Proteome reorganization and amino acid metabolism during germination and seedling establishment in Lupinus albus

Abstract

During germination plants rely entirely on their seed storage compounds to provide energy and precursors for the synthesis of macromolecular structures until the seedling has emerged from the soil and photosynthesis can be established. Lupin seeds use proteins as their major storage compounds, accounting for up to 40% of the seed dry weight. Lupins are therefore a valuable complement to soy as a source of plant protein for human and animal nutrition. The aim of this study was to elucidate how storage protein metabolism is coordinated with other metabolic processes to meet the requirements of the growing seedling. In a quantitative approach, we analysed seedling growth, as well as alterations in biomass composition, the proteome, and metabolite profiles during germination and seedling establishment in Lupinus albus. The reallocation of nitrogen resources from seed storage proteins to functional seed proteins was mapped based on a manually curated functional protein annotation database. Although classified as a protein crop, Lupinus albus does not use amino acids as a primary substrate for energy metabolism during germination. However, fatty acid and amino acid metabolism may be integrated at the level of malate synthase to combine stored carbon from lipids and proteins into gluconeogenesis.

Amino acid metabolism, germination, glyoxylate cycle, Lupinus albus protein annotation database, nitrogen resource allocation, proteomics

Introduction

The genus Lupinus is part of the Fabaceae family and includes the annual herbaceous species L. albus, L. angustifolius, L. luteus, and L. mutabilis cultivated as crops (Lucas et al., 2015). The white lupin (Lupinus albus) originated in the Mediterranean region and produces large, vigorous seeds facilitating rapid growth and completion of the lifecycle before summer drought (Berger et al., 2017). Due to their high protein content of 30–40% and an adequate balance of essential amino acids, L. albus seeds are an excellent protein source for a plant-based diet (Sujak et al., 2006). A low oil and high dietary fiber content in combination with high levels of tocopherols and a low glycemic index add to the positive seed qualities (Boschin et al., 2008; Boschin and Arnoldi, 2011; Fontanari et al., 2012). Lupin cultivation does not require nitrogen or phosphate fertilization due to rhizobial symbiosis and specialized cluster root structures. It can therefore be categorized as particularly environmentally friendly (Lucas et al., 2015; Xu et al., 2020; Pueyo et al., 2021). However, L. albus wild varieties and landraces accumulate high concentrations of toxic quinolizidine alkaloids (≤11%) in their seeds (Rybiński et al., 2018). These specialized metabolites derived from the amino acid lysine protect the plant from pathogen attack but cause symptoms of poisoning in humans, affecting the nervous, circulatory, and digestive systems (Boschin and Resta, 2013). Mutations at the pauper locus decrease the seed alkaloid levels below 0.02% in the cultivated sweet L. albus varieties making them suitable for human consumption and animal feed (Osorio and Till, 2021).

From the plant perspective, the seed storage compounds provide energy and precursors for the synthesis of macromolecular structures until the seedling has emerged from the soil and photosynthesis has been established. In some species including lupins the cotyledons are not shed but converted to photosynthetically active leaves after germination and thus can serve as a storage tissue for a prolonged time. About 35–40% of the dry weight of L. albus seeds consists of carbohydrates, mostly cellulose and oligosaccharides of the raffinose family, which are non-digestible in humans and other monogastric animals (Gdala and Buraczewska, 1996; Sanyal et al., 2023). Plants activate α-galactosidases during germination to hydrolyse raffinose to sucrose and galactose, which can then serve as a source of energy and as a precursor for cellulose synthesis (Elango et al., 2022, Sanyal et al., 2023). Lupinus albus seeds also store lipids in oil bodies (7–14%; Borek et al., 2012). During germination the triacylglycerols are hydrolysed by lipases and the free fatty acids are further metabolized to acetyl-CoA and NADH via β-oxidation in the peroxisomes (Baker et al., 2006; Graham, 2008). The glyoxylate cycle, including peroxisomal steps catalysed by isocitrate lyase, citrate synthase and malate synthase as well as cytosolic steps catalysed by aconitase and malate dehydrogenase, converts acetyl-CoA to succinate for the synthesis of carbohydrates during gluconeogenesis (Baker et al., 2006; Pracharoenwattana and Smith, 2008). For fatty acid respiration citrate is exported from the peroxisomes into the mitochondria and feeds into the TCA cycle to produce ATP (Pracharoenwattana et al., 2005). Storage lipid metabolism has been extensively studied in lupins (Borek et al., 2012, 2013, 2015). In addition to their role in gluconeogenesis and ATP production, it has been demonstrated that lipid-derived carbon skeletons are converted to amino acids (Borek et al., 2003; Borek and Ratajczak, 2010), which is astonishing given the fact that lupin seeds contain 30–40% protein consisting predominantly of storage proteins from the globulin class (Duranti et al., 2008). Thus, the demand for free amino acids during germination should be covered by storage protein breakdown. Surprisingly little research has systematically focused on amino acid metabolism in germinating lupin seeds. Asparagine, which is the major transport form of nitrogen in many plant species including lupins, strongly accumulates during seedling establishment (Lea et al., 2007). Asparagine synthesis assimilates free ammonium via the combined activities of glutamine synthetase and asparagine synthetase (Lea and Fowden, 1975). The most obvious source of ammonium in germinating seeds without external nitrogen supply would be its release during amino acid catabolism. It has been postulated that the oxidation of amino acids provides alternative substrates for mitochondrial respiration in germinating lupins (Borek et al., 2017). Amino acid oxidation pathways are strongly induced during energy deficiency responses in Arabidopsis (Heinemann and Hildebrandt, 2021) and have been suggested to be relevant during seed development (Galili et al., 2014). It has also been shown that excised pea embryo axes and isolated mitochondria are able to use externally supplied glutamate as a respiratory substrate (Morkunas et al., 2000). However, to our knowledge the postulated role of amino acids derived from storage protein degradation as substrates for mitochondrial respiration in germinating lupins has not been demonstrated yet.

The release of high-quality genome data for L. albus by two different groups in 2020 (Hufnagel et al., 2020; Xu et al., 2020) recently opened up the possibility to comprehensively address quantitative changes in the proteome of germinating lupin seeds by proteomics approaches. The genome assembly presented by Hufnagel et al. (2020) covers 451 Mb; 38 258 protein coding genes were predicted based on RNAseq data from 10 different tissue types in combination with the proteome of Medicago truncatula. The assembly by Xu et al. (2020) is larger (559 Mb) and includes 46 596 coding sequences predicted using protein coding sequences of Arabidopsis and several legume species as well as RNAseq data from root, leaf and stem tissue. However, functional annotation of the proteins in the reference database available on uniport.org and gene ontology classifications are still fragmented, which limits the interpretation of proteomics and transcriptomics datasets. We therefore created a manually curated protein annotation database for L. albus including functional categorization according to a modified version of the MapMan annotation system (Thimm et al., 2004). In this study we combine proteomics based on the new annotation database with metabolite and physiological data to investigate nitrogen resource allocation and utilization of seed storage proteins during germination and seedling development in L. albus. Our results provide new insight into the specific challenges and chances associated with the use of proteins as major seed storage compounds.

Materials and methods

Plant material

White lupin (Lupinus albus cv. ‘Nelly’) seeds were stored in the dark at 4 °C until use. Before cultivation, the seeds were soaked in demineralized water at 20 °C for 16 h and then transferred to water-saturated expanded clay substrate (LamstedtDan, 4–8 mm, Fibo ExClay Deutschland GmbH, Lamstedt, Germany). Plants were cultivated in a phytochamber (22–24 °C, 16 h light, 8 h dark, 110 µmol m⁻² s⁻¹ light). At harvest, the seed coats were removed and the remaining plant organs were deep-frozen in liquid nitrogen. The plant material was lyophilized in an Alpha 1-2 LD+ freeze dryer (Christ, Osterode, Germany). The dried material was then ground into powder and used for the investigations. For dry seeds and young seedlings several plants were pooled (10 seeds, five seedlings at day 4, two seedlings at day 8), and five different plants or pools were analysed at each time point as biological replicates. The time points for harvesting were selected based on preliminary experiments to cover germination as well as the subsequent seedling developmental stages characterized by massive degradation of seed storage proteins.

Quantification of total carbon and nitrogen

The total carbon and nitrogen content of the samples was measured according to Andrino et al. (2019), using an Elementar vario MICRO cube C/N analyser (Elementar GmbH, Hanau, Germany).

Quantification of total lipids

The total lipid content was determined using the sulfo-phospho-vanillin method described in Park et al. (2016). Commercial lupin oil was used as a standard.

Quantification of total carbohydrates

The total carbohydrate content was determined using the phenol–sulfuric acid method described by Tamboli et al. (2020). Five milligrams of lyophilized plant powder was dissolved in 1 ml of 2.5 M HCl and incubated for 3 h at 95 °C, shaking. The extracts were diluted (1:50) with demineralized water and 10 µl phenol and 1 ml concentrated sulfuric acid were added. After incubation (10 min, 95 °C, shaking) the absorbance was measured at 490 nm with a plate reader (Multiskan Sky, Thermo Fisher Scientific, Dreieich, Germany).

Quantification of cellulose

The cellulose content was determined using a modification of the method originally described by Updegraff (1969). Five milligrams of lyophilized plant powder was dissolved in 1 ml acetic acid–demineralized water–nitric acid (8/2/1, v/v/v) and incubated shaking for 30 min at 100 °C. After a centrifugation step, the supernatant was evaporated in a vacuum concentrator, and the pellets were washed with demineralized water and then dissolved in 1 ml sulfuric acid (72%). The samples were further diluted to 5 ml with demineralized water. Fifty microliters of the extract was mixed with 150 µl of demineralized water and 500 µl of anthrone solution (0.3% anthrone in concentrated sulfuric acid) and incubated shaking at 100 °C for 20 min. The absorbance at 620 nm was measured and quantified against a cellulose standard.

Quantification of chlorophyll

The quantification of chlorophyll was carried out according to a modified version of the method described by Lichtenthaler (1987). Five milligrams of plant powder was dissolved in 700 ml methanol (100%) and incubated for 20 min at 80 °C with shaking. After centrifugation (10 min, 4 °C, 18 800×g), the chlorophyll content of the supernatant was quantified with a plate reader (Multiskan Sky) (wavelengths: 470, 653, and 666 nm).

Quantification of proteins

The protein extraction protocol for quantification of the total seed/seedling protein content was rigorously tested and optimized for quantitative recovery of all proteins. Four milligrams of lyophilized plant powder was dissolved in 1 ml methanol (100%) and incubated for 20 min at −20 °C. After centrifugation (14 200×g, 5 min, 4 °C) the pellet was dissolved in 1.4 ml of 0.1 M NaOH containing 2% SDS (v/w) and incubated for 1 h at 60 °C, shaking. After centrifugation (7000×g, 10 min, room temperature), the supernatant was diluted with demineralized water (1:10). The Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, Rockford, IL, USA) was used for protein quantification with globulin as a standard. This protein extraction protocol was clearly superior to phenol and trichloroacetic acid (TCA) precipitation in terms of protein recovery from Lupinus albus seeds. Defatting of the plant material did not further increase protein recovery.

Quantification of free amino acids

Free amino acids were quantified according to Batista-Silva et al. (2019). The pre-column derivatization with ophthaldialdehyde (OPA) and fluorenylmethoxycarbonyl (FMOC) was based on the Agilent application note ‘Automated amino acids analysis using an Agilent Poroshell HPH-C18 Column’. Samples were injected onto a 100 mm, 3 mm InfinityLab Poroshell HPH-C18 column (2.7 mm) using an Ultimate 3000 HPLC system (Thermo Fisher Scientific).

Respiration measurements

Respiration rates of seeds and seedlings were measured at 20 °C in demineralized water using an Oxygraph-2k respirometer (Oroboros Instruments Corp., Innsbruck, Austria).

SDS-PAGE

Proteins were separated by glycine SDS-PAGE according to a modified protocol of Laemmli (1970). Exactly 5 mg of plant powder was dissolved in 1 ml methanol (100%) and incubated for 20 min at −20 °C. After centrifugation (10 min, 4 °C, 18 800×g) the pellet was air dried. Then 200 µl of sample buffer (125 mM Tris–HCl, pH 6.8, 4% SDS, 20% glycerol) was added and incubated for 1 h at 95 °C, shaking. The samples were cooled on ice and centrifuged (5 min, 18 800×g, 4 °C). The supernatant was used for glycine-SDS-PAGE. Separation gels contained 4.5% polyacrylamide. Amersham ECL Rainbow Marker—Full Range (High Range, Cytiva RPN756E, Sigma-Aldrich, St Louis, MO, USA) was used as size standard. The same volume of protein extract was used per sample. Gels were stained with colloidal Coomassie Blue. Selected protein bands were cut and digested with trypsin as described in Klodmann et al. (2010), desalted using homemade C18-StageTips (Kulak et al., 2014), and identified by mass spectrometry as detailed in the section entitled ‘Quantitative proteomics by shotgun mass spectrometry’.

Protein extraction and purification for proteome analysis via mass spectrometry

For protein extraction, clean-up, and digestion, we used an adapted ‘single-pot solid-phase-enhanced sample preparations (SP3)’ protocol from Hughes et al. (2019) and Mikulášek et al. (2021). The mechanism relies on hydrophilic interactions between proteins and paramagnetic beads, which can be accumulated on a magnetic rack for fast washing steps and buffer exchanges. The approach is known for its efficient, unbiased, and robust processing of protein samples for proteomic analysis. In brief, 5 mg of lyophilized plant powder (~500 µg plant protein) was dissolved in 250 µl of 2× SDT buffer (8% SDS, 0.2 M dithiothreitol, 0.2 M Tris pH 7.6) and incubated at 60 °C for 30 min to extract, denature, and reduce the proteins. Samples were then sonicated for 10 min and centrifuged at 20 000×g for 10 min. Thirty microliters (~30 µg protein) of the supernatant was transferred to new tubes and mixed with 7.5 µl iodoacetamide (0.1 M) in order to alkylate the reduced disulfide bridges. The samples were incubated for 30 min in the dark. Then 2 µl dithiothreitol (0.1 M) was added to each sample to neutralize excess amounts of iodoacetamide. Preparation of the magnetic beads (Sera-Mag carboxylate-modified beads, hydrophobic solids no. 441521050250, hydrophilic solids no. 241521050250, GE Healthcare Life Sciences), protein binding and washing steps were performed on a magnetic rack as described in Mikulášek et al. (2021). We used a bead to protein ratio of 20:1 (w/w; 600 µg beads). The bound and purified proteins were digested overnight at 37 °C with 0.5 µg trypsin (mass spectrometry grade, Promega) per sample. The peptide-containing supernatants were collected in tubes with low peptide binding. The magnetic beads were rinsed in 60 µl ammonium bicarbonate (50 mM) to recover the remaining peptides. Both eluates were combined and acidified with formic acid. The peptides were desalted with 50 mg Sep-Pak tC18 columns (WAT054960, Waters) and quantified using the Pierce Quantitative Colorimetric Peptide Assay Kit (Thermo Fisher Scientific, Rockford, IL, USA). The samples were finally diluted to a target concentration of 200 ng μl⁻¹ in 0.1% formic acid.

Quantitative proteomics by shotgun mass spectrometry

Four hundred nanograms of peptides was injected via a nanoElute UHPLC (Bruker Daltonic, Bremen, Germany) and separated on an analytical reversed-phase C18 column (Aurora Ultimate 25 cm×75 µm, 1.6 µm, 120 Å; IonOpticks). Using a multistaged linear gradient (eluent A: MS grade water containing 0.1% formic acid, eluent B: acetonitrile containing 0.1% formic acid, gradient: 0 min, 2%; 54 min, 25%; 60 min, 37%; 62 min, 95%; 70 min, 95% eluent B), peptides were eluted and ionized by electrospray ionization using a captive spray source with a flow of 300 nl min⁻¹. A standard application method [data-dependent acquisition parallel accumulation–serial fragmentation (DDA-PASEF) cycle time 1.1 s] was used for data collection without further modification. Ion mobility spectrometry–MS/MS spectra were analysed using MaxQuant (version 2.0.3.0, Cox and Mann, 2008). Protein identification was based on the proteome of Lupinus albus published on uniport.org by Xu et al. (2020) (UP000464885). The search parameters of the software were set as follows: carbamidomethylation (C) as fixed modification, oxidation (M) and acetylation (protein N-term) as variable modifications. The specific digestion mode was set to trypsin (P), with a maximum of two missing cleavage sites allowed. A positive peptide identification had to contain at least seven amino acids. Mass tolerance for fragment ions was set at 40 ppm. The false discovery rate at the protein and peptide spectrum match level was set at 1%. The minimum number of unique peptides per protein group was 1. Both label-free quantification (LFQ) values and intensity-based absolute quantification (iBAQ) values were calculated for the identified proteins.

Data processing and functional annotation

Intensities of unique and razor peptides were used for protein quantification. Further analysis and statistical evaluation based on iBAQ values and LFQ values generated by MaxQuant was performed using Perseus (version 1.6.1.1) (Tyanova et al., 2016). Fold changes of individual proteins can be estimated using LFQ values (Cox et al., 2014). This approach is suitable for identifying proteins that are induced and could therefore be particularly relevant under the conditions tested. However, changes in the absolute amounts of individual proteins can be estimated by iBAQ values. The MaxQuant output table was filtered to remove non-plant contaminants, reversed sequences, and proteins identified only on the basis of modified peptides. Proteins were excluded from further analysis if they were not detected in at least four out of five biological replicates in at least one of the sample groups. Subsequently, the missing protein intensities were considered too low for proper quantification and were randomly replaced with low values of a normal distribution. Significant changes were calculated from the LFQ values using Student’s t-test (P=0.05). A functional annotation database (Supplementary Dataset S1) was built by sequence alignments (BLASTp) of the proteome of Lupinus albus (UP000464885) against proteomes of Lupinus angustifolius (UP000188354), Medicago truncatula (UP000002051) and Arabidopsis (TAIR10). Candidate consensus was achieved by highest bit score and/or manual curation. We then used common bioinformatics tools of the model plant Arabidopsis to obtain further information on the subcellular localization (SUBA.live) and metabolic pathway involvement (mapman.gabipd.org) of all individual proteins.

Results

The Lupinus albus proteome

The genome of L. albus contains approximately 46 500 protein coding genes (Hufnagel et al., 2020; Xu et al., 2020). However, only 5% of the proteins have been functionally annotated so far in the reference database available on uniprot.org (UP000464885, UP000447434). The annotation tool Mercator (https://www.plabipd.de/mercator_main.html; Schwacke et al., 2019; Bolger et al., 2021) offers a fast and convenient way to greatly improve the coverage of the proteome. However, a curated mapping file specific for Lupinus albus is not available yet, leading to gaps in the classification of characteristic lupin proteins such as seed storage proteins. We therefore created an extended annotation database based on sequence alignments (BLASTp) of L. albus proteins against the proteomes of Lupinus angustifolius (UP000188354), Medicago truncatula (UP000002051) and Arabidopsis (TAIR10). We used the larger genome assembly (Xu et al., 2020) for initial sequence alignments and subsequently assigned the genes detected by Hufnagel et al. (2020) via blast. This allowed us to name 71% of the proteins in the Lupinus albus proteome (Fig. 1; Supplementary Dataset S1). Linking each lupin protein to the closest homologue in Arabidopsis makes it possible to apply bioinformatic tools and resources available for the model plant Arabidopsis to L. albus datasets. For functional annotation we divided the L. albus proteome into 20 major categories with two levels of sub-categories based on a modified version of the MapMan annotation system (Thimm et al., 2004; Supplementary Dataset S1). Fifty-four percent of the proteome could confidently be assigned to a functional category (blast bitscore >200) and an additional 3% was tentatively assigned (bitscore <200) (Fig. 1B; Supplementary Dataset S1). The remaining 20 140 proteins are homologues of previously uncharacterized proteins or of proteins lacking a functional annotation in Arabidopsis and therefore remain in the category ‘not assigned’.

Fig. 1.

Lupinus albus proteome annotation. (A) Schematic presentation of the workflow used for curating the L. albus proteome annotation database. Proteins were identified based on sequence alignments and subsequently assigned to functional categories according to a modified version of the MapMan annotation system (Thimm et al., 2004). (B) Number of proteins assigned to the 20 major functional categories of the annotation system. The category ‘not assigned’ containing 20 140 proteins is not shown.

Energy requirements and changes in biomass composition during germination and seedling establishment

In order to monitor the specific metabolic adaptations required for using protein as a major seed storage compound we cultivated L. albus seedlings for 12 d and harvested the plant material at four different time points (0, 4, 8, 12 d after sowing; Fig. 2A). The first two samples represent early seedling development in the dark whereas in the later stages photosynthesis was already active. The total biomass (dry weight) of the plants remained constant over the first 12 d, but its composition changed drastically (Fig. 2B; Supplementary Dataset S2). In the dark phase during the first 4 d there was no significant change in protein content but a decrease in lipid content by 50% from 25.6 ± 2.5 to 12.4 ± 1.3 mg plant⁻¹. We measured the respiration rate of the seedlings to estimate the total energy demand for early seedling establishment up to the onset of photosynthesis, which has to be covered exclusively by storage compound metabolism (Fig. 2C). An average individual seedling consumed a total amount of ~224 µmol O₂ during the first 4 d, which would correspond to the complete oxidation of 2.5 mg oleic acid (the most abundant fatty acid in L. albus seeds; Gdala and Buraczewska, 1996) or 6.7 mg glucose (Supplementary Dataset S2). Thus, the energy required for seedling establishment can easily be provided by lipid and/or carbohydrate degradation. A decrease in total protein content could only be detected at later developmental stages (8–12 d), and at the same time free amino acids strongly accumulated. By far the largest fraction of nitrogen (≥80%) was included in proteins and free amino acids, and the nitrogen demand for chlorophyll synthesis was comparatively low in the developing seedling (Fig. 2B). C/N analysis indicated that the sum of all nitrogen-containing compounds quantified individually (proteins, free amino acids, chlorophyll, ammonium, nitrate) corresponded very well to the total nitrogen content of the plant at the later developmental stages (8–12 d), but a fraction of 20% and 11% was not yet covered in the dry seed and the 4-day-old seedling, respectively (Fig. 2B; Supplementary Dataset S2).

Fig. 2.

Germination and seedling establishment in Lupinus albus. (A) Phenotype of representative plants during the first 12 d of growth. Yellow boxes mark sample groups. (B) Biomass and composition of the seeds/seedlings at 0, 4, 8, and 12 d after sowing. The total nitrogen content was analysed using a C/N-analyser and compared with the sum of calculated nitrogen contents of proteins, free amino acids (AA), and chlorophyll (Chl). Data presented are means ±SD (n=5). (C) Respiration rate of Lupinus albus during four phases (P_I–IV) of germination and seedling development in the first 4 d after sowing. P_I, initial onset of oxygen consumption at low rates; P_II, seed swelling and preparation for germination; P_III, completion of germination (blue box; germination=testa rupture); P_IV, early seedling establishment. Data presented are means ±SD (n=4).

Proteome remodeling from seed to seedling

In order to gain a more precise insight into the metabolic transformations during seedling development we analysed the quantitative composition of the L. albus proteome (Fig. 3; Supplementary Dataset S3). On the basis of our manually curated annotation table, we were able to assign 87% of the seed protein content to functional categories including the large fractions of storage proteins, late embryogenesis abundant (LEA) proteins, and structural proteins of the storage oil bodies (oleosins) (Fig. 3; Supplementary Fig. S1). Twelve days after imbibition the proteome of the seedling was more diverse and Rubisco subunits had become the proteins with the highest abundance (Fig. 3B, C). During germination and initial seedling establishment in the dark, LEA proteins rapidly decreased by 88% (Fig. 3D). Asparagine synthetase as well as the enzymes involved in the glyoxylate cycle and gluconeogenesis showed the highest relative increase in protein content (log₂ fold change: 5.6–6.2; Supplementary Dataset S3). A distinct peak in phytochrome A protein abundance was visible at day 4 shortly before emergence of the cotyledons from the soil (Fig. 3D). The most pronounced proteome remodeling occurred during development of the first true leaves (day 4–8) including massive degradation of storage proteins and synthesis of the photosynthetic apparatus (Fig. 3; Supplementary Fig. S1). Oleosins were completely degraded at this stage.

Fig. 3.

Quantitative composition of the Lupinus albus proteome during germination and seedling establishment. (A) Total protein content and mass fractions covered by the major functional categories at day 0, 4, 8, and 12 after sowing. Mass fractions [protein abundances obtained by MS (intensity-based absolute quantification), multiplied by protein molecular mass] for all individual proteins are provided in Supplementary Dataset S3. (B) Proteomap illustrating the quantitative composition of the L. albus proteome at day 0 and 12 after sowing. Proteins are shown as polygons whose sizes represent their mass fractions; see (A). Proteins involved in similar cellular functions are arranged in adjacent locations and visualized by colors. Letters correspond to the major proteins detected in the most prominent bands after SDS-PAGE. (C) Glycine-SDS-PAGE of proteins extracted from 5 mg lyophilized lupin sample stained with colloidal Coomassie blue. The most prominent bands (marked by letters) were analysed by MS (Supplementary Dataset S3). Major proteins: A, G: oleosin; B, C: lipoxygenase; D, F: late embryogenesis abundant protein; E: cupin; H: conglutin; I: ribulose bisphosphate carboxylase large chain; J: ribulose bisphosphate carboxylase small chain. (D) Protein abundance profiles of selected functional categories at day 0, 4, 8, and 12 after sowing. Numbers in parentheses indicate the number of different proteins detected in the respective category; LEA, late embryogenesis abundant. Abundance profiles for additional functional categories as well as information on the calculation method are provided in Supplementary Dataset S3.

Amino acid metabolism during germination and seedling establishment

A large fraction of the amino acids released by proteolysis is stored in the free amino acid pool leading to a shift between the free and protein-bound amino acid pools (Fig. 4A; Supplementary Dataset S4). Asparagine strongly accumulated in the free pool (from 6 ± 2 to 1198 ± 150 µmol g⁻¹ dry weight) and most of the other amino acids also significantly increased (2.3–134-fold) with the exception of glutamate and arginine, which decreased during development (Fig. 4B; Supplementary Dataset S4). Most of the enzymes involved in amino acid metabolism increased in their abundance during germination and seedling establishment (Figs 3D, 4C; Supplementary Dataset S4). The most pronounced induction could be detected in the pathway required for nitrogen (re)assimilation into asparagine, in photorespiration, and in sulfur amino acid metabolism. Some pathways contain isoenzymes localizing to different subcellular compartments. Since to our knowledge localization studies have not been performed for enzymes involved in L. albus amino acid metabolism so far, assignments of subcellular compartments are based on sequence homologies to Arabidopsis (Supplementary Fig. S2).

Fig. 4.

Amino acid metabolism in Lupinus albus during germination and seedling establishment. (A) Quantitative composition of the free (orange) and protein-bound (blue) amino acid pools at day 0, 4, 8, and 12 after sowing. Amino acids are shown as polygons whose sizes represent the molar fractions. Free amino acid contents were quantified by HPLC, and the quantitative amino acid composition of the proteome was calculated on the basis of molar composition of the proteome (see Supplementary Dataset S4). (B) Free amino acid contents of the seeds/seedlings at day 0, 4, 8, and 12 after sowing. Data presented are means ±SD (n=5). (C) Relative protein abundance of amino acid metabolic pathways during germination and seedling establishment. The colored squares represent the means of all log₂ fold changes in the abundance of enzymes involved in the respective branch of the pathway at day 4, 8, and 12 compared with day 0. The full table of amino acid metabolism is provided in Supplementary Dataset S4 and Supplementary Fig. S3. 2OB, 2‐oxobutyrate; 2OG, 2‐oxoglutarate; 3PG, 3-phosphoglycerate; ACoA, Acetyl-CoA; AdoMet, S-adenosylmethionin; C1, C1-metabolism; CM, chorismate; ET, ethylene; FeS, iron–sulfur cluster; FU, fumarate; GABA, γ‐aminobutyric acid; Glce, glycolate; GX, glyoxylate; Hpyr, hydroxypyruvate; MA, malate; MoCo, molybdenum cofactor; n.d., not detected; OA, oxaloacetic acid; OG, oxoglutarate; Pyr, pyruvate; R5P, ribose-5-phosphate; SC, succinate; SSA, succinic semialdehyde.

Integration of fatty acid and amino acid metabolism

Storage lipids can either be converted to carbohydrates via the glyoxylate cycle and gluconeogenesis or serve as a substrate for mitochondrial ATP production (Fig. 5A). In both cases metabolites have to be cycled back to the peroxisomes to supply acceptor molecules for the acetyl groups from fatty acid β-oxidation. The enzymes involved in lipid breakdown and fatty acid β-oxidation as well as three steps of the glyoxylate cycle (isocitrate lyase, malate synthase, malate dehydrogenase) strongly increased in the developing L. albus seedlings. The abundance of phosphoenolpyruvate carboxykinase required for gluconeogenesis from fatty acid catabolism peaked at day 4 after sowing. However, the isoforms of citrate synthase and aconitase required for completing the glyoxylate cycle were decreased in their abundance whereas isoforms in other subcellular compartments were moderately induced (Fig. 5A; Supplementary Dataset S5). Since lupin seeds use proteins as a major storage compound we reinvestigated a potential contribution of amino acids to seedling carbohydrate and energy metabolism. A moderate but continuous increase in relative protein abundance over the entire 12-day period studied was detectable for pyruvate orthophosphate dikinase (PPDK), an enzyme required for recovering carbon from pyruvate produced by amino acid catabolism (Fig. 5D; Eastmond et al., 2015). The total protein content of the seedlings did not change significantly during the initial phase of development prior to emergence from the soil (Fig. 2B), but interestingly there were clear differences in the quantitative amino acid composition of the proteome. The protein set-up of the seedling at day 4 had a lower content of glycine, alanine, and threonine than that of the dry seed while the fractions of the other 17 amino acids in the proteome remained fairly constant (Fig. 5B; Supplementary Dataset S5). By far the major source of glycine, alanine, and threonine release were two categories of seed proteins, LEA proteins and oleosins (Fig. 5C). These proteins are of high abundance (6–10%) in the dry seed, they were degraded early during germination and seedling establishment, and they are specifically rich in glycine, threonine, and alanine (Fig. 3; Supplementary Datasets S3, S5). Strikingly, transamination of glycine provides glyoxylate, and threonine can directly be converted to glycine via aldol cleavage (Hildebrandt et al., 2015). Thus, the unusual amino acid composition of specific seed proteins might provide an additional source of glyoxylate to support fatty acid metabolism during germination (Fig. 5D). The enzymes required for the production of glyoxylate from glycine and the conversion of threonine to glycine were strongly induced in the developing lupin seedlings. The same trend was visible for subsequent reaction steps metabolizing glyoxylate and acetyl-CoA to phosphoenolpyruvate as a substrate for gluconeogenesis (malate synthase, malate dehydrogenase, phosphoenolpyruvate carboxykinase). Transamination of alanine produces pyruvate, which can either be phosphorylated to phosphoenolpyruvate by PPDK for gluconeogenesis or serve as an amino group acceptor during transamination of glycine to glyoxylate. Via glutamate and glutamine, the ammonium can then be transferred to asparagine for storage and transport. Our datasets so far do not distinguish between the different tissues of the developing seedling. In order to test, whether the enzymes required for a combined metabolism of amino acids and fatty acids to carbohydrates were induced in the same tissue, we analysed the proteome of the cotyledons and the hypocotyl of 4-day-old seedlings separately (Supplementary Dataset S6; Supplementary Fig. S5). All the enzymes involved in this postulated combined pathway increased in abundance in both tissues compared with the dry seed with a more pronounced induction in the cotyledons. In contrast, the amino acid synthesis pathways increased more strongly in the growing hypocotyl (Supplementary Dataset S6; Supplementary Fig. S5).

Fig. 5.

Potential integration of fatty acid and amino acid metabolism during germination and seedling establishment. (A) Relative protein abundance of glyoxylate cycle enzymes. The colored squares represent the means of log₂ fold changes in the abundance of tissue specific isoforms at day 4, 8, and 12 compared with day 0. The full dataset is provided in Supplementary Dataset S5. (B) Protein-bound contents of glycine, alanine, and threonine [µmol g⁻¹ dry weight (DW)] in seeds and 4-day-old seedlings. A complete list of free and protein-bound amino acid contents at all time points is provided in Supplementary Dataset S4. (C) Total amount of glycine, alanine, and threonine released by protein degradation within 4 d after germination. The two major categories (late embryogenesis abundant proteins and oleosins) are shown individually whereas all other categories have been combined. (D) Relative protein abundance of enzymes required for a combined metabolism of amino acids and fatty acids to carbohydrates. The colored squares represent the means of log₂ fold changes in the abundance of tissue specific isoforms at day 4, 8, and 12 compared with day 0. The full dataset is provided in Supplementary Dataset S5. Assignments of subcellular compartments are based on sequence homologies to Arabidopsis (Supplementary Fig. S4). ACO, aconitase; AlaAT, alanine aminotransferase; CI, citrate; CSY, citrate synthase; cy, cytosolic; Eno, enolase; GX, glyoxylate; GXAT, glyoxylate aminotransferase; ICL, isocitrate lyase; MA, malate; MDH, malate dehydrogenase; ME, malic enzyme; MS, malate synthase; mt, mitochondrial; OA, oxaloacetate; PEP, phosphoenolpyruvate; PEPCK, phosphoenolpyruvate carboxykinase; PK, pyruvate kinase; pl, plastidic; PPDK, pyruvate; phosphate dikinase; px, peroxisomal; Pyr, pyruvate; ThrAld, l-threonine aldolase.

Discussion

Energy and carbohydrate metabolism of Lupinus albus during germination

During germination seeds exclusively rely on their storage compounds to supply substrates for ATP production as well as precursors for the synthesis of structural elements and cellular components. At this stage their energy metabolism is chemotrophic in contrast to the characteristic photoautotrophic strategy of plants at later developmental stages. The obvious idea for seeds that use proteins as major storage compounds such as legumes would be to oxidize amino acids as alternative respiratory substrates and to convert them into carbohydrates via the glyoxylate cycle and gluconeogenesis. Both pathways are present in plants and have been shown to be relevant for seedling establishment in Arabidopsis (Eastmond et al., 2015; Henninger et al., 2022). However, our quantitative data indicate that in Lupinus albus there is no net decrease in total protein content up to a stage when the seedling emerges from the soil and develops first true leaves (Fig. 2). Nevertheless, the composition of the proteome clearly changes within the first 4 d of germination and seedling development with a strong decrease in oleosins and LEA proteins and an induction of metabolic enzymes (Fig. 3). We detected only a moderate increase in the pathways characteristic for an energy deficiency response such as lysine, tyrosine, and branched-chain amino acid degradation as well as PPDK, required for efficient recovery of carbon from storage proteins (Eastmond et al., 2015; Pedrotti et al., 2018). In contrast, the total lipid content dropped by more than 50% during the initial phase of seedling establishment and the pathways catalysing fatty acid catabolism and gluconeogenesis were strongly induced. These results agree with previous findings on storage lipid metabolism in L. albus that already demonstrated a pivotal function during germination (Borek et al., 2015). Electron micrographs of white lupin cotyledons show a strong decrease in the number of oil bodies and more but smaller protein storage vesicles after 4 d of in vitro cultivation (Borek et al., 2011). Our estimation based on seedling respiration rates indicates that the lower limit for covering the ATP demand until the onset of photosynthesis would be ~10.5 mg lipid g⁻¹ dry weight (2.5 mg per seed). Thus, in order to secure vigorous seedling establishment, breeding strategies should not aim at low fat varieties with a lipid content below 1.5–2%. Surprisingly, no increase in peroxisomal citrate synthase or cytosolic aconitase was detectable. Citrate synthase is essential for fatty acid respiration when citrate is used as the carbon transport form from the peroxisomes to the mitochondria, which has been postulated for the oilseeds Arabidopsis and Helianthus annuus (Raymond et al., 1992; Pracharoenwattana et al., 2005). Both, citrate synthase and aconitase are required for regeneration of the initial acetyl acceptor in the glyoxylate cycle (Baker et al., 2006; Pracharoenwattana and Smith, 2008). Citrate synthase activity can be modified by redox regulation and allosteric effectors (Schmidtmann et al., 2014; Nishio and Mizushima, 2020). Thus, an increase in enzyme activity could be achieved without a concomitant increase in protein abundance via post-translational mechanisms. However, our results indicate that in germinating seeds of L. albus, the amino acids glycine and threonine, which are strikingly enriched in two classes of seed specific proteins (oleosins and LEA proteins), might provide an alternative source of glyoxylate (Fig. 5). Oleosins are the predominant lipid droplet coat proteins and their degradation is coordinated with triacylglycerol mobilization (Traver and Bartel, 2023). Thus, according to our hypothesis, fatty acids would be delivered together with the suitable amino acids for their efficient utilization during germination. This integration of fatty acid and amino acid metabolism would utilize a linear flux mode through parts of the glyoxylate cycle to combine stored carbon from lipids and proteins into gluconeogenesis. A potential next step involves conducting computational analysis of a network representation of L. albus primary metabolism by means of flux-balance techniques. This approach would allow the making of quantitative predictions of metabolic flux modes during the different stages of early seedling establishment. Since photosynthesis has not yet been established during these developmental phases, the system operates in a chemotroph mode. As such, all metabolic processes are driven by internal energy and matter conversions, which will enable us to confidentially constrain the model based on experimental measurements. Such a model has the potential to elucidate the combined roles of fatty acid degradation and amino acid metabolism to support fluxes into gluconeogenesis and to highlight constraints and capacities associated with the use of proteins as major seed storage compounds. Since we detected no decrease in net protein content during early seedling development, at least part of the carbohydrates produced in the cotyledons is most likely used for amino acid synthesis and the production of a functional protein set-up in the growing seedling.

Nitrogen resource allocation in the Lupinus albus seedling

Amino acids and proteins contain by far the largest fraction of nitrogen in white lupin seeds and seedlings (≥80%). Additional nitrogen containing compounds such as chlorophyll and nucleotides quantitatively require no major input of nitrogen resources. However, about 20% of the seed nitrogen content could not be assigned to a substance class by our approach. This most likely includes non-protein amino acids and polyamines, which are present in dry lupin seeds and might have a protective function against abiotic stress (Aniszewski et al., 2001; Bano et al., 2020). Asparagine strongly accumulated in the lupin seedlings with a peak after the onset of photosynthesis (Fig. 4). Asparagine is a suitable storage and transport form of reduced nitrogen within the plant since it has a favorable C/N ratio, little net charge, and low reactivity under physiological conditions (Lea et al., 2007). It also serves as the major end product of nitrogen fixation in lupins and accounts for 60–80% of the total amino acids in nodulated roots, leaves, and developing seeds of nitrogen-fixing L. albus (Pate et al., 1981). During early seedling establishment, the induction of asparagine synthesis preceded a net decrease in total protein content indicating an incorporation of nitrogen from alternative sources such as degradation of specialized defense compounds in the seeds. The carbon skeletons for early asparagine synthesis might be at least partially provided by storage lipid catabolism, which would explain the observed carbon flux from lipids to amino acids (Borek et al., 2003; Borek and Ratajczak, 2010). Asparagine accumulation continued and even accelerated at later developmental stages (8 to 12 d after sowing) when the seedling had already emerged from the soil and photosynthesis was active, but at this stage was clearly connected to proteolysis and a massive shift of amino acids from the protein bound to the soluble pool. A major function of this shift is most likely remobilization and transport of nitrogen to the growing tissues (Lea et al., 2007). Since the free amino acid concentration in the seedlings reached almost 200 mM, this process might in addition be relevant for osmoregulation. In nitrogen-poor soils the lupin seedlings rely on the nitrogen stored in amino acids until the symbiosis with N₂-fixing rhizobia has been established, which is usually achieved within 2–3 weeks after germination. Our results indicate that supply of nitrogen derived from storage proteins from the cotyledons to the growing seedling peaks during the second week of development, which would suggest a pivotal role of the cotyledons not only for successful germination but also for subsequent establishment on nitrogen-poor soil. Successfully populating these habitats requires the large internal nitrogen store provided by storage proteins in lupins and other legume plants to meet the demands of the growing seedling up to the point when rhizobial symbiosis is fully functional.

Conclusions

Although classified as a protein crop, Lupinus albus does not use amino acids as a primary resource during germination and early seedling establishment. The total protein content decreases substantially only after emergence of the seedling from the soil. The major advantage of storing large quantities of protein in the seeds would therefore be to provide the nitrogen resources required for successfully populating nitrogen deprived habitats for a prolonged period (2–3 weeks) after germination until the symbiosis with N₂-fixing rhizobia has been established. However, our results also point to the metabolic potential of integrating fatty acid and amino acid catabolism at the level of malate synthase during germination. Conversion of glycine to glyoxylate would allow the integration of acetyl-CoA from fatty acid β-oxidation into oxaloacetate for gluconeogenesis without the full glyoxylate cycle. The physiological relevance of this combined pathway during germination will have to be elucidated.

Supplementary data

The following supplementary data are available at JXB online.

Fig. S1. Quantitative composition of the Lupinus albus proteome during germination and seedling establishment.

Fig. S2. Multiple sequence alignments of Arabidopsis thaliana and Lupinus albus isoenzymes of amino acid metabolic pathways localized to different subcellular compartments.

Fig. S3. Amino acid metabolism in Lupinus albus.

Fig. S4. Multiple sequence alignments of Arabidopsis thaliana and Lupinus albus isoenzymes in the glyoxylate cycle localized to different subcellular compartments.

Fig. S5. Potential integration of fatty acid and amino acid metabolism in Lupinus albus cotyledons and hypocotyl during seedling establishment.

Dataset S1. Manually curated Lupinus albus protein annotation database.

Dataset S2. Composition and respiration rate of Lupinus albus seeds and seedlings (dataset corresponding to Fig. 2).

Dataset S3. Proteome of Lupinus albus seeds and seedlings (dataset corresponding to Fig. 3).

Dataset S4. Amino acid metabolism in Lupinus albus seeds and seedlings (dataset corresponding to Fig. 4).

Dataset S5. Potential integration of fatty acid and amino acid metabolism during germination and seedling establishment (dataset corresponding to Fig. 5).

Dataset S6. Proteome of Lupinus albus cotyledons and hypocotyl 4 d after imbibition.

Acknowledgements

We thank Hashani Amarasinghe for preliminary work on this project and Dagmar Lewejohann for skillful technical assistance. Constructive discussions with Marco Herde and Holger Eubel are highly appreciated. We also thank Dr Alberto Andrino de la Fuente for support during C/N analysis.

Author contributions

TMH and HPB initiated the project; TMH, CA, and BH designed the research; BH and TMH developed the Lupinus albus protein annotation database; BH performed and evaluated the shotgun proteomics experiments; JH performed the experiments for individual analysis of cotyledons and hypocotyl; CA performed all the other experiments; TMH, CA, and BH analysed the data; NT contributed to pathway evaluations; TMH wrote the paper with support from CA, BH, NT, and HPB. TMH agrees to serve as the author responsible for contact and ensures communication.

Conflict of interest

The authors have no conflicts of interest to declare.

Funding

Research in TMH’s lab and NT’s group is funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy—EXC-2048/1—project ID 390686111.

Data availability

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE partner repository (Perez-Riverol et al., 2022) with the dataset identifier PXD046789.

Author notes

These authors contributed equally to this work.