 35 days in a growth chamber, under a completely randomized experimental design with six repetitions per genotype, as reported above.
Anthocyanins were extracted from 100 mg of frozen and ground stem matter with 3 ml of 0.1% HCl/methanol solution for 60 min at room temperature. Then, 0.75 ml of water and 2 ml of chloroform were added to 1 ml of extract. Finally, anthocyanins were quantified by reading the absorbance of the upper phase at 536 nm, and their concentration calculated on the basis of the absorbance of cyanidin-3-0-glucoside. Chlorophyll determination was carried out according to Lichtenthaler .
Parental and selected F1 plants were cultivated for 35 days in growth chambers, as reported above, and subsequently plant matter was collected and dried at 32°C for 15 days. A completely randomized experimental design with six repetitions per genotype was performed. Proanthocyanidins were quantified as described by Escaray et al.  following the DMACA-HCl protocol .
Levels of PAs were evaluated in the F2 population in the second fully developed leaves by DMACA staining . According to the number of PA-containing cells, related to the intensity of DMACA staining, all 200 plants were sorted into 5 classes. These classes ranged from Class 1, grouping plants with the lowest number of PA-containing cells, to Class 5, clustering plants whose leaf blade was fully covered with PA-containing cells. Then, six F2 plants, three belonging to Class 5 (27, 120, 186) and three to Class 1 (3, 5, 147) were propagated by cuttings and grown in parallel with the original parental lines for 30 days under outdoor conditions in summer 2012. Conditions employed were: 13.5 to 14 hours of photoperiod with a PPFD from 1000 to 1350 μmol m-2 s-1 and average temperature of 22.5°C (Italian conditions). Proanthocyanidins were quantified and expression levels of the PA structural and regulatory genes were analyzed using a completely randomized experimental design, with three repetitions per genotype.
Qualitative analyses of PAs from parental and F1, as well as from reference species L. uliginosus, were performed by Thin Layer Chromatography (TLC) using cellulose plates according to Nybom . Commercial Cyanidin and Delphynidin (Sigma-Aldrich, St. Louis, MO, USA) and leaf PAs from L. uliginosus were used as standards.
DNA and RNA isolation, cDNA synthesis and ITS amplification of DNA and cDNA
DNA was isolated from leaves according to Doyle and Doyle . RNA was isolated from leaves and stems using the Total RNA Plant Isolation kit (Sigma) according to the supplier’s instructions, after which a further DNase treatment was added. The quality and quantity of RNA were verified by agarose gel electrophoresis and spectrophotometric analysis. The absence of DNA from the RNA samples was tested by the null PCR amplification of the universal rDNA primer pair ITS1/ITS4, as described in Paolocci et al. . The same primer pair was also used to prove the hybrid status of F1 plants by amplifying the genomic DNA from parental and hybrid lines and sequencing the resulting amplicons. Then cDNA from parental, hybrid F1 and F2 plants was synthesized from 3 μg of total RNA, using SuperScript III H-Reverse Transcriptase (Invitrogen) and 100 pmol of random hexamers (Pharmacia Biotech) according to supplier’s instructions.
Primer design and cloning of housekeeping and PA genes
The cDNA sequences related to the alpha elongation factor (EF-1α), to the structural genes of the flavonoid pathway (PAL, CHS, DFR, ANS, ANR, LAR1 and LAR2) and to regulatory MYB and bHLH genes from Lotus spp. and other model species were downloaded from GenBank. Sequence alignment for each target gene was carried out using the BioEdit software . Then a primer pair specific to each target gene was designed, with the help of OligoExpress Software (Applera Biosystems), on the most conserved nucleotidic residues between species to amplify cDNA isolated from leaves of L. tenuis and L. corniculatus parental lines, obtained as reported above. The primer pairs are given in Additional file 1: Table S7. The PCR conditions were those reported in Paolocci et al. . The resulting amplicons were sequenced, analyzed and incorporated to GenBank (Additional file 1: Table S5). Later were aligned to design gene-specific primer pairs on highly conserved residues between the two parental lines for qRT-PCR analysis (Additional file 1: Table S8).
Quantitative RT-PCR analysis
The primer pairs, designed as reported above, were initially checked for their specificity and amplification efficiency in qRT-PCR experiments on leaf cDNA from both parents. To test for specificity a dissociation analysis was performed after each amplification run, followed by amplicon sequencing. Meanwhile, PCR efficiency was assessed by performing a standard curve for each gene using six dilution points, each one replicated four times. Unless specifically stated, only primer pairs that produced the expected amplicon and showed similar PCR efficiency on both parents were used.
To study gene expression, the parents and the selected F1 hybrid plants were grown under growth chamber conditions for 30 days as described above. A completely randomized experimental design with four biological replicates for each genotype was used. For each RNA sample, two RT steps were performed and then pooled. An aliquot of 5 μL of 1:10 diluted pools of cDNA was used in the PCR reaction, which was made up using the Power Sybr-Green PCR core mix (Applied Biosystems) according to the supplier’s instructions in 20 μL of final volume in the presence of 2.5 pmol of each primer. Four biological replicates were performed per sample and gene. Cycling parameters were two initial steps of 50°C for 2 min and 95°C for 2 min, a two-step cycle of 95°C for 15 s and 60°C for 1 min repeated 50 times, a final step of 10 min at 60°C. Afterward, the dissociation protocol was performed. Amplifications were performed on ABI PRISM 5700 SDS apparatus (Applied Biosystems). For each transcript, the average threshold cycle (Ct) was determined. The gene quantification method based on the relative expression of the target gene versus the reference genes EF-1α, was adopted according to Paolocci et al. .
To study gene expression on the F2 population, selected plants were analyzed in parallel with L. corniculatus and L. tenuis parental lines. To this purpose, a completely randomized experimental design with three biological replicates per genotype was used and grown under Italian conditions described above. The qRT-PCR analyses were performed following the procedure reported above.
Double-strand sequence analysis was carried out, either directly on PCR products or on PCR fragments previously cloned in the pGEM-T Easy Vector System I (Promega), using the Big Dye Terminator Cycle Sequencing Kit and an ABI Prism 310 Sequence Analyzer (Applied Biosystems) according to the supplier’s instructions. The sequencing primers were the Sp6 and T7 vector primers for cloned PCR fragments and gene-specific primers for direct sequencing reactions.
Sequence data from this article can be found in the GenBank NCBI data libraries under accession numbers [GenBank: KF164612] (LtITS), [GenBank: KF164611] (LcITS), [GenBank: KF134531] (Lt1αEF cDNA), [GenBank: KF134524] (Lc1αEF cDNA), [GenBank: KF428722] (LtbHLH cDNA), [GenBank: KF428721] (LcbHLH), [GenBank: KF134528] (LcTT2 cDNA), [GenBank: KF134534] (LtPAL cDNA), [GenBank: KF134527] (LcPAL cDNA), [GenBank: KF428720] (LtDFR genomic), [GenBank: KF428719] (LcDFR genomic), [GenBank: KF134530] (LtCHS cDNA), [GenBank: KF134523] (LcCHS cDNA), [GenBank: KF134522] (LcANS cDNA), [GenBank: KF134529] (LtANR cDNA), [GenBank: KF134521] (LcANR cDNA), [GenBank: KF134532] (LtLAR1 cDNA), [GenBank: KF134525] (LcLAR1 cDNA), [GenBank: KF134533] (LtLAR2 cDNA), [GenBank: KF134526] (LcLAR2 cDNA), [GenBank: KF386027] (LtLAR2 genomic) and [GenBank: KF386026] (LcLAR2 genomic).
Statistical analysis was performed using the Infostat program . One-way ANOVA was carried out for all analyses and a Tukey test was used for a multivariate analysis (p < 0.05). For non-parametric data the Kruskal-Wallis test with paired comparison (p < 0.05) was performed. The statistical analysis of the relative gene expression was performed according to Pfaffl et al. .