The present invention relates to a method for identifying at least one candidate metabolic conversion step, the modulation of which causes a change of the amount of at least one metabolite of interest in a plant, said method comprising: (a) establishing a stoichiometric network model for the metabolism of the plant cell, plant or plant part including at least one synthesis pathway for the metabolite(s) of interest; (b) calculating a set of all elementary modes for said stoichiometric network model and selecting a set of elementary modes comprising, preferably all, elementary modes for which the flux to the metabolite(s) of interest is not zero; (c) calculating for each elementary mode of step (b) the correlation between (i) the metabolite(s) of interest and (ii) each partial reaction; and (d) identifying at least one candidate metabolic conversion step, the modulation of which causes a change of the amount of at least one metabolite of interest in a plant cell, plant or plant part, based on the correlation calculated in step (c). The present invention further relates to a method for identifying at least two candidate metabolic conversion steps, the combined modulation of which causes a change of the amount of at least one metabolite of interest in a plant. The present invention also relates to a method for generating a plant cell, plant or plant part, which produces a changed amount of a metabolite of interest, and to plants producible with said method, as well as to a method for the

manufacture of a metabolite of interest comprising obtaining the metabolite of interest from said plant. Moreover, the present invention relates to devices and data carriers related to the present invention and plant cell, plant or plant part comprising expressible constructs selected with the methods of the present invention.

Higher plants are the major source of food and feed, cereal seeds being the basis of nutrition for a large percentage of the human population. However, the composition of cereal seeds, e.g., rice seeds, is not optimal for human and livestock nutrition, since they often comprise suboptimal amounts of compounds essential for animals and man like, e.g., vitamins, amino acids, or unsaturated fatty acids. Moreover, increasing the yield of major crop plants is a goal of outmost importance, in particular for developing countries. Thus, a major goal of genetic breeding is to develop transgenic solutions to improve plant yield traits. An alternative to current-state-of-the-art gene discovery approaches is the growing field of metabolic modeling approaches.

The metabolism of an organism of interest can, in principle, be modelled in silico by establishing a metabolic network model for said organism, e.g. a stoichiometric network model (e.g.

Grafahrend-Belau E., Schreiber, F., Koschutzki D., Junker B.H. (2009) Plant Physiology. 149(1 ), 585-598). This, however, requires profound knowledge on the metabolism of said organism. On the basis of such a model, the flow of metabolites through the network can be calculated in a constraint-based modelling approach like flux-balance analysis for steady state analysis (e.g. Orth J.D., Thiele I., Palsson B.O. (2010) Nature Biotechnology. 28(3), 245-248) or like MOMA (Minimization Of Metabolic Adjustment; Segre D., Vitkup D., Church G.M. (2002) PNAS. 99(23), 151 12-151 17) or ROOM (Regulatory On / Off Minimization; Shlomi T., Berkman O., Ruppin E. (2005) PNAS. 102(21 ), 7695-7700) for simulating the distortions within the network caused by the loss of a metabolic conversion step, e.g., by a knockout. Further iterative optimization of data within a metabolic model was described in the literature (e.g. Masakapalli et al. (2010) Plant Physiol. (152): 602).

There are various public resources available for collection of biochemical data for plant metabolism needed for the reconstruction of different types of metabolic models. The biochemistry of plant metabolism, especially the primary metabolism, has been studied for many years and is reviewed, in principle, in many biochemistry text books. In addition, there are several publicly available databases and online resources existing, that contain biochemical data about metabolic reactions and its occurrence and localization in plants (see Table 1 ). Table 1 : Different data sources for biochemical information about plant metabolism. The resources are characterized by reaction properties needed for the reconstruction of plant- specific metabolic models.

The following databases contain almost all necessary biochemical information for plant-specific metabolic models: MetaCrop (Grafahrend-Belau et al., MetaCrop: a detailed database for crop plant metabolism. Nucleic Acids Research, 36 (S1 ):D954-D958, 2008), PlantCyc (Plant Metabolic Network (PNM), 2012, Internet only) and KEGG (Kanehisa and Goto, Kegg: Kyoto encyclopedia of genes and genomes. Nucleic Acids Research, 28(1 ):27-30, 2000.). All of them support the graphical entrance via organism or pathway specific metabolic network maps whereas the first two contain only plant specific data. KEGG and PlantCyc are highly recommended for getting a system-wide introduction into metabolism: what pathways are present in plants and which reactions are involved. In comparison, MetaCrop is a hand-curated database, which contains additional information about reaction directionality and reaction's compartmental localization and their respective references. But MetaCrop does not contain all known metabolic pathways occurring in plants and therefore also BRENDA (Scheer et al., Brenda, the enzyme information system in 201 1. Nucleic Acids Research, 39 (suppl 1 ):D670- D676, 2010) is very useful by providing organism-specific references for all enzymatic reactions in almost all plant species, if available. The metabolic reconstruction of a model organism / tissue of interest typically contains detailed metabolic information about the metabolic processes and their metabolic connections of a target cell or tissue in the model plant. The metabolic processes are described by the respective pathways and the participating enzymes present in the target cell or organ, and each metabolic reaction is defined by its detailed reaction stoichiometry given by the amount of metabolites that are consumed ('substrates') and the metabolites that are formed ('products'). The metabolic model further contains transport reactions between different compartments and with the environment for exchange of essential nutrients or other important metabolites.

Based on the available biochemical information for the plant of interest, a metabolic model can be reconstructed in order to analyze the network structure, calculate feasible flux distributions or explore dynamic properties of the metabolic system. Several algorithms have been proposed that are able to propose knockout strategies for implementing an optimization of the production of a metabolite of interest, preferably while maintaining a suitable growth rate (see e.g. Burgard A.P., Pharkya P., Maranas CD. (2003) Biotechnology and Bioengineering. 84(6):647-657; Tepper N., Shlomi T. (2010) Bioinformatics. 26(4):536-543).

For a given metabolic reconstruction of, e.g., a bacterium, elementary modes characterize the model's structural properties and allow for denomination of any feasible pathway routes through the entire metabolic network (Guido Melzer et al., (2009) Flux Design: In silico design of cell factories based on correlation of pathway fluxes to desired prototype. BMC Systems Biology. 3, 120). By definition, an elementary mode is a minimal set of reactions that could operate in steady-state and cannot be decomposed in further modes conceptually. Thus, any steady-state flux distribution can be written as linear combination of one or more different elementary modes. It can be used to: (1 ) identify all futile cycles of the metabolic system; (2) determine all possible routes from a given substrate to a product; and (3) predict metabolic routes that produce the favored product with the highest molar yield.

Single gene targets (knock-out as well as over-expression) can be obtained from the set of all elementary modes using a correlation approach entitled Flux Design (Guido Melzer et al., loc. cit; Habib Driouch, Guido Melzer, Christoph Wittmann. (2012) Integration of in vivo and in silico metabolic fluxes for improvement of recombinant protein production. Metabolic Engineering. 14, 47-58). Given the full set of elementary modes, the correlation between a target reaction and each participating reaction of the metabolic reconstruction is determined in all modes and assessed for statistical relevance. In that respect, the statistical relevance is defined by a cut-off value for the regression coefficient R2 = 0.7, and only statistical valid genetic targets are considered further. Based on the amplitude of correlation, the genetic targets are grouped into potential over-expression targets (positive correlation) or knock-out targets (negative

correlation).

Despite the availability of the aforesaid methods, for the complex metabolism of plants, prediction of knockout or overexpression targets suitable for changing the concentration of a metabolite of interest is a challenge still today. Thus, there is a need for the reliable prediction of metabolic effects. The technical problem underlying the present invention could, thus, be seen as the provision of means and methods for making predictions of relevant metabolic effects and for, thereby, allowing to identify metabolic conversion steps in a metabolism for the production of a metabolite of interest. The technical problem is solved by the embodiments characterized in the claims and herein below.

Accordingly, the present invention relates to a method for identifying at least one candidate metabolic conversion step, the modulation of which causes a change of the amount of at least one metabolite of interest in a plant cell, plant or plant part, said method comprising:

(a) establishing a stoichiometric network model for the metabolism of the plant cell, plant or plant part including at least one synthesis pathway for the metabolite(s) of interest;

(b) calculating a set of all elementary modes for said stoichiometric network model and selecting a set of elementary modes comprising, preferably all, elementary modes for which the flux to the metabolite(s) of interest is not zero;

(c) calculating for each elementary mode of step (b) the correlation between (i) the metabolite(s) of interest and (ii) each partial reaction; and

(d) identifying at least one candidate metabolic conversion step, the modulation of which causes a change of the amount of at least one metabolite of interest in a plant cell, plant or plant part, based on the correlation calculated in step (c). As used in the following, the terms "have", "comprise" or "include" or any arbitrary grammatical variations thereof are used in a non-exclusive way. Thus, these terms may both refer to a situation in which, besides the feature introduced by these terms, no further features are present in the entity described in this context and to a situation in which one or more further features are present. As an example, the expressions "A has B", "A comprises B" and "A includes B" may both refer to a situation in which, besides B, no other element is present in A (i.e. a situation in which a solely and exclusively consists of B) and to a situation in which, besides B, one or more further elements are present in entity A, such as element C, elements C and D or even further elements. Further, as used in the following, the terms "preferably", "more preferably", "most preferably", "particularly", "more particularly", "specifically", "more specifically" or similar terms are used in conjunction with optional features, without restricting alternative possibilities. Thus, features introduced by these terms are optional features and are not intended to restrict the scope of the claims in any way. The invention may, as the skilled person will recognize, be performed by using alternative features. Similarly, features introduced by "in an embodiment of the invention" or similar expressions are intended to be optional features, without any restriction regarding alternative embodiments of the invention, without any restrictions regarding the scope of the invention and without any restriction regarding the possibility of combining the features introduced in such way with other optional or non-optional features of the invention.

The method for identifying at least one metabolic conversion step of the present invention, preferably, is an in-silico method. Thus, preferably, most or all of the steps of said method are performed in a computer-assisted mode. Moreover, said method may comprise further steps in addition to the ones explicitly mentioned. Specifically, step a) may, preferably, comprise the further step of generating and/or collecting data required to establish a stoichiometric network model for the plant in question or step d) may, preferably, contain the further steps of validating the metabolic conversion step by constructing and analyzing a plant comprising a modulation of said metabolic conversion step as described herein below. Preferably, said method comprises the further step of determining whether the metabolic conversion step identified in step (d) increases the metabolite of interest in the plant cell, plant or plant part by modulating the said metabolic conversion step in a plant cell, plant or plant part in vivo and, more preferably, comparing the content of said metabolite of interest in said plant cell, plant or plant part to a control plant cell, plant or plant part.

The term "metabolic conversion step", as used herein, relates to any chemical or physical modification of a compound comprised by a plant, plant part, plant organ, or plant cell.

Preferably, the metabolic conversion step is a chemical conversion of a compound into a chemically different compound. More preferably, the metabolic conversion step is an

enzymatically catalyzed chemical reaction. Most preferably, the metabolic conversion step is a chemical reaction catalyzed by a polypeptide having enzymatic properties expressed by the plant cell, i.e. an enzymatic conversion. It is to be understood that the term may refer to any conversion in the metabolism of a plant, including, e.g., anabolism, catabolism, and secondary metabolism. It is also to be understood that the term may also refer to the translocation or transport of a compound within the plant of the present invention. Preferably, included by the term metabolic conversion step are, thus, the transport of a compound in the xylem or phloem of a plant, or the transport from one cell compartment into another, preferably, over one or more membranes. Accordingly, the term "enzyme" as used herein, relates to a biological

macromolecule facilitating at least one metabolic conversion step as specified herein; thus, preferably, the term enzyme includes transporters, redox mediators, and the like. Thus, preferably, the terms enzymatic step, enzymatic conversion, and the like, relate to steps, conversions, etc. mediated by an enzyme as specified above.

As used herein, the term "plant" relates to a whole plant, a plant part, a plant organ, a plant tissue, or a plant cell. Thus, the term includes, preferably, seeds, shoots, stems, leaves, roots (including tubers), and flowers. Preferably, the term "plant" relates to a member of the clade Archaeplastida. Plants that are particularly useful in the methods of the invention include all plants which belong to the superfamily Viridiplantae, preferably Tracheophyta, more preferably Spermatophytina, most preferably monocotyledonous and dicotyledonous plants including fodder or forage legumes, ornamental plants, food crops, trees or shrubs selected from the list comprising Acer spp., Actinidia spp., Abelmoschus spp., Agave sisalana, Agropyron spp., Agrostis stolonifera, Allium spp., Amaranthus spp., Ammophila arenaria, Ananas comosus, Annona spp., Apium graveolens, Arachis spp, Artocarpus spp., Asparagus officinalis, Avena spp. (e.g. Avena sativa, Avena fatua, Avena byzantina, Avena fatua var. sativa, Avena hybrida), Averrhoa carambola, Bambusa sp., Benincasa hispida, Bertholletia excelsea, Beta vulgaris, Brassica spp. (e.g. Brassica napus, Brassica rapa ssp. [canola, oilseed rape, turnip rape]), Cadaba farinosa, Camellia sinensis, Canna indica, Cannabis sativa, Capsicum spp., Carex elata, Carica papaya, Carissa macrocarpa, Carya spp., Carthamus tinctorius, Castanea spp., Ceiba pentandra, Cichorium endivia, Cinnamomum spp., Citrullus lanatus, Citrus spp., Cocos spp., Coffea spp., Colocasia esculenta, Cola spp., Corchorus sp., Coriandrum sativum, Corylus spp., Crataegus spp., Crocus sativus, Cucurbita spp., Cucumis spp., Cynara spp., Daucus carota, Desmodium spp., Dimocarpus longan, Dioscorea spp., Diospyros spp., Echinochloa spp., Elaeis (e.g. Elaeis guineensis, Elaeis oleifera), Eleusine coracana, Eragrostis tef,

Erianthus sp., Eriobotrya japonica, Eucalyptus sp., Eugenia uniflora, Fagopyrum spp., Fagus spp., Festuca arundinacea, Ficus carica, Fortunella spp., Fragaria spp., Ginkgo biloba, Glycine spp. (e.g. Glycine max, Soja hispida or Soja max), Gossypium hirsutum, Helianthus spp. (e.g. Helianthus annuus), Hemerocallis fulva, Hibiscus spp., Hordeum spp. (e.g. Hordeum vulgare), Ipomoea batatas, Juglans spp., Lactuca sativa, Lathyrus spp., Lens culinaris, Linum

usitatissimum, Litchi chinensis, Lotus spp., Luffa acutangula, Lupinus spp., Luzula sylvatica, Lycopersicon spp. (e.g. Lycopersicon esculentum, Lycopersicon lycopersicum, Lycopersicon pyriforme), Macrotyloma spp., Malus spp., Malpighia emarginata, Mammea americana,

Mangifera indica, Manihot spp., Manilkara zapota, Medicago sativa, Melilotus spp., Mentha spp., Miscanthus sinensis, Momordica spp., Morus nigra, Musa spp., Nicotiana spp., Olea spp., Opuntia spp., Ornithopus spp., Oryza spp. (e.g. Oryza sativa, Oryza latifolia), Panicum miliaceum, Panicum virgatum, Passiflora edulis, Pastinaca sativa, Pennisetum sp., Persea spp., Petroselinum crispum, Phalaris arundinacea, Phaseolus spp., Phleum pratense, Phoenix spp., Phragmites australis, Physalis spp., Pinus spp., Pistacia vera, Pisum spp., Poa spp., Populus spp., Prosopis spp., Prunus spp., Psidium spp., Punica granatum, Pyrus communis, Quercus spp., Raphanus sativus, Rheum rhabarbarum, Ribes spp., Ricinus communis, Rubus spp., Saccharum spp., Salix sp., Sambucus spp., Secale cereale, Sesamum spp., Sinapis sp., Solanum spp. (e.g. Solanum tuberosum, Solanum integrifolium or Solanum lycopersicum), Sorghum bicolor, Spinacia spp., Syzygium spp., Tagetes spp., Tamarindus indica, Theobroma cacao, Trifolium spp., Tripsacum dactyloides, Triticosecale rimpaui, Triticum spp. (e.g. Triticum aestivum, Triticum durum, Triticum turgidum, Triticum hybernum, Triticum macha, Triticum sativum, Triticum monococcum or Triticum vulgare), Tropaeolum minus, Tropaeolum majus, Vaccinium spp., Vicia spp., Vigna spp., Viola odorata, Vitis spp., Zea mays, Zizania palustris, Ziziphus spp., amongst others. Preferably, the plant cell, plant or plant part is a rice cell, rice plant, rice plant part, or rice seed.

The term "modulation", as used herein, relates to a change of a stoichiometric or kinetic parameter of a metabolic conversion step from the corresponding parameter found under physiological conditions in a plant cell, plant, or plant part. Physiological conditions are those, which can be observed without modulation of the step. Preferably, the said change is a statistically significant change. The change may be an increase or a decrease. The modulation of a metabolic conversion step and thus, the deviation of a stoichiometric or kinetic parameter can, e.g., be achieved by deleting or mutating a gene encoding a subunit of an enzyme complex catalyzing a partial reaction of an enzymatic step, such that the amount or identity of the final product is altered. A deviation of a kinetic parameter can, e.g., be achieved by deleting the gene coding for an enzyme catalyzing the metabolic conversion step in question, such that the reaction velocity is reduced to the reaction velocity of the uncatalyzed conversion, which is, preferably, zero. Preferably, modulation encompasses decreasing or increasing the activity of an enzyme catalyzing said metabolic conversion. More preferably, modulation is abolishing the activity of an enzyme catalyzing said metabolic conversion step. Preferably, modulation is achieved by modulation of gene expression. Thus, preferably, the term "modulation" means in relation to expression or gene expression, a process or state in which the level of gene expression is changed by said process or state in comparison to the control plant, wherein the expression level may be increased or decreased. The original, unmodulated expression may be of any kind of expression of a structural RNA (rRNA, tRNA) or mRNA with subsequent translation. The term "modulating the activity" in relation to expression or gene expression shall mean any change of the expression of the gene, leading to an altered concentration of the corresponding polynucleotides and/or encoded proteins in the cell. Modulation of an enzymatic activity can be achieved by a variety of methods well known in the art.

Preferably, the modulation is an activation, i.e., preferably, a modulation increasing the activity of an enzyme catalyzing said metabolic conversion. Activation can, preferably, be achieved by application of an activator for the enzyme. More preferably, activation is mediated by introducing into the plant cell one or more molecules of an enzyme catalyzing said metabolic conversion step. Said enzyme may, preferably, be autologous or, more preferably, heterologous. Said enzyme, may be a wildtype enzyme or a mutated enzyme with an increased activity or, preferably, with reduced feedback inhibition. Preferably, modulation comprises introducing a mutated enzyme with increased activity and/or decreased feedback inhibition in addition to the enzyme present in the cell, e.g. by introducing a gene encoding said mutated enzyme with increased activity and/or decreased feedback inhibition due to a mutation, e.g. a point mutation, into the plant cell; or, more preferably, modulation comprises introducing a mutated enzyme with increased activity and/or decreased feedback inhibition replacing the enzyme present in the plant cell, e.g. by replacing all copies of the gene encoding the cellular enzyme by copies comprising a mutation, e.g. a point mutation; or by replacing one or more copies of the gene encoding the cellular enzyme by one or more copies comprising a mutation, e.g. a point mutation, and by inactivating all further genes encoding said cellular enzyme potentially present in said plant cell. Also, the enzyme may be introduced into the plant cell as a polypeptide or, more preferably, as an expressible gene.

The term "expression" or "gene expression" relates to transcription of a specific gene or specific genes or a specific genetic construct. The term "expression" or "gene expression" in particular means the transcription of a gene or genes or genetic construct into structural RNA (rRNA, tRNA) or mRNA with or without subsequent translation of the latter into a protein. The process includes transcription of DNA and processing of the resulting mRNA product. The term

"increased expression" or "overexpression" as used herein means any form of expression that is additional to the original wildtype expression level. Methods for increasing expression of genes or gene products are well documented in the art and include, for example, overexpression driven by appropriate promoters, the use of transcription enhancers or translation enhancers. Isolated nucleic acids which serve as promoter or enhancer elements may be introduced in an appropriate position (typically upstream) of a non-heterologous form of a polynucleotide so as to upregulate expression of a nucleic acid encoding the polypeptide of interest. For example, endogenous promoters may be altered in vivo by mutation, deletion, and/or substitution (see, Kmiec, US 5,565,350; Zarling et al., W09322443), or isolated promoters and/or, preferably, one or more enhancer elements, may be introduced into a plant cell in the proper orientation and distance from a gene of the present invention so as to control the expression of the gene. In another preferred embodiment, modulated expression is achieved through a modulation of the transcription or transcript stability of the corresponding endogenous gene, modulation of transcription of an endogenous gene meaning that the endogenous transcriptional control of the gene is influenced, either directly through mutation of gene control elements, e.g., preferably, the promotor, or indirectly by changing or de novo expression of endogenous or heterologous transcription factors, which bind to the gene control elements and which thereby modify expression. Heterologous transcription factors can be obtained from other organisms, preferably from other plants, or can be specifically designed to bind to target sequences and to activate transcription. Preferably, in order to provide or increase gene expression, a constitutive promoter is used. A constitutive promoter, preferably, is a promoter that is transcriptionally active during most, more preferably during all, phases of growth and development and under most, more preferably all, environmental conditions. Preferably, the constitutive promoter is active in at least one cell, tissue or organ; more preferably, the constitutive promoter is active in all cells, tissues and organs of a plant. Table 2 below shows preferred constitutive promoters.

Table 2: Examples of constitutive promoters

Gene Source Reference

Actin McElroy et al, Plant Cell, 2: 163-171 , 1990

HMGP WO 2004/070039

CAMV 35S Odell et al, Nature, 313: 810-812, 1985

CaMV 19S Nilsson et al., Physiol. Plant. 100:456-462, 1997

GOS2 de Pater et al, Plant J Nov;2(6):837-44, 1992, WO 2004/065596

Ubiquitin Christensen et al, Plant Mol. Biol. 18: 675-689, 1992

Rice cyclophilin Buchholz et al, Plant Mol Biol. 25(5): 837-43, 1994

Maize H3 histone Lepetit et al, Mol. Gen. Genet. 231 :276-285, 1992

Alfalfa H3 histone Wu et al. Plant Mol. Biol. 1 1 :641 -649, 1988

Actin 2 An et al, Plant J. 10(1 ); 107-121 , 1996

34S FMV Sanger et al., Plant. Mol. Biol., 14, 1990: 433-443

Rubisco small subunit US 4,962,028

OCS Leisner (1988) Proc Natl Acad Sci USA 85(5): 2553

SAD1 Jain et al., Crop Science, 39 (6), 1999: 1696

SAD2 Jain et al., Crop Science, 39 (6), 1999: 1696

nos Shaw et al. (1984) Nucleic Acids Res. 12(20):7831 -7846

V-ATPase WO 01/14572

Super promoter WO 95/14098

G-box proteins WO 94/12015 More preferably, however, a regulated or regulable promoter is used. A regulable promoter, preferably, is a promoter, which can be regulated by administration of an external signal to the plant cell, e.g. a chemical compound. Preferred regulable promoters are, e.g. a tet-promoter, a dexamethasone-inducible glucocorticoid receptor (GR)-based promoter, an estradiol-inducible estrogen receptor (ER)-based promoter, an insecticide-inducible ecdysone receptor (EcR)- based promoter, an ethanol-inducible AlcR-based promoter, a copper-inducible Acel-based promoter (for review see Padidam (2003), Curr. Op. Plant Biol. 6:169 and Tang et al. (2004), Plant Sci. 166:827) or the like. A regulated promoter, preferably, is a promoter, which is regulated by the plant cell, e.g. according to diurnal rhythms or according to day/night cycles, or in specific tissues. Examples of promoters regulated according to day/night cycles and/or in specific tissues are known in the art (e.g. Biasing et al. (2005), Plant Cell 17(12):3257; Wang et al. (2002), Plant Sci. 163:273; Lin et al. (2004), DNA Seq. 15(4):269; Sato et al. (1996), Proc. Natl. Acad. Sci. USA, 93(15):81 17). As will be understood by the skilled person, promoters are preferably selected to match the network model used to identify the respective candidate conversion step; e.g. genes encoding enzymes catalyzing conversion steps identified in a model of photosynthetic metabolism are preferably expressed from diurnally regulated promoters; and genes encoding enzymes catalyzing conversion steps identified in a model of seed metabolism are preferably expressed from seed-specific promoters. A "seed-specific promoter", preferably, is a promoter transcriptionally active predominantly in seed tissue; thus, preferably, a seed-specific promoter is a promoter being at least 10fold, more preferably at least 25fold, most preferably at least 10Ofold more active in seed tissue than in any other plant tissue. More preferably, a seed-specific promoter is a promoter transcriptionally active exclusively in seed tissue. Preferably, the seed-specific promoter is active during seed development and/or during germination. Preferably, the seed-specific promoter is endosperm- and/or aleurone- and/or embryo-specific. Preferred examples of seed-specific promoters

(endosperm/aleurone/embryo specific) are shown in Tables 3 to Table 6 below. Further preferred examples of seed-specific promoters are given in Qu and Takaiwa (Plant Biotechnol. J. 2, 1 13-125, 2004). Table 3: Examples of seed specific promoters

Gene source Reference

seed-specific genes Simon et al., Plant Mol. Biol. 5: 191 , 1985;

Scofield et al., J. Biol. Chem. 262: 12202, 1987.;

Baszczynski et al., Plant Mol. Biol. 14: 633, 1990.

Brazil Nut albumin Pearson et al., Plant Mol. Biol. 18: 235-245, 1992.

legumin Ellis et al., Plant Mol. Biol. 10: 203-214, 1988.

glutelin (rice) Takaiwa et al., Mol. Gen. Genet. 208: 15-22, 1986;

Takaiwa et al., FEBS Letts. 221 : 43-47, 1987.

zein Matzke et al Plant Mol Biol, 14(3):323-32 1990

napA Stalberg et al, Planta 199: 515-519, 1996.

wheat LMW and HMW glutenin- Mol Gen Genet 216:81 -90, 1989; NAR 17:461 -2, 1989 1

wheat SPA Albani et al, Plant Cell, 9: 171 -184, 1997

wheat α, β, γ-gliadins EMBO J. 3:1409-15, 1984 barley Itr1 promoter Diaz et al. (1995) Mol Gen Genet 248(5):592-8 barley B1 , C, D, hordein Theor Appl Gen 98:1253-62, 1999; Plant J 4:343-55, 1993;

Mol Gen Genet 250:750-60, 1996

barley DOF Mena et al, The Plant Journal, 1 16(1 ): 53-62, 1998 blz2 EP99106056.7

synthetic promoter Vicente-Carbajosa et al., Plant J. 13: 629-640, 1998.

rice prolamin NRP33 Wu et al, Plant Cell Physiology 39(8) 885-889, 1998 rice a-globulin Glb-1 Wu et al, Plant Cell Physiology 39(8) 885-889, 1998 rice OSH1 Sato et al, Proc. Natl. Acad. Sci. USA, 93: 81 17-8122, 1996 rice a-globulin REB/OHP-1 Nakase et al. Plant Mol. Biol. 33: 513-522, 1997 rice ADP-glucose pyrophos- Trans Res 6:157-68, 1997

phorylase

maize ESR gene family Plant J 12:235-46, 1997

sorghum okafirin DeRose et al., Plant Mol. Biol 32:1029-35, 1996

KNOX Postma-Haarsma et al, Plant Mol. Biol. 39:257-71 , 1999 rice oleosin Wu et al, J. Biochem. 123:386, 1998

sunflower oleosin Cummins et al., Plant Mol. Biol. 19: 873-876, 1992

PRO01 17, putative rice 40S WO 2004/070039

ribosomal protein

PRO0151 , rice WSI 18 WO 2004/070039

PRO0175, rice RAB21 WO 2004/070039

PRO005 WO 2004/070039

PRO0095 WO 2004/070039

oamylase (Amy32b) Lanahan et al, Plant Cell 4:203-21 1 , 1992; Skriver et al,

Proc Natl Acad Sci USA 88:7266-7270, 1991

cathepsin β-like gene Cejudo et al, Plant Mol Biol 20:849-856, 1992

Barley Ltp2 Kalla et al., Plant J. 6:849-60, 1994

Chi26 Leah et al., Plant J. 4:579-89, 1994

Maize B-Peru Selinger et al., Genetics 149;1 125-38,1998

Table 4: Examples of endosperm-specific promoters

Gene source Reference

glutelin (rice) Takaiwa et al. (1986) Mol Gen Genet 208:15-22; Takaiwa et al.

(1987) FEBS Letts. 221 :43-47

zein Matzke et al., (1990) Plant Mol Biol 14(3): 323-32

wheat LMW and HMW Colot et al. (1989) Mol Gen Genet 216:81 -90, Anderson et al. glutenin-1 (1989) NAR 17:461 -2

wheat SPA Albani et al. (1997) Plant Cell 9:171 -184

wheat gliadins Rafalski et al. (1984) EMBO 3:1409-15

barley Itr1 promoter Diaz et al. (1995) Mol Gen Genet 248(5):592-8

barley B1 , C, D, hordein Cho et al. (1999) Theor Appl Genet 98:1253-62; Muller et al.

(1993) Plant J 4:343-55; Sorenson et al. (1996) Mol Gen Genet 250:750-60

barley DOF Mena et al, (1998) Plant J 1 16(1 ): 53-62

blz2 Onate et al. (1999) J Biol Chem 274(14):9175-82

synthetic promoter Vicente-Carbajosa et al. (1998) Plant J 13:629-640

rice prolamin NRP33 Wu et al, (1998) Plant Cell Physiol 39(8) 885-889

rice globulin Glb-1 Wu et al. (1998) Plant Cell Physiol 39(8) 885-889

rice globulin REB/OHP-1 Nakase et al. (1997) Plant Molec Biol 33: 513-522

rice ADP-glucose Russell et al. (1997) Trans Res 6:157-68

pyrophosphorylase

maize ESR gene family Opsahl-Ferstad et al. (1997) Plant J 12:235-46

sorghum kafirin DeRose et al. (1996) Plant Mol Biol 32:1029-35

Table 5: Examples of embryo-specific promoters

Table 6: Examples of aleurone-specific promoters

If polypeptide expression is desired, it is generally desirable to include a polyadenylation region at the 3'-end of a polynucleotide's coding region. The polyadenylation region can, preferably, be derived from the natural gene, from a variety of other plant genes, or from T-DNA, and the like. The 3' end sequence to be added may be derived from, for example, the nopaline synthase or octopine synthase genes, or alternatively from another plant gene or, preferably plant virus, or, less preferably, from any other eukaryotic gene. An intron sequence may also be added to the 5' untranslated region (UTR) or the coding sequence of the partial coding sequence to increase the amount of the mature message that accumulates in the cytosol. Inclusion of a spliceable intron in the transcription unit in both plant and animal expression constructs has been shown to increase gene expression at both the mRNA and protein levels up to 1000-fold (Buchman and Berg (1988) Mol. Cell biol. 8: 4395-4405; Callis et al. (1987) Genes Dev 1 :1 183-1200). Such intron enhancement of gene expression is typically greatest when placed near the 5' end of the transcription unit. Use of the maize introns Adh1 -S intron 1 , 2, and 6, the Bronze-1 intron are known in the art. For general information see: The Maize Handbook, Chapter 1 16, Freeling and Walbot, Eds., Springer, N.Y. (1994).

Also preferably, the modulation is an inactivation or inhibition, i.e., preferably, a modulation decreasing the activity of an enzyme catalyzing said metabolic conversion. Preferably, the inhibition is reversible, more preferably, the inhibition is irreversible, i.e. an inactivation, preferably, a partial inactivation or a complete inactivation. A direct inhibition is achieved by a compound, which binds to the enzyme and thereby inhibits its catalytic activity. Compounds which directly inhibit enzymes in this sense, are, preferably, compounds which block the interaction of the enzyme with other proteins or with its substrates. Alternatively, but

nevertheless preferred, a direct inhibitor of an enzyme may induce an allosteric change in the conformation of the polypeptide constituting the enzyme. The allosteric change may

subsequently block the interaction of the enzyme with other proteins or with its substrates and, thus, interfere with the catalytic activity of the enzyme. Compounds which are suitable as direct inhibitors of enzymes encompass small molecule antagonists (e.g., substrate analogues, allosteric inhibitors), antibodies, aptamers, mutants or variants of the enzyme, a dominant- negative subunit of an enzyme complex, and the like. Thus, preferably, modulation comprises introducing a mutated enzyme with decreased or absent activity in addition to the enzyme present in the cell, e.g. by introducing a gene encoding said mutated enzyme due to a mutation, e.g. a point mutation, into the plant cell; or, more preferably, modulation comprises introducing a mutated enzyme with decreased or absent activity replacing the enzyme present in the cell, e.g. by replacing all copies of the gene encoding the cellular enzyme by copies comprising a mutation, e.g. a point mutation; or by replacing one or more copies of the gene encoding the cellular enzyme by one or more copies comprising a mutation, e.g. a point mutation, and by inactivating all further genes encoding said cellular enzyme potentially present in said plant cell.

Reference herein to an "endogenous" gene not only refers to the gene in question as found in a plant in its natural form (i.e., without there being any human intervention), but also refers to that same gene (or a substantially homologous nucleic acid/gene) in an isolated form subsequently (re)introduced into a plant (a transgene). For example, a transgenic plant containing such a transgene may encounter a substantial reduction of the transgene expression and/or substantial reduction of expression of the endogenous gene. The isolated gene may be isolated from an organism or may be manmade, for example by chemical synthesis.

The term "small molecule antagonist" as used herein refers to a chemical compound that, preferably specifically, interacts with and inhibits the enzyme. A small molecule as used herein preferably has a molecular weight of less than 1000 Da, more preferably, less than 800 Da, less than 500 Da, less than 300 Da, or less than 200 Da. Such small molecules are, preferably, capable of diffusing across cell membranes so that they can enter and reach intracellular sites of action. Suitable chemical compounds encompass small organic molecules. Preferably, the small molecule antagonist is a substrate analogon or an allosteric inhibitor. The term "antibody" as used herein encompasses all types of an antibody which, preferably, specifically binds to an enzyme and inhibits its activity. Preferably, the antibody of the present invention is a monoclonal antibody, a polyclonal antibody, a single chain antibody, a chimeric antibody or any fragment or derivative of such antibodies being still capable of binding to the enzyme and inhibiting its catalytic activity. Such fragments and derivatives comprised by the term antibody as used herein encompass a bispecific antibody, a synthetic antibody, an Fab, F(ab)2 Fv or scFv fragment, or a chemically modified derivative of any of these antibodies. Specific binding as used in the context of the antibody of the present invention means that the antibody does not cross-react with other polypeptides or, preferably, does not inhibit the activity of other polypeptides. Specific binding and/or inhibition can be tested by various well known techniques. Inhibition is preferably tested by an enzymatic assay determining the activity of the enzyme in question in the presence and in the absence of the antibody. Antibodies or fragments thereof, in general, can be obtained by using methods which are well known to the skilled person. Monoclonal antibodies can be prepared the techniques which comprise the fusion of mouse myeloma cells to spleen cells derived from immunized mammals and, preferably, immunized mice. Monoclonal antibodies which specifically bind to the enzyme can be prepared using the well known hybridoma technique, the human B cell hybridoma technique, and the EBV hybridoma technique. Specifically binding antibodies which affect at least one catalytic activity can be identified by assays known in the art.

The term "aptamer" as used herein relates to oligonucleic acid or peptide molecules that bind to a specific target polypeptide. Oligonucleic acid aptamers are engineered through repeated rounds of selection or the so called systematic evolution of ligands by exponential enrichment (SELEX technology). Peptide aptamers are designed to interfere with protein interactions inside cells. Preferably, they usually comprise a variable peptide loop attached at both ends to a protein scaffold. This double structural constraint shall increase the binding affinity of the peptide aptamer into the nanomolar range. Said variable peptide loop length is, preferably, composed of ten to twenty amino acids, and the scaffold may be any protein having improved solubility properties, such as thioredoxin A. Peptide aptamer selection can be made using different systems including, e.g., the yeast two-hybrid system. Aptamers which affect at least one biological activity of an enzyme can be identified by functional assays known in the art.

The term "dominant-negative subunit of an enzyme complex", as used herein, refers to a subunit of an enzyme complex mutated such that it is still able to bind to the enzyme complex, but not catalytically active. Thus, the non-catalytic dominant-negative subunit dislocates a functional subunit from the complex, leading to a decreased, altered, or abolished activity of the complex.

Inhibition of an enzyme according to the present invention is, preferably, achieved by indirect inhibition wherein the number of molecules of said enzyme present in a plant cell is reduced. Preferably, the number of molecules of said enzyme is reduced to zero, i.e. production of enzyme molecules is abolished. Such a reduction of the number of enzyme molecules is, preferably, accomplished by a reduction or prevention of the expression of the gene coding for said enzyme, i.e. by a reduction or prevention of transcription, a destabilization or increased degradation of the transcripts or a reduction or prevention of the translation of the transcripts into enzyme polypeptides. Compounds which are known to interfere with transcription and/or translation of genes as well as stability of transcripts are inhibitory nucleic acids. Such inhibitory nucleic acids, usually, recognize their target transcripts by hybridization of nucleic acid sequences present in both, the target transcript and the inhibitory nucleic acid, being

complementary to each other. Accordingly, for a given transcript with a known nucleic acid sequence, such inhibitors can be designed and synthesized without further ado by the skilled artisan. Suitable assays for testing the activity are known in the art. Specifically, the presence or absence of the target transcript can be measured or the presence or absence of the protein encoded thereby, or its activity, can be measured in the presence and absence of the putative inhibitory nucleic acid. A nucleic acid which, indeed, is an inhibitory nucleic acid can be subsequently identified if in the presence of the inhibitory nucleic acid, the target transcript, the polypeptide, or the enzymatic activity encoded thereby can no longer be detected or is detectable at reduced amounts.

Reference herein to "reducing the number of enzyme molecules" or "reduction or substantial elimination" is taken to mean a decrease in endogenous gene expression and polypeptide levels and/or polypeptide activity relative to control plants. The reduction or substantial elimination is, preferably to a statistically significant extent and, more preferably, in increasing order of preference a reduction of at least 10 %, 20 %, 30 %, 40 % or 50 %, 60 %, 70 %, 80 %, 85 %, 90 %, 95 %, 96 %, 97 %, 98 %, 99 % or more than 99 % compared to that of control plants.

Reference herein to "decreased expression" or "reduction or substantial elimination" of expression is taken to mean a decrease in endogenous gene expression and/or polypeptide levels and/or polypeptide activity relative to control plants. The reduction or substantial elimination is in increasing order of preference at least 10 %, 20 %, 30 %, 40 % or 50 %, 60 %, 70 %, 80 %, 85 %, 90 %, 95 %, 96 %, 97 %, 98 %, 99 % or more than 99 % reduced compared to that of control plants. For the reduction or substantial elimination of expression an endogenous gene in a plant, a sufficient length of substantially contiguous nucleotides of a nucleic acid sequence is required. In order to perform gene silencing, this may be as little as 20, 19, 18, 17, 16, 15, 14, 13, 12, 1 1 , 10 or fewer than 10 nucleotides; alternatively this may be as much as the entire gene (including the 5' and/or 3' UTR, either in part or in whole). The stretch of substantially contiguous nucleotides may be derived from the nucleic acid encoding the protein of interest (target gene), or from any nucleic acid capable of encoding an ortholog, paralog or homolog of the protein of interest. Preferably, the stretch of substantially contiguous nucleotides is capable of forming hydrogen bonds with the target gene (either sense or antisense strand), more preferably, the stretch of substantially contiguous nucleotides has, in increasing order of preference, 50 %, 60 %, 70 %, 80 %, 85 %, 90 %, 95 %, 96 %, 97 %, 98 %, 99 %, 100 % sequence identity to the target gene (either sense or antisense strand). A nucleic acid sequence encoding a (functional) polypeptide is not a requirement for the various methods discussed herein for the reduction or substantial elimination of expression of an endogenous gene. This reduction or substantial elimination of expression may be achieved using routine tools and techniques. A preferred method for the reduction or substantial elimination of endogenous gene expression is by introducing and expressing in a plant a genetic construct into which the nucleic acid (in this case a stretch of substantially contiguous nucleotides derived from the gene of interest, or from any nucleic acid capable of encoding an ortholog, paralog or homolog of any one of the protein of interest) is cloned as an inverted repeat (in part or completely), separated by a spacer (non-coding DNA).

Accordingly, the inhibitor of the invention is, preferably, an inhibitory nucleic acid. More preferably, said inhibitory nucleic acid is selected from the group consisting of: an antisense RNA, a ribozyme, a siRNA, a micro RNA, a morpholino or a triple helix forming agent.

The term "antisense RNA", as used herein, refers to an RNA which comprises a nucleic acid sequence which is essentially or perfectly complementary to the target transcript. Preferably, an antisense nucleic acid molecule essentially consists of a nucleic acid sequence being complementary to at least 100 contiguous nucleotides, more preferably, at least 200, at least 300, at least 400 or at least 500 contiguous nucleotides of the target transcript. How to generate and use antisense nucleic acid molecules is well known in the art (see, e.g., Weiss, B. (ed.): Anti-sense Oligodeoxynucleotides and Antisense RNA : Novel Pharmacological and

Therapeutic Agents, CRC Press, Boca Raton, FL, 1997.). The antisense nucleic acid sequence can be produced biologically using an expression vector into which a nucleic acid sequence has been subcloned in an antisense orientation (i.e., RNA transcribed from the inserted nucleic acid will be of an antisense orientation to a target nucleic acid of interest). Preferably, production of antisense nucleic acid sequences in plants occurs by means of a stably integrated nucleic acid construct comprising a promoter, an operably linked antisense oligonucleotide, and a terminator. In a preferred embodiment of the present invention, the production of antisense nucleic acid sequences in plants occurs by means of a non-stably integrated nucleic acid construct comprising a promoter, an operably linked antisense oligonucleotide, and a terminator, e.g. in a transient fashion. The nucleic acid molecules used for silencing in the methods of the invention (whether introduced into a plant or generated in situ) hybridize with or bind to mRNA transcripts and/or genomic DNA encoding a polypeptide to thereby inhibit expression of the protein, e.g., by inhibiting transcription and/or translation. The hybridization can be by conventional nucleotide complementarity to form a stable duplex, or, for example, in the case of an antisense nucleic acid sequence which binds to DNA duplexes, through specific interactions in the major groove of the double helix. Antisense nucleic acid sequences may be introduced into a plant by transformation or direct injection at a specific tissue site. Alternatively, antisense nucleic acid sequences can be modified to target selected cells and then administered systemically. For example, for systemic administration, antisense nucleic acid sequences can be modified such that they specifically bind to receptors or antigens expressed on a selected cell surface, e.g., by linking the antisense nucleic acid sequence to peptides or antibodies which bind to cell surface receptors or antigens. The antisense nucleic acid sequences can also be delivered to cells using the vectors described herein. In a preferred embodiment of the present invention, the nucleic acid molecules used for silencing in the methods of the invention (whether introduced into a plant or generated in situ) hybridize with or bind to the regulatory region, eg the promoter of the target gene.

According to a further aspect, the antisense nucleic acid sequence is an a-anomeric nucleic acid sequence. An a-anomeric nucleic acid sequence forms specific double-stranded hybrids with complementary RNA in which, contrary to the usual b-units, the strands run parallel to each other (Gaultier et al. (1987) Nucl Acid Res 15: 6625-6641 ). The antisense nucleic acid sequence may also comprise a 2'-o-methylribonucleotide (Inoue et al. (1987) Nucl Ac Res 15, 6131 -6148) or a chimeric RNA-DNA analogue (Inoue et al. (1987) FEBS Lett. 215, 327-330).

The term "ribozyme", as used herein, refers to catalytic RNA molecules possessing a well defined tertiary structure that allows for catalyzing either the hydrolysis of one of their own phosphodiester bonds (self-cleaving ribozymes), or the hydrolysis of bonds in other RNAs, but they have also been found to catalyze the aminotransferase activity of the ribosome. The ribozymes envisaged in accordance with the present invention are, preferably, those which specifically hydrolyze the target transcripts. In particular, hammerhead ribozymes are preferred in accordance with the present invention. How to generate and use such ribozymes is well known in the art (see, e.g., Hean J, Weinberg MS (2008). "The Hammerhead Ribozyme Revisited: New Biological Insights for the Development of Therapeutic Agents and for Reverse Genomics Applications". In Morris KL. RNA and the Regulation of Gene Expression: A Hidden Layer of Complexity. Norfolk, England: Caister Academic Press).

The term "siRNA" as used herein refers to small interfering RNAs (siRNAs) which are complementary to target RNAs (encoding a gene of interest) and diminish or abolish gene expression by RNA interference (RNAi). Without being bound by theory, RNAi is generally used to silence expression of a gene of interest by targeting mRNA. Briefly, the process of RNAi in the cell is initiated by double stranded RNAs (dsRNAs) which are cleaved by a ribonuclease, thus producing siRNA duplexes. The siRNA binds to another intracellular enzyme complex which is thereby activated to target whatever mRNA molecules are homologous (or

complementary) to the siRNA sequence. The function of the complex is to target the

homologous mRNA molecule through base pairing interactions between one of the siRNA strands and the target mRNA. The mRNA is then cleaved approximately 12 nucleotides from the 3' terminus of the siRNA and degraded. In this manner, specific mRNAs can be targeted and degraded, thereby resulting in a loss of protein expression from the targeted mRNA. A complementary nucleotide sequence as used herein refers to the region on the RNA strand that is complementary to an RNA transcript of a portion of the target gene. The term "dsRNA" refers to RNA having a duplex structure comprising two complementary and anti-parallel nucleic acid strands. Not all nucleotides of a dsRNA necessarily exhibit complete Watson-Crick base pairs; the two RNA strands may be substantially complementary. The RNA strands forming the dsRNA may have the same or a different number of nucleotides, with the maximum number of base pairs being the number of nucleotides in the shortest strand of the dsRNA. Preferably, the dsRNA is no more than 49, more preferably less than 25, and most preferably between 19 and 23, nucleotides in length. dsRNAs of this length are particularly efficient in inhibiting the expression of the target gene using RNAi techniques. dsRNAs are subsequently degraded by a ribonuclease enzyme into short interfering RNAs (siRNAs). The complementary regions of the siRNA allow sufficient hybridization of the siRNA to the target RNA and thus mediate RNAi. In mammalian cells, siRNAs are approximately 21 -25 nucleotides in length. The siRNA sequence needs to be of sufficient length to bring the siRNA and target RNA together through

complementary base-pairing interactions. The length of the siRNA is preferably greater than or equal to ten nucleotides and of sufficient length to stably interact with the target RNA;

specifically 10-30 nucleotides; more specifically any integer between 10 and 30 nucleotides, most preferably 10, 1 1 , 12, 13, 14, 15, 16, 17, 18, 19, 20, 21 , 22, 23, 24, 25, 26, 27, 28, 29, or 30 nucleotides. By "sufficient length" is meant an oligonucleotide of greater than or equal to 15 nucleotides that is of a length great enough to provide the intended function under the expected conditions. By "stably interact" is meant interaction of the small interfering RNA with target nucleic acid (e.g., by forming hydrogen bonds with complementary nucleotides in the target under physiological conditions). Generally, such complementarity is 100 % between the siRNA and the RNA target, but can be less if desired, preferably 91 %, 92 %, 93 %, 94 %, 95 %, 96 %, 97 %, 98 %, or 99 %. For example, 19 bases out of 21 bases may be base-paired. In some instances, where selection between various allelic variants is desired, 100 % complementary to the target gene is required in order to effectively discern the target sequence from the other allelic sequence. When selecting between allelic targets, choice of length is also an important factor because it is the other factor involved in the percent complementary and the ability to differentiate between allelic differences. Methods relating to the use of RNAi to silence genes in organisms, including C. elegans, Drosophila, plants, and mammals, are known in the art (see, e.g., WO 0129058; WO 09932619; and Elbashir (2001 ), Nature 41 1 : 494-498).

The term "microRNA" as used herein refers to a self complementary single-stranded RNA which comprises a sense and an antisense strand linked via a hairpin structure. The microRNA comprises a strand which is complementary to an RNA targeting sequence comprised by a transcript to be downregulated. micro RNAs are processed into smaller single stranded RNAs and, therefore, presumably also act via the RNAi mechanisms. How to design and to synthesize microRNAs which cause specific degradation of a transcript of interest is known in the art and described, e.g., in EP 1 504 126 A2 or Dimond (2010), Genetic Engineering & Biotechnology News 30 (6):1.

Another example of an RNA silencing method involves the introduction of nucleic acid sequences or parts thereof (in this case a stretch of substantially contiguous nucleotides derived from the gene of interest, or from any nucleic acid capable of encoding an ortholog, paralog or homolog of the protein of interest) in a sense orientation into a plant. "Sense orientation" refers to a DNA sequence that is homologous to an mRNA transcript thereof.

Introduced into a plant would therefore be at least one copy of the nucleic acid sequence. The additional nucleic acid sequence will reduce expression of the endogenous gene, giving rise to a phenomenon known as co-suppression. The reduction of gene expression will be more pronounced if several additional copies of a nucleic acid sequence are introduced into the plant, as there is a positive correlation between high transcript levels and the triggering of co- suppression. The term "morpholino" refers to a synthetic nucleic acid molecule having a length of from 20 to 30 nucleotides, preferably, about 25 nucleotides. Morpholinos bind to complementary sequences of target transcripts by standard nucleic acid base-pairing. They have standard nucleic acid bases which are bound to morpholine rings instead of deoxyribose rings and linked through phosphorodiamidate groups instead of phosphates. The replacement of anionic phosphates with the uncharged phosphorodiamidate groups eliminates ionization in the usual physiological pH range, so morpholinos in organisms or cells are uncharged molecules. The entire backbone of a morpholino is made from these modified subunits. Unlike inhibitory small RNA molecules, morpholinos do not degrade their target RNA molecules. Rather, they sterically block binding to a target sequence within an RNA and simply getting in the way of molecules that might otherwise interact with the RNA (see, e.g., Summerton (1999), Biochimica et Biophysica Acta 1489 (1 ): 141-58).

The term "triple helix forming agent" as used herein refers to oligonucleotides which are capable of forming a triple helix with DNA and, in particular, which interfere upon forming of the triple- helix with transcription initiation or elongation of a desired target gene. The design and manufacture of triple helix forming agents is well known in the art (see, e.g., Vasquez (2002), Quart Rev Biophys 35: 89-107). For optimal performance, the gene silencing techniques used for reducing expression in a plant of an endogenous gene require the use of nucleic acid sequences from monocotyledonous plants for transformation of monocotyledonous plants, and from dicotyledonous plants for transformation of dicotyledonous plants. Preferably, a nucleic acid sequence from any given plant species is introduced into that same species. For example, a nucleic acid sequence from rice is transformed into a rice plant. However, it is not an absolute requirement that the nucleic acid sequence to be introduced originates from the same plant species as the plant in which it will be introduced. It is sufficient that there is substantial homology between the endogenous target gene and the nucleic acid to be introduced. Abolishing production of enzyme molecules, i.e. reduction by 100 %, is accomplished in a variety of ways. The gene coding for said enzyme can, e.g., be deleted or mutated in a way such that a functional enzyme can no longer be expressed (Knockout-mutation, KO-mutation). Alternatively, said gene may be replaced, e.g. by a non-functional gene, by a mutant copy coding for an inactive variant, or by a gene coding for a selectable marker, e.g., preferably, by homologous recombination. Homologous recombination allows introduction into a genome of a selected nucleic acid at a defined selected position. Homologous recombination is a standard technology used routinely in biological sciences for lower organisms such as yeast or the moss Physcomitrella. Methods for performing homologous recombination in plants have been described not only for model plants (Offringa et al. (1990) EMBO J 9(10): 3077-84) but also for crop plants, for example rice (Terada et al. (2002) Nat Biotech 20(10): 1030-4; lida and Terada (2004) Curr Opin Biotech 15(2): 132-8), and approaches exist that are generally applicable regardless of the target organism (Miller et al, Nature Biotechnol. 25, 778-785, 2007). It is known to the skilled person that such deletion, mutation, or replacement will have to be performed for each copy of the wildtype gene coding for said enzyme available in said plant cell. It is also known to the skilled person that said deletion, mutation, or replacement may, but does not have to, extend to isoenzymes, preferably isoenzymes encoded and/or active in other compartments of the cell. A KO-mutation may also be achieved by insertion mutagenesis (for example, T-DNA insertion or transposon insertion) or by strategies as described by, among others, Angell and Baulcombe ((1999) Plant J 20(3): 357-62), (Amplicon VIGS WO 98/36083), or Baulcombe (WO 99/15682). In a preferred embodiment of the present invention the CRISPR- Cas technology is used for the production of such deletion, mutation or replacement as described for example in Shan et al. (2013), Nat Biotech 31 (8): 686-688.

Preferably, a modulation of enzyme molecules or, preferably the activity of enzyme molecules, is achieved by TILLING. The term "TILLING" is an abbreviation of "Targeted Induced Local Lesions In Genomes" and refers to a mutagenesis technology useful to generate and/or identify nucleic acids encoding proteins with modified expression and/or activity. TILLING also allows selection of plants carrying such mutant variants. These mutant variants may exhibit modified expression, either in strength or in location or in timing (if the mutations affect the promoter for example). These mutant variants may exhibit higher activity than that exhibited by the gene in its natural form. TILLING combines high-density mutagenesis with high-throughput screening methods. The steps typically followed in TILLING are: (a) EMS mutagenesis (Redei GP and Koncz C (1992) In Methods in Arabidopsis Research, Koncz C, Chua NH, Schell J, eds.

Singapore, World Scientific Publishing Co, pp. 16-82; Feldmann et al., (1994) In Meyerowitz EM, Somerville CR, eds, Arabidopsis. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, pp 137-172; Lightner J and Caspar T (1998) In J Martinez-Zapater, J Salinas, eds, Methods on Molecular Biology, Vol. 82. Humana Press, Totowa, NJ, pp 91 -104); (b) DNA preparation and pooling of individuals; (c) PCR amplification of a region of interest; (d) denaturation and annealing to allow formation of heteroduplexes; (e) DHPLC, where the presence of a heteroduplex in a pool is detected as an extra peak in the chromatogram; (f) identification of the mutant individual; and (g) sequencing of the mutant PCR product. Methods for TILLING are well known in the art (McCallum et al., (2000) Nat Biotechnol 18: 455-457; reviewed by Stemple (2004) Nat Rev Genet 5(2): 145-50).

Alternatively, a screening program may be set up to identify in a plant population natural variants of a gene, which variants encode polypeptides with reduced activity and/or which variants may lead to reduced expression of the gene. Such natural variants may also be used, for example, to perform homologous recombination in a recipient plant. E.g. screening for natural variants of a gene by TILLING is sometimes referred to as ecoTILLING. Moreover, targeted mutagenesis may be used to achieve a reduction of enzyme molecules, cf. Shan et al. (2013), Nature Biotechnology 31 , 686-688. Preferably, targeted mutagenesis may be used to achieve a modulation of the number and/or of the activity of enzyme molecules. Described above are examples of various methods for the reduction or substantial elimination of expression in a plant of an endogenous gene. A person skilled in the art would readily be able to adapt the aforementioned methods for silencing so as to achieve reduction of expression of an endogenous gene in a whole plant or in parts thereof through the use of an appropriate promoter, for example.

The term "significant", as used in this specification, relates to statistical significance. Whether a data set supports a hypothesis in a statistically significant way can be determined without further ado by the person skilled in the art using various well known statistic evaluation tools, e.g., determination of confidence intervals, p-value determination, Student ^' s t-test, Mann- Whitney test etc. Preferred confidence intervals are at least 90 %, at least 95 %, at least 97 %, at least 98 % or at least 99 %. The p-values are, preferably, 0.1 , 0.05, 0.01 , 0.005, or 0.0001.

The term "amount" relates to the quantity of a metabolite or compound of the present invention. Preferably, the amount is determined as the concentration of the metabolite in the cell, as the fraction of biomass or dry mass, or any other method suitable for determining a quantity of a specific substance. An increase in amount is preferably a significant increase, more preferably an increase of the amount is an increase by 2-5 %, 5-10 %, 10-20 %, 20-50 %, 50-100 %, 10- 100 %, 100-200 %, or 100-500 % as compared to a control plant. Most preferably, an increase in amount is an increase by at least 2 %, 5 %, 10 %, 20 %, 30 %, 40 %, 50 %, 75 %, 100 %, 200 %, 300 %, 400 %, or at least 500 % as compared to a control plant. The term "biomass" as used herein is intended to refer to the total weight of a plant. Within the definition of biomass, a distinction may be made between the biomass of one or more parts of a plant, which may include any one or more of the following: aboveground parts such as but not limited to shoot biomass, seed biomass, leaf biomass, etc.; aboveground harvestable parts such as but not limited to shoot biomass, seed biomass, leaf biomass, etc.; parts below ground, such as but not limited to root biomass, etc.; harvestable parts below ground, such as but not limited to root biomass, etc.; vegetative biomass such as root biomass, shoot biomass, etc.; reproductive organs; and propagules, such as seed.

As used herein, the term "metabolite of interest" relates to any compound of the primary or secondary metabolism of a plant or a combination thereof. Preferably, said metabolite of interest is a metabolite represented in the network model as described elsewhere herein.

Preferably, the metabolite of interest is a compound not synthesized by the body cells of at least one animal species, preferably at least one mammalian species, more preferably at least one livestock species, most preferably, man. Also preferably, the metabolite of interest is a compound not sufficiently synthesized by the body cells of at least one animal species, preferably at least one mammalian species, more preferably at least one livestock species, most preferably, man. Preferably, the metabolite of interest is an amino acid or a vitamin. Also included as metabolites of interest are, preferably, fatty acids, more preferably, unsaturated fatty acids, most preferably, polyunsaturated fatty acids. Further included as metabolites of interest are, preferably, carbohydrates, more preferably, sugars, starch, and the like. Preferably, the metabolite of interest is a combination of plant metabolites in a fixed ratio. More preferably, the metabolite of interest is the combination of chemical compounds as it is present in a plant cell, plant part, or plant according to the invention; i.e. preferably, the metabolite of interest is biomass, more preferably seed biomass, still more preferably rice seed biomass, most preferably, rice seed biomass as described in OECD (2004): Consensus Document on compositional considerations for new varieties of rice (Oryza sativa): key food and feed nutrients and anti-nutrients. According to the invention, a stoichiometric network model may be simplified by representing one or more chemical compounds constituting the metabolite of interest by its or their biochemical precursor(s); e.g., preferably, as detailed in the examples below. It will be understood by the skilled person that in case improvement of storage compound production by a plant is envisaged, it is desirable to optimize metabolism in a way that ensures both growth of the plant and improved production, wherein improved production, preferably, is an increased amount of storage compound produced as compared to an appropriate control. Thus, preferably, the metabolite of interest is a combination of biomass and storage compound, more preferably at a fixed ratio. Alternatively, also preferably, both biomass and storage compound are metabolites of interest; it will be understood by the skilled person that, in such case, preferably, all elementary modes producing both biomass and storage compound will be selected for further analysis as specified herein below.

Methods for determining biomass in a plant and/or seed are known to the skilled person.

Preferred are the following: The plant aboveground area (or leafy biomass) is preferably determined by counting the total number of pixels on the digital images from aboveground plant parts discriminated from the background, preferably in pictures obtained as described herein in the examples. The values, preferably, are averaged for the pictures taken on the same time point from the different angles and are converted to a physical surface value expressed in square mm by calibration. The skilled person knows that the aboveground plant area measured this way correlates with the biomass of plant parts above ground. The above ground area, preferably, is the area measured at the time point at which the plant reached its maximal leafy biomass.

Root biomass is, preferably, measured as maximum biomass of roots observed during the lifespan of a plant. An increase of root biomass may be measured directly by comparing two values of root biomass, or as an increase in the root/shoot index, measured as the ratio between root mass and shoot mass in the period of active growth of root and shoot. Thus, preferably, the root/shoot index is defined as the ratio of the rapidity of root growth to the rapidity of shoot growth in the period of active growth of root and shoot. More preferably, root biomass is determined using a method as described in WO 2006/029987. Another preferred indicator of plant biomass is the height of the plant. Preferably, height of the plant is measured by determining the location of the centre of gravity, i.e., preferably, determining the height (in mm) of the gravity centre of the leafy biomass.

Preferred methods of determining seed-, and in particular seed biomass-related parameters are the following: Preferably, the seeds are harvested and dried. E.g., in rice, preferably, the mature primary panicles are harvested, counted, and dried, preferably for three days in an oven at 37°C, more preferably to weight constancy; the panicles are then threshed and all the seeds are collected and counted. Preferably, in seeds comprising a husk, i.e. a dry outer covering, the filled husks, which may also be referred to as florets, are separated from the empty ones, preferably using an air-blowing device. The empty husks are, preferably, discarded and the remaining fraction is counted again. Preferably, the filled husks are weighed on an analytical balance. The total number of seeds is, preferably, determined by counting the number of filled husks that remained after the separation step, and the total seed weight is measured by weighing all filled husks harvested from a plant. The total number of seeds (or florets) per plant is, preferably, determined by counting the number of husks (whether filled or not) harvested from a plant. Thousand Kernel Weight (TKW) may be extrapolated from the number of seeds counted and their total weight. The Harvest Index (HI) in the present invention is defined as the ratio between the total seed weight and the above ground area (mm ² ), multiplied by a factor 106. The number of flowers per panicle as defined in the present invention is the ratio between the total number of seeds over the number of mature primary panicles. The "seed fill rate" or "seed filling rate" as defined in the present invention is the proportion (expressed in %) of the number of filled seeds (i.e. florets containing seeds) over the total number of seeds (i.e. total number of florets). In other words, the seed filling rate is the percentage of florets that are filled with seed.

It will be understood by the skilled person that in the method according to the present invention, preferably, pseudo steady-state is assumed for all metabolites for modeling purposes.

Accordingly, in the method, initially a candidate metabolic conversion step is identified, the modulation of which causes a change in the flux to or from a metabolite of interest; however, in case only said specific metabolic conversion step is modulated, an increase in flux to a metabolite of interest will lead to an increase of the concentration of said metabolite; and a decrease in flux to a metabolite of interest will lead to an decrease of the concentration of said metabolite.

The term "network model", as used herein, relates to a representation and simulation of metabolic and physical conversions that determine the physiological and biochemical properties of a plant. Preferably, the network model comprises the metabolic conversions of at least one synthesis pathway, more preferably the synthesis pathways accounting for at least 50 %, more preferably for at least 75 % of the amount of analyte of interest synthesized, still more preferably all synthesis pathways for the metabolite of interest. More preferably, the network model comprises all metabolic conversions having an impact on the amount of the metabolite of interest. The term "having an impact" relates to a metabolic conversion which, when abolished, leads to a deviation from normal of the amount of the metabolite of interest of at least 5 %, at least 10 %, at least 25 %, at least 50 %, at least 100 %, at least 200 %, at least 500 %, or at least 1000 %.

Preferably, the network model comprises or consists of a medium-scale representation of relevant metabolic conversion steps of the anabolic and catabolic pathways of the metabolism of the plant cell, plant or plant part, wherein each metabolic conversion step is defined by its underlying reaction stoichiometry. As used herein, the term "medium-scale representation of relevant metabolic conversion steps" relates to a network model comprising at least 50, preferably at least 60, more preferably at least 70 metabolic conversion steps. Preferably, the medium-scale representation of relevant metabolic conversion steps relates to a network model comprising at most 1 10, more preferably at most 100 metabolic conversion steps. Accordingly, the medium-scale representation of relevant metabolic conversion steps, preferably, comprises from 60 to 1 10, more preferably, from 70 to 100 metabolic conversion steps. Preferably, the stoichiometric network model includes metabolic conversion steps occurring in at least one, preferably at least two, more preferably, at least three cellular compartments, wherein said cellular compartments are, preferably, cytosol, mitochondrion, and/or plastid. Preferably, said metabolic conversion steps comprise the metabolic conversion steps accounting for at least 50 %, more preferably at least 70 % of the biomass, preferably dry biomass, of the respective plant, plant cell, or plant part. Preferably, the network model comprises at least 5 transport reactions between different compartments and/or at least 5 exchange reactions with the external environment. Also preferably, the network model comprises all known metabolic conversions of a plant. The term "known metabolic conversion", preferably, includes metabolic conversions known from in silico predictions of enzymes encoded in the genome of said plant.

Preferably, the network model is organized into compartments, preferably corresponding to the compartments present in the plant cell underlying the model. Preferably, the network model is organized into at least two compartments, preferably selected from cytosol, mitochondrion, plastid, and peroxisome. Preferably, the two-compartment model comprises at least cytosol and mitochondrion and/or plastid. More preferably, the network model is organized into at least three compartments, preferably selected from cytosol, mitochondrion, plastid, and peroxisome.

Preferably, the three-compartment model comprises at least cytosol, mitochondrion, and plastid. Even more preferably, the network model is organized into at least four compartments, preferably being cytosol, mitochondrion, plastid, and peroxisome. Most preferably, the network model is organized into the four compartments cytosol, mitochondrion, plastid, and peroxisome.

Preferably, the cytosol in the network model comprises the Embden-Meyerhof-Parnas (EMP) pathway, the oxidative part of the pentose-phosphate (PP) pathway, and the oxidative part of sucrose synthesis. Preferably, the plastid in the network model comprises photosynthetic Calvin-Benson (CB) cycle, the EMP pathway, the oxidative and non-oxidative PP pathway, starch metabolism, and the part of the photo-respiratory system catalyzed by ribulose bisphosphate oxygenase, phosphoglycolate phosphatase, and glycerate kinase. Preferably, the peroxisome in the network model comprises the reactions of photorespiration not comprised in the plastid and the gyoxylate metabolism. Preferably, the mitochondrion in the network model comprises the tricarboxylic acid cycle. Preferably, pyruvate kinase in the model is comprised in the cytosol and in the plastid; also preferably, supply of cytosolic Acetyl-CoA is attributed to ATP-citrate-lyase in the cytoplasm. Also preferably, the plastid comprises malic enzyme.

Preferably, uni- or bidirectional transport is assumed in the network model for pyruvate, malate, oxaloacetate, succinate, and citrate. Preferably, transport between cytosol and plastid is assumed in the network model for 3-phosphoglycerate, glycerate, glycolate, malate, pyruvate, phosphoenolpyruvate, xylulose 5-phosphate, glucose 6-phosphate and dihydroxyacetone phosphate. Preferably, CO2 is assumed to freely diffuse within the cell, and redox equivalents and ATP are balanced over the whole network model. Preferably, the network model may represent reactions performed by a plant cell in the absence of photosynthesis, in the presence of photosynthesis, or both. As will be understood by the skilled person, the network model representing reactions performed by a plant cell in the absence of photosynthesis is, preferably, representative of a plant during nighttime, i.e. after sunset and before sunrise. Preferably, said network model representing reactions performed by a plant cell in the absence of photosynthesis is an approximation of the metabolism of non- photosynthetic tissue, e.g. seed tissue, more preferably seed storage tissue. Also preferably, the network model represents reactions performed by a plant cell under stress conditions, e.g. under drought stress, nitrogen-deprivation stress, or salt stress. Preferably, the network model representing reactions performed by a plant cell under stress conditions deviates only in a small subset of reactions from the network model representing reactions performed by a plant cell in the presence of photosynthesis and/or the absence of photosynthesis. It will be understood that, preferably, in particular the representation of biomass and/or storage compound(s) may be different in a network model representing reactions performed by a plant cell under stress conditions as compared to a network model representing reactions performed by a plant cell under non-stress conditions.

The term "stoichiometric network model", as used herein, relates to a network model comprising data related to the stoichiometry of educts and products of the metabolic conversions comprised in said network model.

Preferably, the stoichiometric network model also comprises data related to the composition of the plant, plant part, plant tissue, or plant cell of interest. It is, thus, understood by the skilled person that a stoichiometric network model, preferably, is specific for a specific plant, plant part, or plant tissue having said composition. More preferably, the stoichiometric network model is a stoichiometric network model of a monocot plant, or even more preferably of rice, most preferably of rice seeds. In a preferred embodiment, the stoichiometric network model comprises the data of Table 7 and/or Table 9 below, more preferably consists of the data of Table 7 and/or Table 9 below. In a further preferred embodiment, the stoichiometric network model comprises the data of Table 7 below, more preferably consists of the data of Table 7 below; in a most preferred embodiment, the stoichiometric network model is a stoichiometric network model of rice, preferably of rice seeds, and comprises the data of Table 7 below, more preferably consists of the data of Table 7 below. In a further preferred embodiment, the stoichiometric network model comprises the data of Table 9 below, more preferably consists of the data of Table 9 below; in a most preferred embodiment, the stoichiometric network model is a stoichiometric network model of a leaf of a plant, preferably of a leaf of a C3 plant, and comprises the data of Table 9 below, more preferably consists of the data of Table 9 below. Preferably, the stoichiometric network model does not comprise kinetic data related to the metabolic conversions. Preferably, the stoichiometric network model is implemented in a data processor, more preferably a computer.

The term "mode", as used in the context of the network model of the present invention, relates to a set of metabolic conversion steps, wherein said metabolic conversion steps form an unbranched or branched chain of metabolic interconversions of one or more substrates to one or more products. The term "elementary mode" is known to the skilled person and relates to a thermodynamically and stoichiometrically possible pathway reducing a given metabolism into all unique, non-decomposable biochemical pathways. Accordingly, the term "elementary mode" relates to a set of metabolic conversion steps that could operate in steady-state and cannot conceptually be decomposed into further modes without losing the property of being able to operate in steady state. Preferably, an elementary mode is an unbranched subpart of a stoichiometric network model, which is circular or, more preferably, linear. Also, preferably, the elementary mode may comprise exchange reactions between one or more compartments and/or the external medium. Preferably, an elementary mode is an unbranched subpart of a stoichiometric network model starting with an external metabolite or a metabolite of interest, comprising internal conversion steps at equilibrium, and ending at a metabolite of interest or at an external metabolite. According to the present invention, preferably, all elementary modes of the stoichiometric network model are calculated. As detailed in the examples below, a set of elementary modes producing biomass may be selected before continuing with further analysis.

The term "calculating the correlation between the metabolite of interest and each partial reaction", as used herein, relates to determining whether the amount of metabolite of interest and the amount of product produced in a metabolic conversion step consistently correlate over the elementary modes analyzed. Preferably, the term relates to determining whether the amount of metabolite of interest and the amount of product produced in a metabolic conversion step consistently correlate over the elementary modes analyzed and to determining whether such correlation is statistically significant. More preferably, calculating the correlation between the metabolite of interest and each partial reaction comprises

determining slope and regression coefficient of the correlation of each partial reaction of the stoichiometric network model with the amount of the metabolite(s) of interest. Preferably, said determining slope and regression comprises the following steps (see, e.g. Melzer et al. (2009), BMC Systems Biology 3:120):

1 ) establishing a, preferably normalized, stoichiometric matrix including the elementary modes of the selected set of elementary modes;

2) for each metabolic conversion step comprised in the stoichiometric matrix established in step 1 ), determining, for each elementary mode, the value pair (i) molar amount of product produced in said metabolic conversion step vs.(ii) molar amount of metabolite of interest produced by said elementary mode; and

3) calculating the slope and regression coefficient for said value pairs over all elementary modes selected.

As will be understood by the skilled person, not every metabolic conversion step will have a significant correlation with the metabolite of interest. Accordingly, preferably, only a metabolic conversion step having a regression coefficient of at least 0.6, preferably at least 0.65, most preferably, at least 0.7 will be considered as a candidate metabolic conversion step.

Preferably, according to the method of the present invention, steps (c) and (d) are repeated using one or more different sets of elementary modes. Selection of said different sets of elementary modes is, preferably, based on the normalized molar amount of metabolite of interest produced by the respective elementary mode. Preferably, a subset of high-producing elementary modes, preferably top 25 %, more preferably top 10 %, most preferably top 5 % producing elementary modes are selected; and/or a subset of highest-producing elementary modes, preferably top 4 %, more preferably top 3 %, even more preferably top 2 %, most preferably top 1 % producing elementary modes, are selected. As will be appreciated by the skilled person, sets of elementary modes may also be selected according to different criteria, e.g. a set of suboptimal producing elementary modes, a set of elementary modes containing a specific pathway or set of reactions, and/or a set of elementary modes not containing a certain reaction or set of reactions may be selected. Preferably, calculations of correlation and/or identification of candidate metabolic conversion steps are performed separately for each set of elementary modes selected.

The term "identifying at least one candidate metabolic conversion step", as used herein, relates to selecting, based on the calculation of correlation as described herein, at least one candidate metabolic conversion step from the metabolic conversion steps comprised in the network model as potentially relevant, i.e. as a metabolic conversion step, the modulation of which causes a change of the amount of at the least one metabolite of interest. Preferably, the metabolic conversion step is identified as a metabolic conversion step, the activation of which increases the amount of the metabolite of interest in case the slope of the correlation as determined in step 3) above is positive; conversely, the metabolic conversion step is identified as a metabolic conversion step, the inactivation of which increases the amount of the metabolite of interest in case the slope of the correlation as determined in step 3) above is negative. As will be understood by the skilled person, also preferably, the metabolic conversion step is identified as a metabolic conversion step, the inactivation or reduction of which decreases the amount of the metabolite of interest in case the slope of the correlation as determined in step 3) above is positive; and, conversely, the metabolic conversion step is identified as a metabolic conversion step, the activation of which decreases the amount of the metabolite of interest in case the slope of the correlation as determined in step 3) above is negative. Preferably, at least two, more preferably, at least five, most preferably, at least ten candidate metabolic conversion steps are identified.

Preferably, in a method according to the present invention comprising calculating correlation between the metabolite of interest and partial reactions for at least two sets of elementary modes, identifying at least one candidate metabolic conversion step further comprises first determining at least one, preferably at least two, more preferably at least five, most preferably, at least ten preliminary candidate metabolic conversion steps for each set of elementary modes and comparing (i) identity and (ii) kind of correlation (positive or negative) of the preliminary candidate metabolic conversion steps between said sets of elementary modes. Preferably, preliminary candidate metabolic conversion steps are selected as candidate metabolic conversion steps ("nominated") if they are identified and have the same kind of correlation in at least 50 % of the sets of elementary modes analyzed. More preferably, preliminary candidate metabolic conversion steps are selected as candidate metabolic conversion steps if they are identified and have the same kind of correlation in at least 65 % of the sets of elementary modes analyzed, e.g., preferably, in 2 out of 3. Even more preferably, preliminary candidate metabolic conversion steps are selected as candidate metabolic conversion steps if they are identified and have the same kind of correlation in at least 90 %, most preferably, 100 %, of the sets of elementary modes analyzed, e.g. in 3 out of 3. Advantageously, it was found in the work underlying the present invention that, if appropriate modifications are applied, the method of elementary mode analysis/flux design can be successfully applied to higher plants. Moreover, it was surprisingly found that a medium-scale representation of a plant's metabolism is sufficient for the application of said method and that building blocks of biomass production can be represented by their biochemical precursors, leading to a drastic reduction of calculation effort. It was further established that by selecting different sets of elementary modes and comparing results obtained therewith, highly significant correlations between a metabolic conversion step and a metabolite of interest can be identified. In a further aspect, it was found that by applying the correlation analysis of the present invention to more than one candidate metabolic conversion step, combinations of metabolic conversion steps which can advantageously be combined, can be identified. Overall, the new method of the present invention was found to have the highest sensitivity of all methods evaluated in identifying suitable metabolic conversion steps for improving plant traits. Importantly, in cases where the same candidate conversion steps were identified with different methods known in the art, the proposed direction of modulation proposed by the method of the present invention was identical to those proposed by the other methods evaluated; however, results could be obtained with a significantly lower calculation effort.

The definitions made above apply mutatis mutandis to the following embodiments.

The present invention further relates to a method of identifying at least two candidate metabolic conversion steps, the combined modulation of which causes a change of the amount of at least one metabolite of interest in a plant cell, plant or plant part, said method comprising:

(a) providing at least two candidate metabolic conversion steps,

(b) performing steps (a) to (b) of the method for identifying at least one candidate metabolic conversion step of the present invention,

(c) calculating for each elementary mode of step (b) the correlation between (i) the metabolite of interest, (ii) the first candidate metabolic conversion step, and (iii) the second candidate metabolic conversion step; and

(d) identifying at least two candidate metabolic conversion steps, the combined modulation of which causes a change of the amount of said at least one metabolite of interest based on the correlation calculated in step (c).

The method for identifying at least two candidate metabolic conversion steps, the combined modulation of which causes a change of the amount of at least one metabolite of interest of the present invention, preferably, is an in-silico method. Thus, preferably, most or all of the steps of said method are performed in a computer-assisted mode. Moreover, said method may comprise further steps in addition to the ones explicitly mentioned. Specifically, step (a) may, preferably, comprise steps required for providing at least two candidate metabolic conversion steps as specified herein below, step (b) may, preferably, comprise the further step of generating and/or collecting data required to establish a stoichiometric network model for the metabolism in question, and/or step (d) may, preferably, contain the further steps of validating the combined modulation of said metabolic conversion steps by constructing and analyzing a plant cell, plant or plant part comprising said combined modulation of said metabolic conversion steps as described herein above.

The term "combined modulation", as used herein, relates to a modulation of at least two metabolic conversion steps being present in one plant (i.e., preferably, plant cell, plant or plant part) at the same time. It will be understood that the modulation of said metabolic conversion steps may be the same or may be different, i.e., preferably, the activity of one of said at least two metabolic conversion steps may be increased and the activity of a second of said at least two metabolic conversion step may be decreased; or the activity of two of said at least two metabolic conversion steps may be decreased; or the activity of two of said metabolic conversion steps may be increased. It will be understood by the skilled person that the aforesaid applies to methods comprising more than two candidate metabolic conversion steps mutatis mutandis. Preferably, the modulation of said metabolic conversion steps is effective during the same growth phase or physiologic state of the plant. More preferably, the modulation of said metabolic conversion steps is effective at essentially the same time most preferably, at the same time.

As used herein, the term "providing at least two candidate metabolic conversion steps" relates to identifying at least two metabolic conversion steps which are suspected to cause a change of the amount of at least one metabolite of interest. Methods of providing candidate metabolic conversion steps are well-known in the art and include, preferably, at least one of network- based pathway analysis (NBPA, see, e.g. Schuster et al. (1999), Trends Biotechnol 17:53), flux coupling analysis (FCA, see, e.g. Burgard et al. (2004), Genome Res 14:301 ), and Monte Carlo sampling approach (MCSA, see, e.g. Wiback et al. (2004), J Theoret Biol 228:437). More preferably, said at least two candidate metabolic conversion steps are provided according to the method for identifying at least one candidate metabolic conversion step, the modulation of which causes a change of the amount of at least one metabolite of interest, of the present invention. Preferably, said candidate metabolic conversion steps are suspected to cause said change of the amount of at least one metabolite of interest when modulated singly in a plant. More preferably, said candidate metabolic conversion steps are suspected to cause said change of the amount of at least one metabolite of interest when modulated in a combined way in a plant. Preferably, at least two, more preferably, at least five, most preferably, at least ten candidate metabolic conversion steps are provided.

Preferably, the step of "calculating for each elementary mode of step (b) the correlation between (i) the metabolite of interest, (ii) the first candidate metabolic conversion step, and (iii) the second candidate metabolic conversion step" is performed essentially as described herein above. However, preferably, a combined correlation is calculated for the indicated groups of three values over the selected set of elementary modes. Accordingly, identification of candidate metabolic conversion steps, the combined modulation of which causes a change of the amount of said at least one metabolite of interest is also essentially performed as outlined herein above. The present invention also relates to a method for generating a plant cell, plant or plant part which produces a changed amount of a metabolite of interest when compared to a control, said method comprising:

(a) identifying at least one candidate metabolic conversion step, the modulation of which causes a change of the amount of at least one metabolite of interest in a plant cell, plant or plant part, by the method for identifying at least one candidate metabolic conversion step, the modulation of which causes a change of the amount of at least one metabolite of interest of the present invention; or identifying at least two candidate metabolic conversion steps, the combined modulation of which causes a change of the amount of at least one metabolite of interest in a plant cell, plant or plant part, by the method for identifying at least two candidate metabolic conversion steps, the combined modulation of which causes a change of the amount of at least one metabolite of interest of the present invention; and

(b) stably modulating the said metabolic conversion step or metabolic conversion steps such that the amount of the metabolite of interest is changed in vivo in a plant cell, plant or plant part.

The method for generating a plant cell, plant or plant part of the present invention, preferably, is an in vitro method. Moreover, it may comprise steps in addition to those explicitly mentioned above. For example, further steps may relate, e.g., to introducing a compound modulating the said metabolic conversion step in step b). Moreover, one or more of said steps may be performed by automated equipment. Preferably, the generation of said plant cell does not rely exclusively on natural phenomena such as crossing and selection.

As used herein, the term "stably modulating" relates to modulating as defined herein above over an extended period of time. Preferably, stably modulating relates to modulating a metabolic conversion for at least one week, at least two weeks, at least three weeks, at least four weeks, at least one month, at least two months, at least three months, at least six months, at least one year, or more than one year. This kind of stable modulation can, e.g. be achieved by applying an inhibitor to the plant, which is not removed from metabolism to a significant extent over the said period of time, or by introducing a regulable gene into said plant providing for the intended modulation of the amount of the metabolite of interest and applying an inducer or repressor of said inducible gene to said plant for said period of time. More preferably, stably modulating relates to modulating a metabolic conversion starting at a selected point in time and continuing at least until the plant, plant tissue, plant part, or plant cell is harvested or until the end of the growing season. This kind of stable modulation can, e.g. be achieved by introducing a regulable gene into said plant providing for the intended modulation of the amount of the metabolite of interest and applying an inducer or a repressor of said inducible gene to said plant. It is understood by the skilled artisan that said application of an inducer may have to be repeated in order to maintain induction of the inducible gene and, thereby, the modulation of the metabolite of interest. This kind of modulation can, e.g., also be obtained by introducing a genetic construct into said plant, which can be induced to undergo a genetic rearrangement, wherein said genetic rearrangement produces a modified genetic construct being constitutively active in modulating said metabolite of interest. Most preferably, stably modulating relates to modulating a metabolic conversion in a manner stably inherited over at least two generations. Such stable modulation can, e.g. be achieved by introducing a gene coding for an enzyme modulating the amount of a metabolite of interest or by deleting or mutating a gene coding for an enzyme modulating the amount of a metabolite of interest as described herein above. It is understood that stable modulation according to the present invention can also be achieved by indirect methods as described herein above. The present invention also relates to a method for the manufacture of a metabolite of interest comprising the steps of the aforementioned method for generating a plant cell, plant or plant part of the present invention and the further step of obtaining the metabolite of interest from the generated plant cell, plant or plant part. Preferably, said method is a method for the

manufacture of an increased amount of a metabolite, preferably biomass or storage compound, more preferably, biomass and storage compound.

The present invention further relates to a plant cell, plant or plant part obtainable by the method for generating a plant cell, plant or plant part of the present invention, which produces an increased amount of a metabolite of interest when compared to a control.

Preferably, the present invention further relates to a plant cell, plant or plant part obtainable by the method for generating a plant cell, plant or plant part of the present invention, which produces a decreased amount of a metabolite of interest when compared to a control.

Preferably, in such a case the metabolite of interest is a metabolite, the reduction of which is desired for nutritional or other purposes.

The present invention also relates to a device, preferably a data processing device, comprising a data processor having tangibly embedded least one of the algorithms of the invention.

The term "device" as used herein relates to a system of means comprising at least the aforementioned means operatively linked to each other as to allow the identification of at least one candidate metabolic conversion step of the present invention. How to link the means in an operating manner will depend on the type of means included into the device. Preferably, the device is capable of generating an output file containing at least one candidate metabolic step according to the invention identified based on applying said algorithm on the stoichiometric network of the present invention.

The present invention further relates to a data carrier comprising the data defining the stoichiometric network model of the present invention.

As used herein, the term data carrier relates to a physical object comprising the data of the present invention in a form legible, preferably directly or indirectly, to a human or a data processing device. Preferably, data are stored in analogous form; more preferably, data are stored in digital form. Preferably, data are stored electronically or magnetically on the data carrier. It is understood that, preferably, a data carrier is not of any predetermined form or configuration. Preferably, the data carrier is a radiofrequency identification (RFID) chip, a memory chip, a CD or DVD, a hard disk, or the like. It is understood by the skilled person that data may be stored in an encrypted form on the data carrier.

All references cited in this specification are herewith incorporated by reference with respect to their entire disclosure content and the disclosure content specifically mentioned in this specification.

In view of the above, the following embodiments are preferred:

Embodiment 1 : A method for identifying at least one candidate metabolic conversion step, the modulation of which causes a change of the amount of at least one metabolite of interest in a plant cell, plant or plant part, said method comprising:

(a) establishing a stoichiometric network model for the metabolism of the plant cell, plant or plant part including at least one synthesis pathway for the metabolite(s) of interest;

(b) calculating a set of all elementary modes for said stoichiometric network model and selecting a set of elementary modes comprising, preferably all, elementary modes for which the flux to the metabolite(s) of interest is not zero;

(c) calculating for each elementary mode of step (b) the correlation between (i) the metabolite(s) of interest and (ii) each partial reaction; and

(d) identifying at least one candidate metabolic conversion step, the modulation of which causes a change of the amount of at least one metabolite of interest in a plant cell, plant or plant part, based on the correlation calculated in step (c).

Embodiment 2: The method of embodiment 1 , further comprising repeating calculating correlation according to step c) at least once with a set of elementary modes different from the set or sets of elementary modes used in earlier calculations according to step c).

Embodiment 3: The method of embodiment 1 or 2, wherein said set of elementary modes different from the set or sets of elementary modes used in earlier calculations is

(i) a set of high-producing elementary modes of step (b),

(ii) a set of highest-producing elementary modes of step (b),

(iii) a set of suboptimal producing elementary modes of step (b),

(iv) a set of elementary modes containing a specific pathway or set of reactions, and/or

(v) a set of elementary modes not containing a certain reaction or set of reactions.

Embodiment 4: The method of embodiment 2 or 3, comprising separately calculating said correlations and separately determining at least one preliminary candidate metabolic conversion step each for at least two of said sets of elementary modes, and identifying a candidate metabolic conversion step if said metabolic conversion step is identified as a preliminary candidate metabolic conversion step in at least two, or, preferably, at least three, of said sets of elementary modes. Embodiment 5: The method of any one of embodiments 1 to 4, further comprising the step of determining statistical significance of the correlation determined in step (c) before proceeding to step (d).

Embodiment 6: The method of embodiment 5, wherein statistical significance is determined by applying a cut-off value for the regression coefficient of at least 0.6, or, preferably, at least 0.7. Embodiment 7: The method of any one of embodiments 1 to 6, wherein said elementary modes are branched or unbranched subparts of said stoichiometric network model starting with an external metabolite or a metabolite of interest, comprising internal conversion steps at equilibrium, and ending at a metabolite of interest or at an external metabolite. Embodiment 8: The method of any one of embodiments 1 to 7, wherein as a result of the calculation of step c), (i) a positive correlation indicates a metabolic conversion step, the activation of which increases the amount of said metabolite of interest, and the inhibition of which decreases the amount of said metabolite of interest; and (ii) a negative correlation indicates a metabolic conversion step, the inhibition of which increases the amount of said metabolite of interest, and the activation of which decreases the amount of said metabolite of interest.

Embodiment 9: The method of any one of embodiments 1 to 8, wherein said modulation of a metabolic conversion step encompasses decreasing or increasing the activity of at least one enzyme catalyzing the metabolic conversion step in the plant cell.

Embodiment 10: The method of any one of embodiments 1 to 9, wherein said change of the amount of at least one metabolite of interest is an increase of at least one metabolite of interest.

Embodiment 1 1 : The method of any one of embodiments 1 to 10, wherein said

stoichiometric network model for the metabolism of the plant cell, plant or plant part comprises a medium-scale representation of relevant metabolic conversion steps of the anabolic and catabolic pathways of the metabolism of the plant cell, plant or plant part and wherein each metabolic conversion step is defined by its underlying reaction stoichiometry.

Embodiment 12: The method of any one of embodiments 1 to 1 1 , wherein at least two candidate metabolic conversion steps are identified. Embodiment 13: The method of any one of embodiments 1 to 12, wherein the amount of at least two metabolites of interest is increased, preferably wherein the amount of biomass and of a storage compound, preferably starch, is increased. Embodiment 14: The method of any one of embodiments 1 to 13, wherein said plant cell, plant or plant part is a plant cell, plant or plant part of a monocotyledonous plant, preferably, a rice cell, rice plant, rice plant part, or rice seed. Embodiment 15: The method of any one of embodiments 1 to 14, wherein said metabolite of interest is an amino acid, a fatty acid, or, preferably, a carbohydrate.

Embodiment 16: A method for identifying at least two candidate metabolic conversion steps, the combined modulation of which causes a change of the amount of at least one metabolite of interest in a plant cell, plant or plant part, said method comprising:

(a) providing at least two candidate metabolic conversion steps suspected to cause said change of the amount of at least one metabolite of interest,

(b) performing steps (a) to (b) of the method for identifying at least one candidate metabolic conversion step according to any one of embodimentsl to 15,

(c) calculating for each elementary mode of step (b) the correlation between (i) the metabolite of interest, (ii) the first candidate metabolic conversion step, and (iii) the second candidate metabolic conversion step; and

(d) identifying at least two candidate metabolic conversion steps, the combined modulation of which causes a change of the amount of said at least one metabolite of interest based on the correlation calculated in step (c).

Embodiment 17: The method of embodiment 16, wherein said at least two candidate metabolic conversion steps are provided according to the method of any one of embodiments 1 to 15.

Embodiment 18: A method for generating a plant cell, plant or plant part which produces a changed amount of a metabolite of interest when compared to a control, said method comprising:

(a) identifying at least one candidate metabolic conversion step, the modulation of which causes a change of the amount of at least one metabolite of interest in a plant cell, plant or plant part, by the method of any one of embodiments 1 to 16; or identifying at least two candidate metabolic conversion steps, the combined modulation of which causes a change of the amount of at least one metabolite of interest in a plant cell, plant or plant part, by the method of embodiment 16 or 17; and

(b) stably modulating the said metabolic conversion step or metabolic conversion steps such that the amount of the metabolite of interest is changed in vivo in a plant cell, plant or plant part.

Embodiment 19: The method of embodiment 18, wherein said change of amount of metabolite of interest is an increase of said amount of metabolite of interest. Embodiment 20. A method for the manufacture of a metabolite of interest comprising the steps of the method of embodiment 18 or 19 and the further step of obtaining the metabolite of interest from the generated plant cell, plant or plant part. Embodiment 21 : A plant cell, plant or plant part obtainable by the method according to embodiment 18, which produces a changed amount of a metabolite of interest when compared to a control. Embodiment 22: A device comprising a data processor having tangibly embedded least one of the algorithms of the invention.

Embodiment 23: The device of embodiment 22, wherein the device is a data processing device.

Embodiment 24: A data carrier comprising the data defining the stoichiometric network model of the invention and/or an executable code to execute at least one of the algorithms of the invention.

Embodiment 25: A plant cell, plant or plant part comprising

(a) at least one expressible construct for at least one enzyme catalyzing a metabolic conversion step, wherein said metabolic conversion step is identified as a metabolic conversion step, the activation of which causes a change of the amount of at least one metabolite of interest, by the method of any one of embodiments 1 to 17; or

(b) comprising (i) at least one expressible construct for at least one inhibitory nucleic acid for at least one enzyme catalyzing a metabolic conversion step, or (ii) a knock-out mutation for at least one gene encoding an enzyme catalyzing a metabolic conversion step; wherein said metabolic conversion step is identified as a metabolic conversion step, the inactivation of which causes a change of the amount of at least one metabolite of interest, by the method of any one of embodiments 1 to 17.

Embodiment 26: The plant cell, plant or plant part of embodiment 25, wherein said plant cell, plant or plant part comprises an increased amount of starch as compared to a control plant cell, plant or plant part not comprising expressible construct or constructs.

Figure legend

Fig. 1 : schematic representation of artificial plasmid pPROL::AGPase, obtained as described in example 8; LB. left border, RB: right border.

Examples:

1. Model Reconstruction: Metabolic Reconstruction of rice seed metabolism The metabolic model of rice seed metabolism was reconstructed using various public plant metabolic databases (KEGG, RiceCyc, and MetaCrop) to identify pathways and transport processes of relevance. Given a set of potential metabolic reactions for rice seed metabolism, each reaction was checked regarding its occurrence in rice seeds and its compartmental localization, since this information, as provided by public resources, is not necessarily reliable. The final reconstruction of the rice seed model contained pathways and reactions from central carbon metabolism including the following metabolic pathways: sucrose breakdown and starch synthesis, glycolysis, pentose phosphate pathway, TCA cycle, and glyoxylate cycle. In total, the model contained 75 reactions and 62 metabolites distributed into the following compartments: cytosol (24 reactions and 18 metabolites), mitochondrion (13 reactions and 1 1 metabolites), plastid (26 reactions and 24 metabolites) and external environment. The model further contained transport reactions between different compartments (8 reactions and 8 metabolites) and with the external environment (5 reactions and 5 metabolites) (see Table 7).

Table 7: Model reaction list, [c]: cytosolic; [p]: plastidary; [m]: mitochondrial

Reaction Name Reaction Formula

Sucrose import ==> SUCR[c]

Sulfate import ==> S04[c]

DHAP transporter (p) DHAP[c] <==> DHAP[p]

G6P transporter (p) G6P[p] <==> G6P[c]

3PG transporter (p) 3PG[p] <==> 3PG[c]

PYR transporter (p) PYR[p] <==> PYR[c]

PEP transporter (p) PEP[c] <==> PEP[p]

CIT transporter (m) CIT[m] ==> CIT[c]

PYR transporter (m) PYR[c] <==> PYR[m]

NADH dehydrogenase

UBQOL[m] + NAD[p] <==> NADH[p] + UBQON[m]

(UBQON)

NADH kinase 2.4 ADP[p] + NADH[p] ==> NAD[p] + 2.4 ATP[p]

NAD[p] + NADPH[p] + ATP[p] ==> NADH[p] + NADP[p] +

NAD(P)+ kinase

ADP[p]

ATP maintenance ATP[p] ==> ADP[p] + ATP_maint[p]

S04[c] + 6 ferredoxin_red[c] ==> H2S[c] + 6

Sulfite reductase (ferredoxin)

ferredoxin_ox[c]

2 ferredoxin_ox[c] + NADPH[p] ==> 2 ferredoxin_red[c] +

Ferredoxin-NADP+ reductase

NADP[p]

6-Phosphofructokinase (p) F6P[p] + ATP[p] ==> FBP[p] + ADP[p]

Fructose-bisphosphatase (p) FBP[p] ==> F6P[p]

Fructose-bisphosphate

FBP[p] <==> DHAP[p] + GAP[p]

aldolase (p)

Triose-phosphate isomerase

DHAP[p] <==> GAP[p]

(P)

[GAP dehydrogenase (NADP), GAP[p] + NADP[p] + ADP[p] <==> NADPH[p] + ATP[p] + Phosphoglycerate kinase] 3PG[p] [Phosphoglyceromutase,

3PG[p] <==> PEP[p]

Enolase] (p)

Pyruvate, phosphate dikinase

PYR[p] + 2 ATP[p] <==> PEP[p] + 2 ADP[p]

(P)

Pyruvate kinase (p) PEP[p] + ADP[p] ==> PYR[p] + ATP[p]

[Glucose-6-phosphate

dehydrogenase, 6-phospho- gluconolactonase, G6P[p] + 2 NADP[p] ==> RU5P[p] + C02[p] + 2 NADPH[p] phosphogluconate

dehydrogenase (decarb.)]

Ribose 5-phosphate

R5P[p] <==> RU5P[p]

isomerase

Ribulose-phosphate 3-

XU5P[p] <==> RU5P[p]

epimerase (p)

Transaldolase GAP[p] + S7P[p] <==> E4P[p] + F6P[p]

Transketolase E4P[p] + XU5P[p] <==> F6P[p] + GAP[p]

Transketolase S7P[p] + GAP[p] <==> R5P[p] + XU5P[p]

Pyruvate dehydrogenase (p) PYR[p] + NAD[p] ==> ACCOA[p] + NADH[p] + C02[p]

Starch synthesis 2 G6P[p] + 2 ATP[p] ==> 2 ADP[p] + STA[p]

Glucose-6-phosphate

G6P[p] <==> F6P[p]

isomerase (p)

Malate dehydrogenase (m) MAL[m] + NAD[p] ==> OAA[m] + NADH[p]

Malate dehydrogenase (OAA

MAL[m] + NAD[p] ==> C02[p] + NADH[p] + PYR[m] decarboxylating)

Malate dehydrogenase (OAA

MAL[m] + NADP[p] <==> NADPH[p] + C02[p] + PYR[m] decarboxylating) (NADP+)

[Invertase, Fructokinase,

SUCR[c] + 2 ATP[p] ==> G6P[p] + F6P[p] + 2 ADP[p] Hexokinase]

6-Phosphofructokinase (p) F6P[c] + ATP[p] ==> FBP[c] + ADP[p]

Fructose-bisphosphate

FBP[c] <==> DHAP[c] + GAP[c]

aldolase (c)

Triose-phosphate isomerase

GAP[c] <==> DHAP[c]

(c)

OAA transporter (m) OAA[m] <==> OAA[c]

GAP dehydrogenase (NADP) GAP[c] + NADP[p] ==> 3PG[c] + NADPH[p]

[GAP dehydrogenase

GAP[c] + NADP[p] + ADP[p] <==> NADPH[p] + ATP[p] + (phosph.), Phosphoglycerate

3PG[c]

kinase]

[Phosphoglyceromutase,

3PG[c] <==> PEP[c]

Enolase] (c)

Pyruvate kinase (c) PEP[c] + ADP[p] ==> PYR[c] + ATP[p] [Glucose-6-phosphate

dehydrogenase, 6-phospho- gluconolactonase, G6P[c] + 2 NADP[p] ==> RU5P[c] + C02[p] + 2 NADPH[p] phosphogluconate

dehydrogenase (decarb.)]

Ribulose-phosphate 3-

RU5P[c] <==> XU5P[c]

epimerase (c)

ATP citrate synthase CIT[c] + ATP[p] ==> ACCOA[c] + OAA[c] + ADP[p]

PEP carboxykinase (ATP) OAA[c] + ATP[p] ==> C02[p] + PEP[c] + ADP[p]

Glucose-6-phosphate

G6P[c] <==> F6P[c]

isomerase (cyt.)

Malate dehydrogenase (c) OAA[c] + NADH[p] <==> NAD[p] + MAL[m]

PEP carboxylase PEP[c] + C02[p] ==> OAA[c]

Citrate synthase (c) OAA[c] + ACCOA[c] ==> CIT[c]

[Isocitrate lyase, Malate

CIT[c] + ACCOA[c] ==> SUCC[m] + MAL[m]

synthase]

Pyruvate, phosphate dikinase

PYR[c] + 2 ATP[p] <==> PEP[c] + 2 ADP[p]

(c)

Pyruvate dehydrogenase (m) PYR[m] + NAD[p] ==> ACCOA[m] + NADH[p] + C02[p]

Citrate synthase (m) OAA[m] + ACCOA[m] ==> CIT[m]

Isocitrate dehydrogenase

CIT[m] + NAD[p] <==> AKG[m] + C02[p] + NADH[p] (NAD)

Isocitrate dehydrogenase

CIT[m] + NADP[p] <==> AKG[m] + C02[p] + NADPH[p] (NADP)

[oxoglutarate dehydrogenase, AKG[m] + NAD[p] + ADP[p] ==> SUCC[m] + C02[p] + succinate-CoA ligase (ADP)] ATP[p] + NADH[p]

[Succinate dehydrogenase

(UBQON), Fumarate SUCC[m] + UBQON[m] ==> MAL[m] + UBQOL[m] hydratase]

Biomass export BMgrain[c] ==>

CO2 export C02[p] ==>

ATP maintenance export ATP_maint[p] ==>

Table 8: Abbreviations used

Metabolite ID Metabolite Name KEGG Compound ID

SUCR Sucrose C00089

S04 Sulfate C00059

DHAP Dihydroxyacetone phosphate C001 1 1

G6P Glucose-6-phosphate C00092

3PG 3-Phospho glycerate C00197

PYR Pyruvate C00022 PEP Phosphoenolpyruvate C00074

CIT Citrate C00158

UBQOL Ubiquinol C00390

NAD NAD+ C00003

NADH NADH C00004

UBQON Ubiquinone C00399

ADP ADP C00008

ATP ATP C00002

NADPH NADPH C00005

NADP NADP+ C00006

ATP_maint ATP maintenance - ferredoxin_red Ferredoxin (reduces form) C00138

H2S Hydrogen sulfide C00283

ferredoxin_ox Ferredoxin (oxidized form) C00139

F6P Fructose-6-phosphate C00085

FBP Fructose-1 ,6-bisphosphate C00354

GAP Glyceraldehyde 3-phosphate C001 18

RU5P Ribulose 5-phosphate C00199

C02 Carbon dioxide C0001 1

R5P Ribose-5-phosphate C001 17

XU5P Xylulose 5-phosphate C00231

S7P Sedoheptulose 7-phosphate C05382

E4P Erythrose 4-phosphate C00279

ACCOA Acetyl coenzyme A C00024

STA Starch C00369

MAL Malate C00149

OAA Oxaloacetate C00036

SUCC Succinate C00042

AKG Alpha-ketoglutarate C00026

An essential part of the metabolic reconstruction is the definition of the biomass composition. By definition, the biomass composition is defined by its building blocks (such as carbohydrates, lipids, proteins, and the like) and their absolute amount of total biomass content [mmol gDW ^"1 ]. The detailed biomass composition of rice seeds was taken from (OECD (2004), Consensus Document on compositional considerations for new varieties of rice (Oryza sativa): key food and feed nutrients and anti-nutrients), but the respective building blocks were represented by their biochemical precursors in order to limit the complexity of metabolic reconstruction. In particular, biomass was represented by the following: 1.18581 ACCOA[c] + 0.05048 ACCOA[p] + 0.05013 PYR[c] + 0.22287 PYR[p] + 0.08899 PEP[p] + 0.16538 AKG[m] + 0.01241 R5P[p] + 0.04713 E4P[p] + 0.13709 OAA[c] + 0.10217 3PG[p] + 3.31091 NADPH[p] + 1.829 ATP[p] + 0.027 H2S[c] + 0.010552 NADH[p] + 4.09877 STA[p] BM[c] + 3.31091 NADP[p] + 0.010552 NAD[p] + 1 .829 ADP[p] 2. Elementary Modes Calculation and Analysis

Given the metabolic reconstruction of rice seed metabolism, the full set of all elementary modes was calculated. For further analysis, the subset of all biomass producing elementary modes was taken into account. As it turned out that the resulting theoretical states from this subset of relevant elementary modes are too diverse for analysis, the following two physiological scenarios were compared: 1 . Growth Strategy: fixed ratio between starch and biomass, resulting in 12.7 million elementary modes

> Starch synthesis is included in overall biomass synthesis reaction

2. Yield Strategy: variable ratio between starch and biomass production, resulting in 14.8 million elementary modes

> Starch synthesis is not included in overall biomass synthesis reaction; all elementary modes that produce starch and biomass are selected

3. Flux Design - Prediction of single gene targets Here, the method of Flux Design was extended to cope with the complexity of the metabolic reconstruction of the rice seed model. Flux Design was performed using two target reactions (seed biomass production, starch synthesis) and different sets of elementary modes were analyzed:

• Top 1 % producing elementary modes

· Top 5 % producing elementary modes

• All elementary modes that can produce the target reaction

Target Identification Flux Design (Amplification and Attenuation of pathways): The correlation of a metabolic reaction with the seed biomass synthesis reaction delivered genetic targets for amplification (positive correlation) and attenuation (negative correlation). A first exploration of the results from this correlation approach revealed that there is no general conclusion across the entire solution space deducible, which was already known from a similar study of protein production in fungi (Melzer et al. (2009), BMC Systems Biology. 3:120; Driouch et al. (2012), Metabolic

Engineering 14:47). Since there is a large variability between low and high biomass modes, a physiology based pre-selection was conducted. Besides, the full set of biomass producing elementary modes the following subsets were also taken into account:

(1 ) Top 5 % biomass producing elementary modes, and

(2) Top 1 % biomass producing elementary modes.

The application of Flux Design approach delivered many potential genetic targets which are present in at least one of the predefined sets of elementary modes. Due to high number of potential candidates, an additional prioritization was performed, based on the appearance of each reaction in one of the predefined subsets:

Priority 1 : consistent correlation in all subsets

Priority 2: consistent correlation in Top 5 % and Top 1 % biomass producing elementary modes Priority 3: inconsistent correlation in Top 5 % and Top 1 % biomass producing elementary modes.

Potential genetic targets from Flux Design. Based on the Flux Design approach, the target reactions are identified that show a significant correlation to biomass synthesis reaction (+: positive; -: negative) in at least one of the predefined subsets. Depending on their appearance in these subsets, they are assigned to defined groups with a clear priority.

4. Multi Target Strategy

The single gene targets delivered by Flux Design approach can be further used to generate plants comprising multiple modifications. In order to check whether any two reactions could affect the target reaction in the same metabolic reaction, the supportive visualization in a 3D plot is used. It can help to decide whether these two reactions should be used for generating combined mutations.

5. Comparison of Flux Design to other correlation approaches

Besides Flux Design, there are other correlation approaches available in the art that can also be applied to metabolic models. In the following, the results from Flux Design were compared to selected correlation approaches known in the art, using the metabolic reconstruction of rice seeds.

5.1 Network-based Pathway Analysis (NBPA) identifies correlated reaction sets (also called 'Co-sets') according to the reactions' co-occurrence relationship on network-based pathways, such as elementary modes or extreme pathways (Schuster et al. (1999), Trends in

Biotechnology 17: 53). The Co-sets are further analyzed for finding potential knockout or overexpression targets. For the comparison of different algorithms, the elementary modes approach was used to be comparable to the other correlation approaches.

5.2 Flux Coupling Analysis (FCA) is a linear optimization method that can be used to determine whether any two metabolic fluxes are (1 ) directionally coupled, (2) partially coupled, or (3) fully coupled (Burgard et al. (2004), Genome Research 14:301 ). The correlated reaction sets are then able to identify potential knockout or overexpression targets that show a high correlation with a defined target reaction (e.g. biomass synthesis). 5.3 Monte Carlo Sampling Approach (MCSA) generates a sample flux distribution set of available steady-states through uniform random sampling (Wiback et al. (2004), Journal of Theoretical Biology 228:437). The sampling set can be analyzed to measure the linear correlation coefficients of all reaction pairs which in turn give clues on potential knockout or overexpression targets.

Given a subset of target reactions (delivered by Flux Design), the results were compared to the other mentioned correlation approaches. Flux Design was able to find all target reactions while NPBA and FCA found approximately 70 % of all targets, and MCSA could only identify 17 % of all targets. If more than one approach identifies a specific target reaction, the sign of correlation (positive or negative) should be identical. In summary, Flux Design has the highest efficiency in target prediction in comparison to the other tested correlation approaches.

6. Validation of targets via expression in crop plants

6.1 Cloning of the target-encoding nucleic acid sequence

The nucleic acid sequence is amplified by PCR using as template custom-made cDNA libraries in case of eukaryotic source organisms or genomic DNA in case of prokaryotes.

PCR is performed using a commercially available proofreading Taq DNA polymerase in standard conditions, using 200 ng of template in a 50 μΙ PCR mix and appropriate primers, which include the AttB sites for Gateway recombination. The amplified PCR fragment is purified also using standard methods, the first step of the Gateway procedure, the BP reaction, is then performed, during which the PCR fragment recombined in vivo with the pDONR201 plasmid to produce, according to the Gateway terminology, an "entry clone". Plasmid pDONR201 can be purchased from Invitrogen (Life Technologies GmbH, Frankfurter είΓθββ 129B, 64293

Darmstadt, Germany), as part of the Gateway ^® technology.

The entry clone comprising the target gene is then used in an LR reaction with a destination vector used for Oryza sativa transformation. This vector contained as functional elements within the T-DNA borders: a plant selectable marker; a screenable marker expression cassette; and a Gateway cassette intended for LR in vivo recombination with the target nucleic acid sequence of interest already cloned in the entry clone. A promoter functional in rice, e.g. a rice seed endosperm -specific promoter is located upstream of this Gateway cassette.

After the LR recombination step, the resulting expression vector can be transformed into Agrobacterium strain LBA4044 according to methods well known in the art.

In cases where down-regulation of the target gene is envisaged, a hairpin loop is created by producing a tandem sense and antisense fragment (about 500 bp) of each selected gene by commercial gene synthesis. A MARS sequence is added between both orientations to ensure correct folding. The synthetic fragment is subcloned into a vector containing the desired promoter, and later transferred to an Agrobacterium expression vector via Gateway cloning (Life Technologies) as described above. 6.2 Plant transformation

Rice transformation

The Agrobacterium containing the expression vector is used to transform Oryza sativa plants. Mature dry seeds of the rice japonica cultivar Nipponbare are dehusked. Sterilization is carried out by incubating for one minute in 70% ethanol, followed by 30 to 60 minutes, preferably 30 minutes in sodium hypochlorite solution (depending on the grade of contamination), followed by a 3 to 6 times, preferably 4 time wash with sterile distilled water. The sterile seeds are then germinated on a medium containing 2,4-Dichlorophenoxyacetic acid (callus induction medium). After incubation in light for 6 days scutellum-derived calli are transformed with Agrobacterium as described herein below.

Agrobacterium strain LBA4404 containing the expression vector is used for co-cultivation.

Agrobacterium is inoculated on AB medium (Chilton et al. (1974), Proc. Natl. Acad. Sci. USA 71 :3672) with the appropriate antibiotics and cultured for 3 days at 28°C. The bacteria are then collected and suspended in liquid co-cultivation medium to a density (Οϋβοο) of about 1 . The calli are immersed in the suspension for 1 to 15 minutes. The callus tissues are then blotted dry on a filter paper and transferred to solidified, co-cultivation medium and incubated for 3 days in the dark at 25°C. After washing away the Agrobacterium, the calli are grown on 2,4-D- containing medium for 10 to 14 days (growth time for Oryza sativa indica: 3 weeks) under light at 28°C - 32°C in the presence of a selection agent. During this period, rapidly growing resistant callus develops. After transfer of this material to regeneration media, the embryogenic potential is released and shoots developed in the next four to six weeks. Shoots are excised from the calli and incubated for 2 to 3 weeks on an auxin-containing medium from which they are transferred to soil. Hardened shoots are grown under high humidity and short days in a greenhouse.

Transformation of rice cultivar indica can also be done in a similar way as give above according to techniques well known to a skilled person.

35 to 90 independent TO rice transformants are generated for one construct. The primary transformants are transferred from a tissue culture chamber to a greenhouse. After a quantitative PCR analysis to verify copy number of the T-DNA insert, only single copy transgenic plants that exhibit tolerance to the selection agent are kept for harvest of T1 seed. Seeds are then harvested three to five months after transplanting. The method yields single locus transformants at a rate of over 50 % (Aldemita and Hodges (1996), Planta 199(4):612; Chan et al. (1993), Plant Mol. Biol. 22(3):491 ; Hiei et al. (1994), Plant J. 6(2):271 ). 6.3. Phenotypic evaluation procedure

6.3.1 Evaluation setup

35 to 90 independent TO rice transformants are generated. The primary transformants are transferred from a tissue culture chamber to a greenhouse for growing and harvest of T1 seed. A suitable number of events, typically around Nine events, of which the T1 progeny segregated 3:1 for presence/absence of the transgene, are retained. For each of these events, approximately six T1 seedlings containing the transgene (hetero- and homo-zygotes) and approximately six T1 seedlings lacking the transgene (nullizygotes) are selected by monitoring visual marker expression. The transgenic plants and the corresponding nullizygotes are grown side-by-side at random positions. Greenhouse conditions are of shorts days (12 hours light), 28°C in the light and 22°C in the dark, and a relative humidity of 70%. Plants are grown under non-stress conditions with watering at regular intervals to ensure that water and nutrients are not limiting and to satisfy plant needs to complete growth and development. Alternatively, plants are grown under stress conditions, e.g. as follows:

Drought screen

T1 or T2 plants are grown in potting soil under normal conditions until they approach the heading stage. They are then transferred to a "dry" section where irrigation is withheld. Soil moisture probes are inserted in randomly chosen pots to monitor the soil water content (SWC). When SWC goes below certain thresholds, the plants are automatically re-watered continuously until a normal level is reached again. The plants are then re-transferred again to normal conditions. The rest of the cultivation (plant maturation, seed harvest) is the same as for plants not grown under abiotic stress conditions. Growth and yield parameters are recorded as detailed for growth under normal conditions.

Nitrogen use efficiency screen

T1 or T2 plants are grown in potting soil under normal conditions except for the nutrient solution. The pots are watered from transplantation to maturation with a specific nutrient solution containing reduced N nitrogen (N) content, usually between 7 to 8 times less. The rest of the cultivation (plant maturation, seed harvest) is the same as for plants not grown under abiotic stress. Growth and yield parameters are recorded as detailed for growth under normal conditions.

Salt stress screen

T1 or T2 plants are grown on a substrate made of coco fibers and particles of baked clay (Argex) (3 to 1 ratio). A normal nutrient solution is used during the first two weeks after transplanting the plantlets in the greenhouse. After the first two weeks, 25 mM of salt (NaCI) is added to the nutrient solution, until the plants are harvested. Growth and yield parameters are recorded as detailed for growth under normal conditions.

6.3.2 Statistical analysis: F test

A two factor ANOVA (analysis of variance) is used as a statistical model for the overall evaluation of plant phenotypic characteristics. An F test is carried out on all the parameters measured of all the plants of all the events transformed with the gene of the present invention. The F test is carried out to check for an effect of the gene over all the transformation events and to verify for an overall effect of the gene, also known as a global gene effect. The threshold for significance for a true global gene effect is set at a 5% probability level for the F test. A significant F test value points to a gene effect, meaning that it is not only the mere presence or position of the gene that is causing the differences in phenotype.

6.3.3 Parameters measured

From the stage of sowing until the stage of maturity the plants are passed several times through a digital imaging cabinet. At each time point digital images (2048x1536 pixels, 16 million colours) are taken of each plant from at least 6 different angles as described in

WO2010/031780. These measurements are used to determine parameters as required. T1 events can be further evaluated in the T2 generation following the same evaluation procedure as for the T1 generation, e.g. with less events and/or with more individuals per event.

Biomass-related parameter measurement

The plant aboveground area (or leafy biomass) is determined by counting the total number of pixels on the digital images from aboveground plant parts discriminated from the background. This value is averaged for the pictures taken on the same time point from the different angles and is converted to a physical surface value expressed in square mm by calibration.

Experiments show that the aboveground plant area measured this way correlates with the biomass of plant parts above ground. The above ground area is the area measured at the time point at which the plant had reached its maximal leafy biomass.

Increase in root biomass is expressed as an increase in total root biomass (measured as maximum biomass of roots observed during the lifespan of a plant); or as an increase in the root/shoot index, measured as the ratio between root mass and shoot mass in the period of active growth of root and shoot. In other words, the root/shoot index is defined as the ratio of the rapidity of root growth to the rapidity of shoot growth in the period of active growth of root and shoot. Root biomass can be determined using a method as described in WO 2006/029987.

A robust indication of the height of the plant is the measurement of the location of the centre of gravity, i.e. determining the height (in mm) of the gravity centre of the leafy biomass. This avoids influence by a single erect leaf, based on the asymptote of curve fitting or, if the fit is not satisfactory, based on the absolute maximum.

Parameters related to development time

The early vigour is the plant aboveground area three weeks post-germination. Early vigour is determined by counting the total number of pixels from aboveground plant parts discriminated from the background. This value is averaged for the pictures taken on the same time point from different angles and is converted to a physical surface value expressed in square mm by calibration. AreaEmer is an indication of quick early development when this value is decreased compared to control plants. It is the ratio (expressed in %) between the time a plant needs to make 30 % of the final biomass and the time needs to make 90 % of its final biomass, with the origin of time set to 25 days after sowing. The "time to flower" or "flowering time" of the plant can be determined using the method as described in WO 2007/093444.

Seed-related parameter measurements

The mature primary panicles are harvested, counted, bagged, barcode-labelled and then dried for three days in an oven at 37°C. The panicles are then threshed and all the seeds are collected and counted. The seeds are usually covered by a dry outer covering, the husk. The filled husks (herein also named filled florets) are separated from the empty ones using an air- blowing device. The empty husks are discarded and the remaining fraction is counted again. The filled husks are weighed on an analytical balance.

The total number of seeds is determined by counting the number of filled husks that remained after the separation step. The total seed weight is measured by weighing all filled husks harvested from a plant. The total number of seeds (or florets) per plant is determined by counting the number of husks (whether filled or not) harvested from a plant.

Thousand Kernel Weight (TKW) is extrapolated from the number of seeds counted and their total weight. The Harvest Index (HI) in the present invention is defined as the ratio between the total seed weight and the above ground area (mm ² ), multiplied by a factor 10 ⁶ . The number of flowers per panicle as defined in the present invention is the ratio between the total number of seeds over the number of mature primary panicles. The "seed fill rate" or "seed filling rate" as defined in the present invention is the proportion (expressed as a %) of the number of filled seeds (i.e. florets containing seeds) over the total number of seeds (i.e. total number of florets). In other words, the seed filling rate is the percentage of florets that are filled with seed. 7. Elementary Flux Mode (EFM) analysis using the Arabidopsis leaf as a model

Core metabolism

The initial draft of the central carbon metabolism of A. thaliana leaves used for metabolic network reconstruction considered curated knowledge collected in metabolic pathway databases: Kyoto Encyclopedia of Genes and Genomes (Kanehisa et al. (2008) Nucleic Acids Res 36:D480, KEGG release 65.0), MetaCrop (Grafahrend-Belau et al. (2008), Nucleic Acids Res. 36:D954, MetaCrop release 2.0) and AraCyc (Mueller et al. (2003), Plant Physiol.

132(2):453, AraCyc release 8.0). This provided gross information on the genomic pathway repertoire. Where needed, the network was updated with experimental data and primary literature as described in detail below. These individual additions and specifications considered enzyme localization, cofactor usage, and intercompartmental transport.

Genome-based Metabolic Network Reconstruction

The metabolic network of Arabidopsis thaliana leaves was reconstructed to provide a solid basis for subsequent simulation studies by in silico elementary flux mode analysis. The model reactions are listed in Table 9.

Core metabolism. The network created comprised the glycolytic Embden-Meyerhof-Parnas (EMP) pathway, the pentose phosphate (PP) pathway, the tricarboxylic acid (TCA) cycle, glyoxylate metabolism, the Calvin-Benson (CB) cycle for photosynthesis, photorespiration, starch degradation, as well as anabolic pathways for biomass synthesis. The latter considered compartment-specific supply of individual precursors (de Oliveira Dal'Molin et al. (2010), Plant Physiol. 152(2):579; Mintz-Oron et al. (2012), Proc. Natl. Acad. Sci. USA 109(1 ):339). As natural carbon sources, internal starch and atmospheric CO2 were included as substrates, respectively. The final network consisted of 70 reactions and 48 metabolites, whereby 20 reactions belonged to intercompartmental and extracellular transport, respectively. Based on thermodynamic properties, 43 reactions were constrained as irreversible.

Compartmentation. The metabolic network was organized into four compartments respectively: cytosol, peroxisome, mitochondrion and plastid ( Plaxton et al. (1996), Annu. Rev. Plant.

Physiol. Plant Mol. Biol. 47:185; Schnarrenberger and Martin (2002), Eur. J. Biochem

269(3):868; Kruger and von Schaewen (2003), Curr. Opin. Plant. Biol. 6(3):236; Fernie et al. (2004) Curr. Opin. Plant. Biol. 7(3):254; Orzechowski (2008), Acta Biochim. Pol. 55(3):435; MetaCrop release 2.0; AraCyc release 8.0). The cytosol comprised the reactions of the EMP pathway, the oxidative part of the PP pathway and of sucrose synthesis, respectively. The plastid contained the photosynthetic CB cycle, a second copy of the EMP pathway, the oxidative and the non-oxidative PP pathway, starch metabolism, and a part of the photo- respiratory system, catalyzed by ribulose bisphosphate oxygenase, phosphoglycolate phosphatase and glycerate kinase. The peroxisome comprised the remaining reactions of photorespiration and the glyoxylate metabolism. The TCA cycle was located in the

mitochondrion. Pyruvate kinase was assumed as cytosolic and as plastidic reaction (Plaxton et al. (1996), Annu. Rev. Plant. Physiol. Plant Mol. Biol. 47:185), whereas pyruvate

dehydrogenase was assigned to mitochondrion and plastid, respectively ( Schnarrenberger and Martin (2002), Eur. J. Biochem 269(3):868). The supply of cytosolic acetyl-CoA was attributed to ATP-citrate lyase in the cytoplasm, which uses citrate as a substrate (Fatland et al. (2005), Plant Cell 17(1 ): 182). Malic enzyme, specific for photosynthetic tissues, was integrated in the plastid. Intercompartmental and external transport. Uni- or bi-directional transport between cytosol and mitochondrion was assumed for pyruvate, malate, oxaloacetate, succinate and citrate, respectively (Laloi (1999), Cell Mol. Life Sci. 56(1 1-12):918; Grafahrend-Belau et al. (2008), Nucleic Acids Res. 36:D954). Transport between cytosol and plastid was considered for 3- phosphoglycerate, glycerate, glycolate, malate, pyruvate, phosphoenolpyruvate, xylulose 5- phosphate, glucose 6-phosphate and dihydroxyacetonephosphate, respectively (Kruger and von Schaewen (2003), Curr. Opin. Plant Biol. 6(3):236, Furumoto et al. (201 1 ), Nature

476(7361 ):472). So far, a transporter for acetyl-CoA has not been discovered and was therefore not incorporated (Schwender et al. (2006), J. Biol. Chem. 281 (45):34040). C0 ₂ was assumed to freely diffuse within the cell, and reducing equivalents and ATP were balanced over the entire network (Picault et al. (2004), Trends Plant Sci. 9(3):138; Fuhrer and Sauer (2009), J. Bacteriol. 191 (7):21 12; Facchinelli and Weber (201 1 ), Front. Plant Sci. 2:50; Sweetlove and Fernie (2013), Annu. Rev. Plant Biol. 64:723). Energy Household. To account for widely abundant iso-enzymes, capable of utilizing either

NADPH or NADH or both molecules as cofactor, an oxidoreductase for interconversion of

NADPH and NADH was included (Cheung et al. (2013), Plant J. 75(6):1050). The ratio of ATP

NADPH produced in a photosynthetic cell depends on relative activity of the CB cycle and of photorespiration (Foyer et al. (2012), J. Exp. Bot. 63(4):1637). This results in a generally accepted plasticity of the photosynthetic light reactions for energy production (Allen

(2003)Trends Plant Sci. 8(1 ):15; Kramer et al. (2004), Trends Plant Sci. 9(7):349; Cruz et al.

(2005), J. Exp. Bot. 56(41 1 ):395; Munekage et al. (2008) Plant Cell Physiol. 49(1 1 ):1688; Foyer et al. (2012), J. Exp. Bot. 63(4):1637). As it is still unresolved how this mechanism functions exactly, an average ATP to NADPH ratio of 1 .5 was chosen (de Oliveira Dal'Molin et al. (2010),

Plant Physiol. 152(2):579). In order to provide sufficient ATP for maintenance, a surplus of ATP was set as constraint for the modelling (Kromer et al. (2006), Metab. Eng. 8(4):353).

Table 9:Reaction Network for Arabidopsis thaliana leaf metabolism

In silioo transport reactions

'BM[c] -->'

'-> CQ2EX[p]'

'-> STA[p]'

'CQ2[p]->'

'L-aspartate[c] -->'

'Proline[m] -->'

'STA[p] ->'

'SUCR[c] ->'

'LGNCEL[c] ->'

ΊΡΡΡ[ρ] -->'

'Cysteine[p] -->'

'Methionine[p] -->'

Threonine[p] -->'

'Lysine[p] -->'

Tryptophan[p] -->'

Biomass Synthesis

'(1.238) ACCOA[p] + (0.075) PYR[c] + (0.358) C02[p] + (0.820) PEP[p] + (1 .245) AKG[m] + 0.016 R5P[p] + 0.004 DHAP[p] + (0.410) E4P[p] + (0.455) OAA[c] + (0.827) 3PG[c] + 0.009 3PG[p] + (0.1 15) PYR[p] + (8.687) NADPH[p] + (0.384) NAD[p] + (14.341 ) ATP[p] + (0.1 12) GAP[p] + (0.131 ) F6P[c] + (0.530) G6P[p] + (2.097) G6P[c] + (0.068) FUM[m] + (0.050) MAL[c] - -> BM[c] + (8.687) NADP[p] + (0.384) NADH[p] + (14.341 ) ADP[p]'

Plastidic Metabolism

'G6P[p] <-> F6P[p]'

'F6P[p] + ATP[p] -> FBP[p] + ADP[p]'

'FBP[p] -> F6P[p]'

'FBP[p] <==> DHAP[p] + GAP[p]'

ΗΑΡ[ρ] <==> GAP[p]'

'GAP[p] + NADP[p] + ADP[p] <==> NADPH[p] + ATP[p] + 3PG[p]'

'3PG[p] <==> PEP[p]' 'ΡΕΡ[ρ] + ADP[p] -> PYR[p] + ΑΤΡ[ρ]'

'PYR[p] + (2) ATP[p] -> PEP[p] + (2) ADP[p]'

'PYR[p] + NAD[p] -> ACCOA[p] + NADH[p] + CQ2[p]'

'MAL[c] + NADP[p] -> NADPH[p] + C02[p] + PYR[p]'

'G6P[p] + (2) NADP[p] -> (2) NADPH[p] + RU5P[p] + CQ2[p]'

^■ R5P[p] <==> RU5P[p]'

'RU5P[p] <==> XU5P[p]'

'GAP[p] + S7P[p] <==> E4P[p] + F6P[p]'

Έ4Ρ[ρ] + XU5P[p] <==> F6P[p] + GAP[p]'

'S7P[p] + GAP[p] <==> R5P[p] + XU5P[p]'

'RU5P[p] + ATP[p] -> RBP[p] + ADP[p]'

'CQ2EX[p] + RBP[p] -> (2) 3PG[p]'

'RBP[p] -> GLYOX[c] + 3PG[p]'

ΗΑΡ[ρ] + E4P[p] -> S7P[p]'

'STA[p] + 4 ATP[p] -> 4 ADP[p] + 2 G6P[p]'

Cytosolic Metabolism

'G6P[c] <-> F6P[c]'

'F6P[c] + ATP[p] <==> DHAP[c] + GAP[c] + ADP[p]'

'GAP[c] <==> DHAP[c]'

'GAP[c] + NADP[p] -> 3PG[c] + NADPH[p]'

'GAP[c] + NAD[p] + ADP[p] <==> ATP[p] + 3PG[c] + NADH[p]'

■3PG[c] <==> PEP[c]'

'PYR[c] + (2) ATP[p] -> PEP[c] + (2) ADP[p]'

'PEP[c] + ADP[p] -> PYR[c] + ATP[p]'

'G6P[c] + (2) NADP[p] -> (2) NADPH[p] + RU5P[c] + CQ2[p]'

'RU5P[c] <==> XU5P[c]'

'RU5P[c] <==> R5P[c]'

'(2) GLYOX[c] + (3) ATP[p] + (2) NADH[p] -> 3PG[p] + CQ2[p] + (3) ADP[p] + (2) NAD[p]'

OAA[c] + ATP[p] -> CQ2[p] + PEP[c] + ADP[p]'

'PEP[c] + CQ2[p] -> OAA[c] '

Penxosomal metabolism

'GLYOX[c] + ACCOA[c] -> MAL[c]'

'MAL[c] + NAD[p] <==> OAA[c] + NADH[p]'

OAA[c] + ACCOA[c] -> CIT[c] '

'CIT[c] -> GLYOX[c] + SUCC[m] '

'CIT[c] + ATP[p] -> ACCOA[c] + OAA[c] + ADP[p]'

Mitochondrial metabolism

'PYR[m] + NAD[p] + OAA[m] -> NADH[p] + C02[p] + CIT[m]'

'CIT[m] + NAD[p] <==> AKG[m] + C02[p] + NADH[p]'

'CIT[m] + NADP[p] <==> AKG[m] + C02[p] + NADPH[p]'

'AKG[m] + NAD[p] + ADP[p] -> SUCC[m] + C02[p] + ATP[p] + NADH[p]'

'SUCC[m] + NAD[p] -> FUM[m] + NADH[p]'

'FUM[m] -> MAL[m]' 'MAL[m] + NAD[p] -> OAA[m] + NADH[p]'

'MAL[m] + NAD[p] -> CQ2[p] + NADH[p] + PYR[m]'

Transporters

HAP[c] <==> DHAP[p]'

'MAL[c] <==> MAL[m]'

^■ G6P[p] <==> G6P[c]'

'3PG[p] <==> 3PG[c]'

'PYR[p] <==> PYR[c]'

'PYR[c] <==> PYR[m]'

'PEP[c] <==> PEP[p]'

'XU5P[p] <==> XU5P[c]'

OAA[c] + MAL[m] <==> MAL[c] + OAA[m]'

OAA[c] + CIT[m] <==> CIT[c] + OAA[m]'

Energy metabolism

'(9) Hv[p] + (2) NADP[p] + (3) ADP[p] -> (2) NADPH[p] + (3) ATPfpj ⁷

ΆΤΡ[ρ] -> ADP[p] + ATP_maint[p]'

'(2.4) ADP[p] + NADH[p] -> NAD[p] + (2.4) ATP[p]'

' -> Hv[p]'

' ATP_maint[p] -> '

'NADP[p] + NADH[p] <==> NADPH[p] + NAD[p]'

Extending correlation-based analysis for growth improvement to complex networks. Beyond the natural boundary of A. thaliana to grow, its potential to grow and accumulate specific trait compounds was explored. For this purpose, the set of elementary flux modes was analyzed with Flux Design to search for genetic targets towards improved performance (Melzer et al. (2009), BMC Syst. Biol. 3:120). It turned out that the high network complexity of the

compartmented model hampered straightforward target identification, similar to what is observed in other eukaryotes (Melzer et al. (2009), BMC Syst. Biol. 3:120). Therefore, the correlation analysis was narrowed down to specific subsets of modes, involving only biomass forming modes, the top 5% and the top 1 % modes with regard to growth. Shortly, these subsets were tested for their predictive power. In the case of A. thaliana, the best subset, i.e. the subset, which provided the highest number of statistically significant targets, was the top 5% subset of modes. In total, 10 correlation targets were identified. Four of these targets directly correlated to growth, as they uniquely produced a particular growth precursor, and were also found in other subsets and physiology types. Additionally, six further targets, all located in and around the mitochondrion, were predicted as potential growth targets. Moreover, genetic targets were predicted for ten biotechnologically interesting products. Generally, Arabidopsis exhibited a high natural capacity to over-produce each of the studied molecules.

8. Validation of a specific target identified by the method of the present invention The in-silico modeling by Flux Design revealed the starch synthesis reaction as potential overexpression target. In the context of the rice seed model, starch synthesis is defined as a lump reaction converting G6P into starch by utilizing energy in the form of ATP (cf. Table 7). Referring to the original pathway of starch synthesis, the only reaction leading from G6P to starch utilizing ATP as energy source was found to be catalyzed by the enzyme ADPglucose pyrophosphorylase (AGPase, also known as glucose-1 -phosphate adenylyltransferase, EC 2.7.7.27). Therefore, this enzyme was tested as a target for overproduction in rice.

Higher plant AGPases are heterotetramers composed of two closely related types of subunits (S2L2; Morell et al. (1987), Plant Physiol 85: 182; Okita et al. (1990), Plant Physiol 93: 785;

Preiss et al. (1991 ), Biochem Soc Trans 19: 539; Smith-White and Preiss (1992), J Mol Evol 34: 449; Ballicora et al. (2004), Photosynth Res 79: 1 ). In order to simplify the expression of a heteromeric enzyme, an alternative bacterial version was chosen, as most bacterial ADP-GIc PPases are homotetramers composed of four identical subunits (Haugen et al. (1976) J Biol Chem 251 : 7880; Ballicora et al. (2003), Microbiol Mol Biol Rev 67: 213).

AGPase has been previously successfully tested in rice (Sakulsingharoj et al., 2004 Plant Science 167 (6) 1323) and was used here. Biochemical data, available for bacterial AGPases suggested the E. coli enzyme as the one with highest turnover and affinity towards its substrates, and as such was selected for the approach. Biochemical studies conducted with a modified version of the E. coli glgC gene with R67K, P295D, and G336D suggested the removal of enzyme inhibition by Pi as well as constitutive activity without the requirement of an activator, therefore, the modified sequence was designed as from "Rice Science: Innovations and Impact for Livelihood" (Mew et al., 2003, IRRI Conference Report: "Rice Science: Innovations and Impact for Livelihood", Chapter: Manipulating starch and storage protein biosynthesis during endosperm development to increase rice yield, p345, ISBN: 9712201848) was selected here.

8.1 Cloning of the target-encoding nucleic acid sequence The ORF encoding a R67K, P295D, and G336D mutein of AGPase from E. coli (protein: SEQ ID NO: 2; polynucleotide: SEQ ID NO: 1 ) was synthesized by PCR by a commercial provider, including gateway cloning primers (attB sequences) at both ends. At the 5' end of this nucleic acid sequence, a sequence encoding a plastidial target sequence (SEQ ID NO: 5) was added during synthesis to generate a DNA fragment having the nucleic acid sequence of SEQ ID NO: 3, encoding a polypeptide having the sequence of SEQ ID NO: 4. The DNA fragment was used in the first step of the Gateway procedure, the BP reaction, during which the fragment recombined in vivo with the pDONR201 plasmid to produce, according to the Gateway terminology, an "entry clone", pAGPase. The entry clone comprising SEQ ID NO: 3 was then used in an LR reaction with a destination vector used for Oryza sativa transformation. This vector contained as functional elements within the T-DNA borders: a plant selectable marker; a screenable marker expression cassette; and a Gateway cassette intended for LR in vivo recombination with the nucleic acid sequence of interest already cloned in the entry clone. A rice Prolamin promoter (pPROL, SEQ ID NO: 6) for endosperm specific expression was located upstream of this Gateway cassette.

After the LR recombination step, the resulting expression vector pPROL::AGPase (Fig. 1 ) was transformed into Agrobacterium strain LBA4044 according to methods well known in the art and described above.

8.3. Phenotypic evaluation 8.3.1 Evaluation setup

35 to 90 independent TO rice transformants were generated. The primary transformants were transferred from a tissue culture chamber to a greenhouse for growing and harvest of T1 seed. Nine events, of which the T1 progeny segregated 3:1 for presence/absence of the transgene, were retained. For each of these events, approximately 10 T1 seedlings containing the transgene (hetero- and homo-zygotes) and approximately 10 T1 seedlings lacking the transgene (nullizygotes) were selected by monitoring visual marker expression. The transgenic plants and the corresponding nullizygotes were grown side-by-side at random positions.

Greenhouse conditions were of shorts days (12 hours light), 28°C in the light and 22°C in the dark, and a relative humidity of 70%. Plants were grown under non-stress conditions and were watered at regular intervals to ensure that water and nutrients were not limiting and to satisfy plant needs to complete growth and development.

8.3.2 Statistical Analysis A liner model using generalised least squares was used as a statistical model for the evaluation of plant phenotypic characteristics. The data comprised observations from several plants of number of events transformed with the gene of interest. A significance test was carried out to verify an overall effect of the gene, also known as a global gene effect and event level effects. The threshold for significance for a true global gene effect was set at a 5% probability level. Significant effects were obtained in several events for several phenotypes (yield parameters) pointing to a gene effect, meaning that it is not only the mere presence or position of the gene that is causing the differences in phenotype.

8.3.3 Parameters measured

From the stage of sowing until the stage of maturity the plants were passed several times through a digital imaging cabinet. At each time point digital images (2048x1536 pixels, 16 million colours) were taken of each plant from at least 6 different angles. Biomass-related parameter measurement

The plant aboveground area (or leafy biomass) was determined by counting the total number of pixels on the digital images from aboveground plant parts discriminated from the background. This value was averaged for the pictures taken on the same time point from the different angles and was converted to a physical surface value expressed in square mm by calibration.

Experiments show that the aboveground plant area measured this way correlates with the biomass of plant parts above ground. The above ground area is the area measured at the time point at which the plant had reached its maximal leafy biomass.

Seed-related parameter measurements

The mature primary panicles were harvested, counted, bagged, barcode-labelled and then dried for three days in an oven at 37°C. The panicles were then threshed and all the seeds were collected and counted. The seeds are usually covered by a dry outer covering, the husk. The filled husks (herein also named filled florets) were separated from the empty ones using an air- blowing device. The empty husks were discarded and the remaining fraction was counted again. The filled husks were weighed on an analytical balance. The total number of seeds was determined by counting the number of filled husks that remained after the separation step. The total seed weight was measured by weighing all filled husks harvested from a plant. The total number of seeds (or florets) per plant was determined by counting the number of husks (whether filled or not) harvested from a plant. Thousand Kernel Weight (TKW) was extrapolated from the number of seeds counted and their total weight. The Harvest Index (HI) in the present invention is defined as the ratio between the total seed weight and the above ground area (mm ² ), multiplied by a factor 10 ⁶ . The number of flowers per panicle as defined in the present invention is the ratio between the total number of seeds over the number of mature primary panicles. The "seed fill rate" or "seed filling rate" as defined in the present invention is the proportion (expressed as a %) of the number of filled seeds (i.e. florets containing seeds) over the total number of seeds (i.e. total number of florets). In other words, the seed filling rate is the percentage of florets that are filled with seed.

8.4. Results of the phenotypic evaluation of the transgenic plants

The results of the evaluation of transgenic rice plants in the T1 generation and expressing a nucleic acid encoding the AGPase polypeptide of SEQ ID NO: 4 under non-stress conditions are presented below in Table 10. Plant phenotype was not significantly affected, except for improved seed yield and, to a lesser extent, seed size.

Table 10: Data summary for transgenic rice plants; for each parameter, the overall percent change is shown for T1 generation plants.

Parameter Overall change Trend of effect

in %

AreaMax -1 .0 neutral

Time to Flower 2.2 neutral

Total weight of seeds per plant 7.9 positive

TKW 2.0 positive