Introduction

Lentil is a self-pollinating annual cool season legume. Similar to other legumes, lentils contain proteins, carbohydrates, minerals, and vitamins that are important for human nutrition. Lentils are also a good source of phenolic compounds such as flavan-3-ols, proanthocyanidins (also known as condensed tannins) [1–4], flavones [3, 5], flavonols [3–5], stilbenes [5], phenolic acids [3, 5, 6], and anthocyanidins [7]. The range of background colours and patterns of lentil seed coats determines the market classes for whole lentil seeds. Approximately 25 % of the lentil crop in Canada, the world’s major producer, is marketed as “green lentil”, which typically has a pale green seed coat covering a yellow cotyledon [8].

The green seed coat colour in lentil is not stable and changes over time to yellow, yellow–brown, medium brown, and dark brown depending upon the storage conditions and duration [9]. Figure 1 shows how the seed coat colour of CDC Improve (CDC stands for the Crop Development Center at University of Saskatchewan) darkened after 8 years of storage. Seed coat colour is an important grading factor that affects the market value of lentils. According to the Canadian Grain Commission, green lentils with good natural colour are graded as No. 1 lentils [10]. An increase in seed coat colour variability, or darkening of the seed coat, will decrease the grade of the sample, thereby reducing the offered price [11]. For No. 3 grade green lentil (severe discolouration, i.e. dark brown [10]), the average price is approximately half that of No. 1 grade large green lentil (e.g. see http://www.statpub.com/stat/prices/spotbid.html [12]). A better understanding of the biochemicals involved in seed coat darkening will inform breeding strategies aimed at overcoming this process, thereby preserving economic value.

Fig. 1
figure 1

Change in the seed coat colour of CDC Improve lentil from green (left—2014) to dark brown (right—2006) after eight years of storage

Full size image

Different pigments such as chlorophylls and phenolic compounds form the colourful world of plants. While chlorophylls are characterized by a green colour, phenolic compounds such as flavan-3-ols, proanthocyanidins, flavones, and flavonols are colourless, white, and pale yellow. Anthocyanidins make intense colours from orange and red to blue [13]. Greener lentil seeds have higher amounts of chlorophyll a and b [11]. Chlorophyll breaks down with age and is converted to colourless compounds in chloroplasts. These colourless compounds will change to differently structured compounds that are finally stored in vacuoles [14, 15]. Seed coat darkening is a phenomenon that can significantly affect the greenness of the seed coat, and phenolic compounds play an important role. The effect of storage on the total amount of phenolics (by spectrophotometric methods) or subclasses of phenolic compounds (applying chromatographic methods) has been studied in various plant materials (Supplementary Table 1). Spectrophotometric analysis of total phenolics and total proanthocyanidins in lentil seed demonstrated a decrease after storage [9, 16]. Several chromatographic analyses have been applied to other plant materials, but, to our knowledge, no studies have examined how different subclasses of phenolic compounds in lentil seed coats might change as a result of storage.

Recently, we reported the optimization of a targeted liquid chromatography–mass spectrometry (LC–MS) method for use in profiling phenolic compounds in lentil seed coats [17]. The objective of the current study was to use this LC–MS method to compare phenolic compound profiles of fresh green seeds with those that have experienced long-term storage (LTS).

Materials and methods

Plant material

A preliminary study analysed seeds of the recombinant inbred line (RIL) LR-18-183 from the LR-18 (CDC Robin × 964a-46) [18] population harvested in 2009 and 2014. Seeds were obtained from the Crop Development Centre (CDC), University of Saskatchewan, Saskatoon, Canada. Breeder seed of six green lentil cultivars (Table 1) was also obtained from the same source. Fresh seeds were harvested in 2014, and for comparison, older seeds (harvested in 2000–2007) were obtained from the long-term storage facility. The storage facility was a routine type that is commonly used for seed storage; as a result the temperature and humidity were not controlled and were subject to seasonal fluctuation. Both fresh and LTS seeds were analysed four months after harvest of the 2014 seeds in three replicates for each cultivar as described below.

Table 1 Genotype names and storage duration of lentils used in this experiment
Full size table

Reagents and standards

Tables 2 and 3 show the phenolic compounds analysed by LC–MS. Since our method [17] was published, we have added new compounds. Specifically, resveratrol-3-β-mono-d-glucoside and quercetin-3-O-rhamnoside were obtained from Santa Cruz Biotechnology, Inc. (California, USA), and Extrasynthese (Genay, France), respectively, whereas vanillic acid-4-β-d-glucoside, kaempferol-3-O-robinoside-7-O-rhamnoside, (±)-catechin-2,3,4-13C3, and vanillin-(ring)-13C6 were purchased from Sigma-Aldrich (Missouri, USA). Note the last two compounds were used as internal standards (IS). Catechin-3-glucoside and kaempferol dirutinoside (Table 2), as well as several oligomers of proanthocyanidins (Table 3), which were found in the lentil seed matrix but did not exist commercially, were analysed based upon previous reports [2, 17]. Note that for catechin-3-glucoside and kaempferol dirutinoside the type of sugar was determined based on previous literature [2, 3], yet the exact bond location could not be confirmed. Also, for oligomers of proanthocyanidins, the order of C’s (catechin or epicatechin) and G’s (gallocatechin or epigallocatechin) was arbitrary.

Table 2 Phenolic compound characteristics in LC-selected reaction monitoring (LC-SRM), including subclass, retention time, molecular ion, and optimum fragment ion in positive or negative mode
Full size table
Table 3 Proanthocyanidin characteristics in LC-single ion monitoring (LC-SIM), including retention time and molecular ion in positive mode
Full size table

Sample preparation

Samples were prepared based on previous sets of optimization tests with minor modifications [17]. In summary, for each replicate, 1000 µL of the extraction solvent [acetone/water (70:30 v/v)] was added to ~250 mg of freeze-dried whole lentil seeds in a micro-centrifuge tube. Two ceramic sphere beads (¼ inch) were added to each tube and the seeds pulverized to a fine paste using a Mini-Beadbeater-16 (BioSpec Products, Inc., OK, USA) for 5.5 min. Samples were then shaken for 1 h on a Thermomixer (Eppendorf, Germany) at a speed of 1400 rpm at room temperature. Each tube was centrifuged (12,000 rpm for 5 min) and the supernatant extracted and transferred to a new tube. The centrifugation and sample transfer step was then repeated. A portion of the final supernatant was diluted 10 times by Milli-Q water and transferred to glass vials for analysis.

HPLC–MS

Previously optimized chromatographic conditions [17] were applied using an Agilent 1100 (Agilent, Germany) high-performance liquid chromatograph (HPLC) with G1315 PDA UV/Vis detector coupled to a Thermo Finnigan TSQ Quantum Ultra (Thermo Fisher Scientific Inc., UK) triple quadrupole MS equipped with a heated electrospray ionization (HESI) interface. Peak areas were obtained with Thermo Xcalibur 2.1 software. The chromatographic column was a Core–shell Kinetex pentafluorophenyl (PFP), 100 × 2.1 mm id, 2.6 μm particle size (Phenomenex, Torrance, CA). The mobile phases consisted of H2O/formic acid (FA) (99:1, v/v) for solvent A and H2O/acetonitrile (ACN)/FA (9:90:1, v/v/v) for solvent B. The flow rate was 0.35 mL/min, and the injection volume was 5 µL. The gradient used is given in Supplementary Table 2. The optimization results for retention time as well as molecular and fragment ions of the standards and potential compounds from different subclasses of phenolic compounds that were analysed in this experiment are reported in Tables 2 and 3. Relative quantification of phenolic compounds was done using selected reaction monitoring (SRM) in positive or negative mode (Table 2) and single ion monitoring (SIM) in positive mode (Table 3). For SRM analyses, three consecutive functions were defined with time ranges of 6.5, 10.5, and 13 min duration, respectively, in the mass spectrometry software. For SIM analyses, one function with several transitions was used. The reproducibility of the LC–MS method was checked as described previously [17]. To ensure that the targeted method was not missing any significant changes in the phenolic compound profiles, full-scan LC–MS spectra (m/z 140–1500) with UV detection (250–600 nm) were also acquired in both positive and negative modes (Fig. 2).

Fig. 2
figure 2

a Chromatograms obtained using mass spectrometry and total ion current detection, and b chromatograms obtained using UV detection (250–600 nm) using seeds of CDC Improve lentil [fresh is top trace, LTS is lower trace in both (a) and (b)]. The chromatographic conditions described in Supplementary Table 2

Full size image

Data analyses

The analysis of lentil seeds in this research is based upon relative quantification (i.e. area ratio). Relative quantification is commonly used for comparative analyses, especially in metabolomics applications [20], whenever a quantitative measure of the relative amount but not the absolute amount is required, as is the case here in examining changes in phenolic compounds. Thus, with relative quantification, the analyte signal intensity is normalized to that of an internal standard by dividing the integrated area of each phenolic compound to the integrated area of a related IS and reported as area ratio for a given analyte. The area ratio was described per mg weight of each seed sample variety, and the average for three replicates was reported.

A comparison among means was done using R software (v. 2.15.3) [21]. The experimental design was based upon randomized complete blocks.

Results

Samples of LR-18-183 used in the preliminary test were harvested in 2009 and 2014 and compared using LC–MS/MS [17] to investigate changes in the phenolic profile that occurred during lentil storage (Supplementary Figure 1). The study was used to assess whether the changes would be sufficient to warrant a more rigorous test of the rest of the population for this type of analysis. Supplementary Figure 1 shows only small differences between the samples over a period of 5 years. Note that during the course of the year, the daily average temperature varies between −14 and +19 °C in Saskatoon [22]. Thus, a plausible explanation for the minimal change is the influence of the very low temperatures (below 0 °C) experienced during winter storage in Saskatoon (6 months); that is, during the winter months it is expected that little to no change is occurring. Because the available RILs that were less than 5 years old showed only small changes, it was therefore decided to investigate further using older seed of several cultivars with green seed coats. This was done in order to maximize the amount of time for change to occur, which in turn should help to simplify data interpretation. Consequently, the oldest breeder seed available was used and the storage durations varied from 7 to 14 years (Table 1).

Signal responses were found for all phenolic compounds analysed using SRM and SIM methods and in both fresh and LTS seeds for all six genotypes including phenolic acids, stilbenes, flavones, flavonols, and flavan-3-ols. Moreover, dimers, trimers, tetramers, and pentamers of proanthocyanidins were observed in lentil samples.

Figure 3a–f shows mean area ratio of phenolic compounds per mg of fresh and old samples of CDC Imigreen, CDC Impower, CDC Greenland, CDC Improve, CDC Meteor, and Plato genotypes. Vanillic acid-4-β-d-glucoside (phenolic acids subclass) and luteolins (flavones subclass) were elevated in LTS seeds compared to fresh seeds of most of the genotypes. Mean area ratio of the flavonols subclass per mg of different genotypes remained mostly unchanged between fresh and LTS samples; though kaempferol-3-O-robinoside-7-O-rhamnoside and kaempferol dirutinoside showed lower values in some of the long-term stored genotypes (e.g. CDC Impower, CDC Greenland, and Plato), and quercetin-3-O-rhamnoside showed higher values (e.g. CDC Meteor and Plato). A significant reduction after storage in mean area ratio of flavan-3-ols (including catechin, gallocatechin, and catechin-3-glucoside) per mg of all six genotypes observed. A similar reduction was noticed in the 24 detected proanthocyanidins (including dimers, trimers, tetramers, and pentamers of catechin/epicatechin and gallocatechin/epigallocatechin) for all the genotypes.

Fig. 3
figure 3figure 3

Mean area ratio per mg of fresh and old samples in a CDC Imigreen, b CDC Impower, c CDC Greenland, d CDC Improve, e CDC Meteor, and f Plato varieties. Error bars are standard errors of three replicates. C and G represent catechin/epicatechin and gallocatechin/epigallocatechin

Full size image

Full-scan analyses of the fresh and LTS lentil seeds did not show the addition or omission of any major peaks in the chromatograms obtained using total ion current plots or UV detection (e.g. see traces for fresh and LTS samples of CDC Improve, Fig. 2a, b). As a result, we were able to rule out the possibility of significant changes occurring in the phenolic compounds profile from species that were not targeted by our SRM and SIM methods.

Discussion

Marketability of green lentil depends largely on a stable green colour of the seed coats, and storage can dramatically affect the greenness and marketability of this crop. Phenolic compounds are very involved with seed darkening [9, 23–32]. Using LC–MS, we compared the phenolic profiles of fresh seeds and the oldest available breeder seed samples (to maximize the ageing time). The type of phenolic compounds observed in our samples is consistent with previous analyses of lentil seeds [1–6].

Storage increased the amount of vanillic acid-4-β-d-glucoside, luteolin, and luteolin-4′-O-glucoside. Srisuma et al. [27] and Aaby et al. [28] reported the rise of phenolic acids, such as ferulic, sinapic, and ellagic acids, in other plant materials. These increases could be related to the degradation of more complex phenolic compounds such as those with several sugar conjugates [33].

Although we observed some increases (e.g. quercetin-3-O-rhamnoside) and decreases (e.g. kaempferol-3-O-robinoside-7-O-rhamnoside and kaempferol dirutinoside) in flavonols when comparing fresh and LTS samples, these compounds, on average, remained essentially unchanged during LTS conditions. This is similar to reports for pinto bean, for which the storage ratio average of analysed flavonols in non-aged and aged seeds was not significantly different [30]. The pattern for the changes in mean area ratio per mg sample indicated that flavonols with a largest number of sugars decrease slightly whereas those with only one sugar increase slightly. These data support the hypothesis that more complex phenolic compounds (e.g. those with a larger number of sugars) may be breaking down to produce more compounds with a smaller number of sugars.

The LC–MS analysis of the lentil samples showed a significantly declining trend for all 27 flavan-3-ols and oligomers of proanthocyanidins after LTS. Some of the variations in the degree of change among the different genotypes may have to do other variables (e.g. environmental factors), but since all six genotypes showed the same decreasing trend after storage, it is extremely likely that these changes are attributed to storage effects. In other plant materials, flavan-3-ols and oligomers of proanthocyanidins analysed separately by chromatography [26, 29, 32] and/or combined with MS methods [28] as well as total proanthocyanidins analysed by spectrophotometric methods [9, 25, 30, 31] also declined after storage.

Green lentil seed coats darken over time during storage to dark brown, and this results in leakage of different soluble materials into the imbibition medium [9]. Nevertheless, phenolic compounds were not found in large quantities among the leaked materials. The authors proposed polymerization of low molecular weight flavan-3-ols and oligomers of proanthocyanidins as the possible reason [9]. Different models have been introduced for the biosynthesis of proanthocyanidins, including conversion of flavan-3-ols to quinone methides or their protonated carbocation, which results in polymerization into colourless proanthocyanidins [34]. Proanthocyanidins can be produced either through the endoplasmic reticulum or via plastids such as chloroplasts [35]. Re-differentiation of chloroplasts, which occurs especially under stress or ageing processes [36], causes swelling of chloroplasts and formation of tannosomes, which are structures that contain thylakoids and tannins [37]. This could explain why the green colour is replaced by a darker colour. Proanthocyanidins produced in the endoplasmic reticulum and tannosomes from chloroplasts are mostly stored in vacuoles with a minor percentage in cell walls [38]. Proanthocyanidins have oxidizable OH groups (OH groups that are adjacent to each other), which make them good substrates for oxidative enzymes such as polyphenol oxidase (PPO) and peroxidase (POD) [39]. Note that PPO is located in plastids, whereas POD occurs in plastids, mitochondria, and cytosol [38]. Stress or ageing could increase reactive oxygen species (ROS) in the stored plant materials. ROS could react with cell membrane lipids and cause breakdown of membranes and decompartmentalization of organelles [40]. This will intermix oxidative enzymes with their potential substrates, i.e. proanthocyanidins [38], a process that will produce short-lived highly reactive intermediates, such as semiquinones and quinones [38, 41]. As a result, some non-enzymatic reactions with other phenolic compounds in particular, but also with proteins and polysaccharides in the cell wall, could be expected to occur [41]. Cross-linking of proanthocyanidins with other phenolic compounds and especially with the cell wall has been proposed as the source of the brownish compounds that give the seed coat colour [35, 42]. Some quinone products have been observed in vitro [43, 44], but an LC–UV–MS analysis did not provide any evidence for this in our lentil samples. A comparison of chromatograms obtained using total ion current and UV detection of fresh and LTS samples of green lentil seeds did not show significant differences. We cannot rule out the possibility that lower abundance species may change because not all peaks were large enough to be separated from the noise in these chromatograms. More likely, numerous possible new quinone species could form from ROS, but these species are highly reactive and could react with several different phenolic compounds, thereby producing many possible products. Thus, the overall signal would be diluted into many small signals that would become indistinguishable from the baseline. This is in contrast to in vitro experiments in which only a small number of compounds were present and therefore specific favoured pathways could produce identifiable peaks.

Proanthocyanidins make up the largest proportion of phenolic compounds in lentil seed coat tissue [5]. The bursting of vacuoles caused by cell death might transport proanthocyanidins from the vacuole to the cell wall [14]. Pang et al. [45] suggest that the proanthocyanidins are initially soluble when they are in vacuoles and then become insolubilized after attaching to the cell wall. Proanthocyanidins are good H-donors and can easily make hydrogen bonds [46]. Having several reactive sites, oligomers and polymers of proanthocyanidins can be encapsulated within the gel structure of cell wall polysaccharides. Proanthocyanidins and cell wall polysaccharides bind through H-bonding and hydrophobic interactions [47]. All of these will result in stronger binding of proanthocyanidins with cell wall materials and make them extremely hard to extract [48], which is the reason for the significant reduction in their storage ratio in our lentil samples.

Overall, this work addresses the fact that LTS darkens the green colour in lentil seeds; this reduces marketability, as the value of green lentil is based on the visual characteristics of the seed coat. Increases in phenolic acids and flavones occur in green lentil seeds after storage, possibly because of the breakdown of more complex species into smaller subunits. A significant decrease in 27 flavan-3-ols and proanthocyanidins also occurs. During storage, enzymatic and non-enzymatic reactions will polymerize proanthocyanidins and result in cross-linking of these major phenolic compounds with cell wall materials. This will produce dark pigments and reduce their extractability. The findings of this study help to narrow down the genes of interest responsible for lentil seed darkening. However, enzymatic analysis should be approached as a more explicit indication of how phenolic compounds change over time.