Oilseed rape seeds with ablated defence cells of the glucosinolate-myrosinase system. Production and characteristics of double haploid MINELESS plants of Brassica napus L.

Oilseed rape and other crop plants of the family Brassicaceae contain a unique defence system known as the glucosinolate-myrosinase system or the 'mustard oil bomb'. The 'mustard oil bomb' which includes myrosinase and glucosinolates is triggered by abiotic and biotic stress, resulting in the formation of toxic products such as nitriles and isothiocyanates. Myrosinase is present in specialist cells known as 'myrosin cells' and can also be known as toxic mines. The myrosin cell idioblasts of Brassica napus were genetically reprogrammed to undergo controlled cell death (ablation) during seed development. These myrosin cell-free plants have been named MINELESS as they lack toxic mines. This has led to the production of oilseed rape with a significant reduction both in myrosinase levels and in the hydrolysis of glucosinolates. Even though the myrosinase activity in MINELESS was very low compared with the wild type, variation was observed. This variability was overcome by producing homozygous seeds. A microspore culture technique involving non-fertile haploid MINELESS plants was developed and these plants were treated with colchicine to produce double haploid MINELESS plants with full fertility. Double haploid MINELESS plants had significantly reduced myrosinase levels and glucosinolate hydrolysis products. Wild-type and MINELESS plants exhibited significant differences in growth parameters such as plant height, leaf traits, matter accumulation, and yield parameters. The growth and developmental pattern of MINELESS plants was relatively slow compared with the wild type. The characteristics of the pure double haploid MINELESS plant are described and its importance for future biochemical, agricultural, dietary, functional genomics, and plant defence studies is discussed.

Introduction

The presence of myrosin cells and the binary system of the glucosinolate–myrosinase are typically characteristic of the Brassicaceae family. The role of the glucosinolate–myrosinase system, also referred to as ‘the mustard oil bomb’, has been well established in defence-related responses of cruciferous plants (Bones and Rossiter, 1996; Rask ; Kissen ; Wittstock and Burow, 2010; Winde and Wittstock, 2011). Myrosinase (β-thioglucosidase; EC 3.2.1.147) constitutes a group of isoenzymes that have been reported to be present in myrosin cells of seeds, seedlings, and mature tissues of plants (Guignard, 1890; Bones and Iversen, 1985; Thangstad ; Höglund ; Kissen ). Myrosin cells are distributed throughout plant tissues (Bones ; Andréasson ), act as toxic mines for pests and diseases (Matile, 1980; Lüthy and 1984; Bones and Rossiter, 1996), and are considered to be of major biological importance. Upon tissue disruption by insect herbivores, myrosinase hydrolyses glucosinolates to give isothiocyanates (ITCs), thiocyanates, nitriles, epithionitriles, and oxazolidine-thiones (OZTs) (Bones and Rossiter, 1996, 2006; Grubb and Abel, 2006; Halkier and Gershenzon, 2006; Hopkins ). This triggering of the myrosinase–glucosinolate system is known as a ‘mustard oil bomb’ or, more recently, as a toxic mine, and this term will be used herein. The ecological significance of toxic mines in Brassica seeds and seedlings has, for example, been demonstrated by challenging B. juncea cotyledons during seedling development against the generalist herbivore, Spodoptera eridania (Wallace and Eigenbrode, 2002); as an allelochemical in B. nigra (Lankau and Strauss, 2007); and testing of seed nutritional quality against the yellow meal worm/common beetle generalist (Tenebrio molitor) (Davis , 1983; Eriksson ).

Cruciferous plants synthesize glucosinolates, a class of nitrogen- and sulphur-containing secondary metabolites that share a core structure consisting of a β-thioglucoside moiety and a sulphonated oxime, but differ due to a variable side chain derived from one of several amino acids (Grubb and Abel, 2006; Yan and Chen, 2007). The formation of glucosinolate hydrolysis products depends upon the nature of the glucosinolates, the reaction conditions, and the occurrence of protein cofactors. This includes epithiospecifier protein (ESP), nitrile-specifier proteins (NSPs), and thiocyanate-forming protein (TFP), which in turn is dependent on ferrous/ferric ions (Foo ; Lambrix ; Kissen and Bones, 2009). ESP diverts the aglycone towards nitriles or epithionitriles, rather than ITCs (Bones and Rossiter, 1996, 2006; Burow , 2007; Matusheski ; Kissen and Bones, 2009). The conversion of alkenyl glucosinolates to epithionitriles by ESP was observed to be ferrous ion dependent (Zabala ). NSPs promote simple nitrile formation at physiological pH values, but do not catalyse epithionitrile or thiocyanate formation (Burow ; Kissen and Bones, 2009; Wittstock and Burow, 2010). In the presence of myrosinase and TFP, thiocyanates are formed with only the following glucosinolates: allylglucosinolate, benzylglucosinolate, and 4-methylthiobutylglucosinolate (Burow ; Wittstock and Burow, 2010).

The Arabidopisis myrosinases have been well characterized, with TGG1, TGG4, and TGG5 showing activation in the range of 1–5 mM ascorbic acid in vitro, though higher concentrations inhibited enzyme activity (Andréasson ; Wittstock and Burow, 2010). Moreover, TGG1, TGG4, and TGG5 were observed to be highly stable across broad pH and temperature ranges.

The glucosinolate–myrosinase system is a defence against both biotic and abiotic stress responses (Zhao ; Ahuja ; Winde and Wittstock, 2011). Glucosinolates have been shown to be modulated by abiotic stress factors where, for example, the indolic class of glucosinolates decreases in concentration in B. napus after exposure to elevated CO2 (Himanen ; Ahuja ). ITCs, one of the glucosinolate degradation products, are cues for the plant localization and feeding stimulants for herbivores specialized on Brassicaceae (Rosa ; Raybould and Moyes, 2001; Tripathi and Mishra, 2007; Pope ; Kissen ; Ahuja ; Björkman ). In the case of generalist herbivores, the defensive function of this dual system depends on glucosinolate breakdown (Wittstock and Burow, 2010). The insect herbivory of the generalist Trichoplusia ni is strongly deterred by higher glucosinolate levels, faster breakdown rates, and specific chemical structures (Kliebenstein ). In contrast, Plutella xylostella herbivory was not correlated with variation in the glucosinolate–myrosinase system. In B. juncea, lines that have a high glucosinolate and myrosinase content were defended better against the generalist S. eridania compared with lines with reduced concentrations of glucosinolate and lower expression of myrosinase (Li ). In contrast to insect herbivory, the glucosinolate breakdown products such as thiocyanates, vinyl-OZT, and ITCs, produced during the processing of oilseedrape meal (Mawson ), have a harmful effect on animal thyroid function. Hydrolysis of progoitrin [(2R)-hydroxy-3-butenyl glucosinolate], one of the major glucosinolates in oilseed rape, normally produces oxazolidine-2-thiones, the most important being the 5-vinyl-1,3-oxazolidine-2-thione (Fenwick and Heaney, 1983; Mabon ). The use of animal feed containing these glucosinolates has a negative effect on animal nutrition because of their goitrogenic properties (Elfving, 1980; Mithen, 2001). In order to overcome the goitrogenic problem of oilseed rape, plant breeders developed an oilseed crop called double zero Canola (a contraction of ‘Canadian’ and ‘oil’) (Fahey ). However, a radical approach has been adopted to overcome the goitrogenic problem and a transgenic ablation strategy of myrosinase has been developed with the aim of limiting glucosinolate hydrolysis in ruminants. This produced so-called transgenic MINELESS plants for B. napus cv. Westar (Borgen ). Myrosin cells were ablated by using the myrosin Myr1.Bn1 promoter and expressing the cytotoxic RNase barnase in seed myrosin cells. The designation MINELESS was assigned to highlight the genetic ablation of myrosin cells. Transgenic MINELESS plants seem to display significant advantages in many ways. First, these seeds can be used for trials to evaluate their potential as low toxicity–high protein feedstuffs. Secondly, they can be used to evaluate the role of the glucosinolate–myrosinase system in plant–insect interactions using a crop plant rather than the model Arabidopsis thaliana.

Although the myrosinase activity of transgenic MINELESS seeds was low in comparison with the wild-type B. napus cv. Westar, there was considerable variation amongst single seeds (Borgen ). This was expected due to the segregation of the transgenes and possibly also due to cell- and time-specific variation of the Myr1Bn1 promoter. In order to overcome the problem of seed variability, it was decided to use microspore culture, a well-known technique for the production of pure double haploid (DH) plants of transgenic MINELESS B. napus. The basis of the microspore culture lies in microspore embryogenesis, and is based on the ability of a single haploid cell, the microspore, to de-differentiate and regenerate into a whole plant, after being exposed to different stresses (Shariatpanahi ). Haploid and double haploid plants regenerated from microspore embryos are invaluable breeding tools to acquire complete homozygosity in a single generation (Kučera ). With a view to carrying out biochemical and molecular research with regard to androgenesis and embryogenesis, and due to the generally high response of B. napus genotypes, the microspore culture of B. napus has become an important model system (Custers ; Friedt and Snowdon, 2010). Nevertheless, not all the genotypes respond in the same way to induced culture conditions (Malik ). Since the isolated microspores and microspore-derived embryos are haploid, the transformation of these explants and chromosome doubling can produce diploid transgenic plants in only one generation. These plants are homozygous not only for the transgene, but also for the entire genome (Abdollahi ).

Alterations in plant morphology and other related parameters due to the changed genetic environment in MINELESS plants might have occurred, as the genetic, environmental, agronomic, and physiological factors or their interaction are suggested to contribute towards yield and its formation (Thurling, 1974; Diepenbrock, 2000; Sidlakaus and Bernotas, 2003; Shi ; Zhang ). Eighty-five quantitative trait loci (QTLs) for seed yield along with 785 QTLs for eight yield-associated traits, from 10 natural environments and two related populations of B. napus, were identified (Shi ). In evaluating B. napus DH lines and their corresponding parents for silique traits, the additive effects were demonstrated to be more important than epistatic effects for silique length (Zhang ). However, the additive effects together with epistasis were responsible for genetic variations of seeds per silique and seed weight. The increase in duration of the vegetative growth period and precipitation rate resulted in higher seed yield, while the increase in daily temperatures and growing days showed a negative effect on seed yield of spring oilseed rape (Sidlakaus and Bernotas, 2003). Oilseed rape is important not only due to the presence of the glucosinolate–myrosinase system, but also as an important agricultural crop. In oilseed rape, growth attributes such as biomass production, leaf area, number of siliques per plant, seed number per silique, and silique length have been considered to be the important determinants for yield analysis (Diepenbrock, 2000; Zhang ). Analysis of growth attributes assists in estimating plant growth at various developmental stages, which finally accounts for yield variation. Moreover, seed yield is considered as a complex character, and several components may contribute to it by having positive or negative effects upon the trait.

Thus the main objective of the present study was to produce and characterize DH seeds of the transgenic MINELESS plants. In order to accomplish this objective, experiments were performed to determine the difference and importance of DH transgenic MINELESS seeds and plants to the parental B. napus cv. Westar (designated as the wild type here). Homozygous seeds and wild-type seeds were characterized at several levels, and plants were compared for growth and yield parameters. The study confirmed production of pure DH MINELESS B. napus seeds, with a low and constant myrosinase activity. The results also revealed changes in glucosinolate concentrations and their hydrolysis products in MINELESS seeds, emphasizing the modification of the glucosinolate–myrosinase defence system.

Materials and methods

Plant material, microspore isolation, embryo culture, kanamycin selection, plant regeneration, colchicine application, and production of double haploid MINELESS seed

Microspores were prepared from the donor plants of transgenic B. napus MINELESS and wild-type B. napus cv. Westar under in vitro culture conditions, as previously described (Hansen, 2003). Plants were grown in pots with fertile soil in environmentally controlled rooms, with a 16 h photoperiod and ∼200 μmol m−2 s−1 photosynthetically active radiation at 15 °C in the light and 10 °C in the dark. Wild-type and MINELESS plants were kept in separate rooms to avoid cross-pollination. One week prior to microspore isolation, the room temperature was lowered to between 5 °C and 10 °C. Young buds (3.0–4.5 mm) from healthy plants were transferred to tea baskets, sterilized, rinsed, and microspores were released in NLN-13 medium (Lichter, 1982; Cao ; Hansen, 2003; Zhang ) to give a final suspension of ∼40 000–50 000 microspores ml−1 medium. A heat shock (32 °C) treatment of 48 h was given to the freshly isolated microspores, with further incubation at 24 °C for 21 d. Visual observations on microspores and developing embryos were made under an inverted microscope. Microspore-derived embryos (MDEs) were desiccated for ∼2–3 weeks and the direct regeneration involved transfer of MDEs to solid Gamborg B5/MS medium (pH 5.8) (Murashige and Skoog, 1962) and formation of plantlets. Some of the desiccated MDEs were tested for kanamycin resistance. Kanamycin selection of MINELESS MDEs was performed with both types of medium, MS (100 μg ml−1) and B5 (200 μg ml−1).

The plants were raised under aseptic conditions by transferring plantlets on solid agar containing MS medium, 3% (w/w) sucrose (pH 5.8) at 22 °C under a light regime of 16 h light/8 h photoperiod and at a light intensity of 70–80 μmol m−2 s−1. Plants were hardened with liquid 1/2 MS salt mixture. For DH generation, plants were transferred to autoclaved soil under the same temperature and light/dark cycles at a light intensity of 50–60 μmol m−2 s−1. Cotton swabs were dipped in freshly prepared 0.1% colchicine solution and placed on internodes of young plants for an overnight treatment. The diploid flowering stalks were selfed by bagging. Plant height was measured for 10 plants of each of the wild-type and MINELESS plants. Sterile/non-fertile and fertile flowers were observed under a stereomicroscope (Nikon SMZ1500, Nikon) and recorded using a 990 digital camera. Selfed seeds were harvested from DH MINELESS plants after they attained complete maturity.

The MINELESS MDEs, DH MINELESS seeds, and 5-day-old cotyledons (5-d COTY) were compared with the wild-type MDEs, seeds, and 5-d COTY for myrosinase enzyme activity. Myrosinase and ESP expression, and glucosinolate and hydrolysis product analysis was carried out only on seeds. The structural analysis of myrosin cells from seeds was carried out by using immunocytochemical techniques and image analysis by confocal microscopy. In addition, sections from seeds of both the wild type and MINELESS were stained with toluidine blue and subjected to image analysis by light microscopy. Wild-type and MINELESS plants were grown and compared for growth, development, and yield attributes.

Extraction of myrosinase, specific myrosinase activity, and protein assays

Samples from single MDEs, seeds, and 5-d COTY of the wild type and MINELESS were crushed and proteins were extracted in 100 μl of imidazole-HCl buffer (10 mM, pH 6.0). The samples were centrifuged at 4 °C for 20 min, and supernatants were transferred to dialysis membranes (12 000–14 000 Da MWCO) (Spectra/Por), dialysed overnight in imidazole-HCl buffer, centrifuged at 4 °C for 20 min, and supernatants were used for myrosinase activity and protein assays. The specific myrosinase activity was measured in MDEs, DH seeds, and 5-d COTY using the GOD-Perid assay (Bones and Slupphaug, 1989). Myrosinase assays were carried out using citrate buffer (50 mM, pH 5.5), sinigrin (15 mg ml−1), and GOD-Perid reagent (Roche, Basel, Switzerland). In order to calculate the specific myrosinase activity, the total protein content of samples was measured using Bradford reagent (BioRad Laboratories, UK). The specific myrosinase activity is described as nmol glucose generated min−1 mg−1 protein. The protein extracts of MDEs, seeds, and 5-d COTY were further processed for SDS–PAGE and immunoblot analysis.

SDS–PAGE and immunoblot analysis

A 10 μg aliquot of protein samples was solubilized in standard SDS buffer, prior to separation by SDS–PAGE using 12% TRIS-HCl–SDS gels, in a BioRad vertical electrophoresis system. Biotinylated protein molecular weight markers (SP-1400, Vector) were used as standards. Gels were blotted for 1 h onto 0.2 μm nitrocellulose membrane (Trans-blot® Transfer Medium) using a Minitransfer Trans-Blot Electrophoretic Transfer cell module as described by the manufacturer (BioRad). The blots were probed with a myrosinase-specific monoclonal antibody 3D7 (Lenman ), followed by a rabbit anti-mouse Ig (G+A+M) (P161; Dako, Glostrup, Denmark) conjugated with biotin and enhanced by streptavidin conjugated with horseradish peroxidase (Vector, SA-5004). A chemiluminescent detection system was used for visualizaion (Super Signal West Pico, Pierce, Rockford, IL, USA). ESP was detected by an anti-ESP polyclonal antibody characterized earlier (Foo ) as in a previous study (Borgen ).

Analysis of myrosinase isoforms

Following isoelectric focusing (IEF; Bones and Slupphaug, 1989; Bones ) myrosinase isoforms were detected by barium sulphate deposition. Wild-type and MINELESS seeds were weighed and crushed with a pestle and mortar in liquid nitrogen and proteins were extracted in 50 mM TRIS-HCl buffer, pH 8.0. IEF (PhastGel 4–6.5) was carried out using a Pharmacia PhastSystem. Following separation of the isoforms the gel was transferred to a glass plate, and retained on the gel bed of the PhastSystem. A solution (0.5 ml) of 12 mM sinigrin, 1 mM ascorbic acid, and 18 mM barium acetate in 50 mM citrate buffer, pH 5.5 was added to the gel. The gel was covered with another glass plate without any air bubbles and kept for 30–60 min at 40 °C; images were recorded using a GelDoc system (Biorad).

Measurement of soluble and insoluble myrosinase enzyme activity

Soluble and insoluble myrosinase activity from seeds was determined by enzyme kinetic analysis (Travers-Martin ). The wild-type and MINELESS seeds were finely pulverized in liquid nitrogen with a Retsch mill and samples were extracted in TRIS–EDTA buffer (200 mM TRIS-HCl, 10 mM EDTA, pH 5.5). For the measurement of soluble myrosinase activity, endogenous glucosinolates were removed by passing the supernatant through Sephadex A25 (Sigma-Aldrich, St Louis, MO, USA) columns. The pellet (insoluble myrosinase fraction) obtained after the first extraction was dissolved in the extraction buffer and assayed by adding 2 mM glucosinolate sinigrin and a mixture of glucose oxidase peroxidase, 4-aminoantipyrine, phenol, and imidazole as colorimetric indicators (Travers-Martin ). The coloured product formed was measured at 492 nm.

Confocal laser scanning microscopy and immunolabelling

Seeds of the wild-type and DH MINELESS plants were incubated in water for 24 h at room temperature, fixed, dehydrated, and embedded as described (Thangstad ). Blocks were sectioned (1 μm and 600–700 Å) and observed using fluorescence and confocal microscopes. For light microscopy, the sections were stained with toluidine blue. The slides were examined and photographed with a research microscope (E800; Nikon, Tokyo, Japan) equipped with a digital camera (SPOT RT; Diagnostic Instruments, Burroughs, MI, USA). Semi-thin sections (2–3 μm) were also used for the detection of myrosinase with the polyclonal antibody K089 directed against myrosinase (Thangstad ). Positive cells were visualized with fluorescein isothiocyanate (FITC)-conjugated streptavidin (DAKO, Denmark). Confocal laser scanning microscopy was performed on a Zeiss LSM 510 (Zeiss, Jena, Germany). FITC fluorescence excited by an argon laser excitation light (488 nm) and differential interface contrast (DIC) images were captured using LSM510 3.0 software.

Liquid chromatography–mass spectrometry (LC-MS) analysis of glucosinolates

The total glucosinolate concentrations and glucosinolate profile were quantified according to the described method (Heaney ). Samples were dissolved in deionized water and were analysed by HPLC-DAD (Agilent HP 1100 Series) coupled with an Agilent 1100 Series LC/MSD trap mass spectrometer (Agilent Technologies, Palo Alto, CA, USA) with an APCI interface. The desulpho-glucosinolates were separated on a Supelcosil LC 18 (250×2.1 mm; 5 μm) (Supelco, Bellefonte, PA, USA) at a flow rate of 0.30 ml min−1 at room temperature and detected at 229 nm. The mobile phase consisted of solvent A (deionized water) and solvent B (acetonitrile; Merck, Darmstadt, Germany). The following gradient procedure was used: 0–2 min, 3% B; 2–17 min, 3–40% B; 17–22 min, 40% B; 22–22.10 min, 40–100% B; 22.10–32 min, 100% B; 32–32.10 min, 3% B; 32.10–60 min, 3% B. The mass spectrometer was configured in the positive ion chemical ionization mode. The APCI settings were: nebulizer pressure, 60 psi; drying gas flow, 5 l min−1; drying gas temperature, 350 °C; APCI vap temperature, 400 °C; Corona current 4000 nA. The individual glucosinolates were identified using a diode array UV spectrometer and by mass spectrometry. The identity of the compounds was confirmed by comparison with retention times, UV spectra, and MS spectra of authentic standards of desulpho-glucosinolates (John Rossiter's laboratory, Imperial College, London). The glucosinolate data were analysed and processed using Agilent Chem Station software. The data were corrected for UV response factors for different types of glucosinolates relative to the response factor of benzyl glucosinolate (glucotropaeolin) used as an internal standard, as described in the European Community (EC, 1990) standard.

Gas chromatography–MS (GC-MS) analysis of glucosinolate hydrolysis products

Single seeds were crushed with a micropistil in 250 μl of distilled H2O in a 2 ml Eppendorf tube, and another 250 μl of distilled H2O was added. The mixture was immediately pipetted into a 2 ml screw top vial with a PTFE/silicone septum and left for 2 h at ambient temperature for hydrolysis. A mixture of 0.5 ml of hexane:dichloromethane (3:2) with an internal standard (12 μg of butyl-ITC) was injected through the septum into the vial, and the sample was vortexed for 50 s. After centrifugation at 3100 rpm for 2 min, the solvent phase was pipetted into a 2 ml screw top vial with a PTFE/silicone septum, and concentrated under a nitrogen flow to a volume of 50 μl. Samples were stored at 4 °C in the dark prior to GC-MS analysis the following day. All GC-MS analyses were carried out on a Varian Star 3400 CX gas chromatograph coupled with a Varian Saturn 3 mass spectrometer. The gas chromatograph was equipped with a Supelco SPB-1 capillary column (30×0.25 m i.d.; 0.25 μm film thickness), the flow rate of the carrier gas He (15 psi) was 50 ml min−1, and the injector, transfer line, and detector temperatures were 220, 240, and 240 °C, respectively. Mass spectra were acquired in EI mode, and a mass range of m/z 39–250 was recorded.

The quantification of compounds was based on peak area detection of the internal standard by calculating linear standard curves based on 1× to 40× concentrated standard solutions generated under nitrogen flow for the analysis of the data of solvent extracts (corresponding to an internal standard concentration of butyl-ITC ranging from 95 ng to 3800 ng). All glucosinolate hydrolysis products were tentatively identified by a combination of a mass spectra database search (NIST/EPA/NIH Mass Spectral Library-NIST05, 2005), use of the deconvolution and identification software AMDIS (vers. 2.64, 2006), and comparison with mass spectral data from the literature. Most of the extracted hydrolysis products identified have been reported before (Brown and Morra, 1995; Smolinska ), except for the indolic nitriles 1-methoxy-indoleaceto-NIT-nitrile (NIT) and indoleaceto-NIT-NIT.

Analysis of growth, development, and yield parameters

Growth and yield analysis studies were conducted under contained greenhouse conditions (S3 security class) at the Plant BioCenter, Norwegian University of Science and Technology, Trondheim, Norway. The experiment was designed as a comparative study between wild-type and MINELESS plants for phenological/morphological, growth and development parameters during plant development and yield attributes at plant maturity. The seeds of wild-type and DH MINELESS plants were sown in pots containing soil/peat mixture P-jord (Hasselfors Garden, Örebro, Sweden). The plants (one plant per pot) were raised in environmentally controlled growth rooms (S3 security class) with a 16 h photoperiod provided by daylight and cool and warm white fluorescent tubes. Day and night temperatures were 22 °C and 18 °C, respectively. Plant morphological and phenological characteristics, such as the dates of seedling emergence, leaf appearance, onset of flowering, plant height, and leaf traits, which comprised leaf number, leaf area, and total leaf weight plant per plant for four development stages were recorded. The four growth stages studied were: (i) before initiation of flowering 30 DAE (days after emergence); (ii) after initiation of flowering (44 DAE); (iii) >50% flowering (62 DAE); and (iv) the end of flowering (74 DAE). Leaf area was measured for individual leaves from both the wild-type and MINELESS plants using a scanner and by analysing images using software Compu Eye, Leaf & Symtom Area, as designed (Bakr, 2005). Leaf area was measured from the leaves present on both the main stem and branches. Fresh and dry matter accumulation data were recorded for the main stem, leaves, branches, and roots at these growth stages. In order to measure the dry weights, leaves, stems, branches, and roots were dried at 60 °C for 72 h. The percentage moisture was calculated as a percentage of the fresh weight.

Following the growth period when the plants had attained maturity, yield data were recorded for 15 plants for each of the wild-type and MINELESS plants: plant height, number of primary branches per plant, number of siliques per plant, silique length and diameter, undeveloped/shrivelled seeds per silique, total number of seeds per silique, biomass of the main stem, branches, leaves, and roots, above-ground biomass, total plant biomass, seed yield per plant, 1000-seed weight of developed seeds, and harvest index (HI) per plant. The data on silique length, diameter, and number of seeds per silique are an average of 10 siliques per plant.

Data analysis

The plant height of microspore-derived plants (Fig. 1), myrosinase activity, glucosinolates and glucosinolate hydrolysis products of the wild-type and MINELESS seeds were compared by using the t-test from SigmaPlot 11. However, the growth, development, and yield data have been statistically analysed by analysis of variance (ANOVA; with general linear models in SYSTAT v. 11, Systat software Inc., 2004), where a model with genotype (wild type/MINELESS) and developmental stage as fixed effects has been applied. Statements about statistical significance of main effects and the interactions refer to the F-test in such an ANOVA. For analysis of growth and developmental data, the experiments were designed in such a way that the main effects of genotype are based on observation of 12 plants, and of developmental stage on six plants, while the interactions are based on three observations per plant. Also, Pearson's correlation (r) was used to show the relationship between yield attributes.

Fig. 1.

Microspore-derived homozygous wild-type and MINELESS plants. (A) Two-celled stage (dyad) after the first division. (B) Twenty-one-day-old transgenic MINELESS embryos in NLN-13 medium. (C) Plant regeneration on MS medium. (D) Regenerated and surviving MINELESS embryos on B5 kanamycin (200 μg ml−1) selection medium. (E) Wild-type and transgenic MINELESS haploid plants showing the difference in plant height, where M and W denote the MINELESS and wild type, respectively. (F) A fully fertile flower of a DH MINELESS plant showing six anthers as observed under a stereo-microscope. (G) Plant height of wild-type and MINELESS homozygous plants (n=10) (**P <0.001). Bars=200 μm in (A) and 2 mm in (F). Error bars represent the SE.

Results

Production of DH MINELESS seeds

Microspore culture was successfully carried out for the transgenic MINELESS and wild-type B. napus plants. Immediately after isolation, microspores were checked for the presence of late uninucleate and early binucleate stages by microscopy. The first cell division in isolated microspores was visible after 3–4 d of culture (Fig. 1A), followed rapidly by several more divisions over the next 5–6 d. After 21 d of microspore isolation, embryos were mature enough for regeneration transfer. MDEs had a distinct elongated root and shoot axis with two conspicuous cotyledons surrounding the shoot apex (Fig. 1B). The MDEs were kept overnight in NLN13 with addition of abscisic acid. This was followed by 2–3 weeks of desiccation in the dark and they were then transferred to B5 medium under light. All MDEs regenerated and reached the plantlet stage after transfer to MS medium (Fig. 1C). Most of the plantlets developed vigorous shoots and successfully developed into plants. Kanamycin concentrations (100 μg ml−1 and 200 μg ml−1) on both MS and B5 media were sufficient for the selection of transgenic MINELESS MDEs, which survived on the selection medium for 4–5 weeks (Fig. 1D). Microspore-derived plants were completely sterile/non-fertile. The wild-type and the transgenic MINELESS plants were morphologically different in terms of height and vegetative growth. MINELESS plants were shorter with enhanced vegetative growth as compared with the wild type (Fig. 1E). The average height of MINELESS plants was 95 cm as compared with 130 cm for the wild type (Fig. 1G). The application of 0.1% colchicine doubled the chromosome number of microspore-derived MINELESS plants. A homozygous population of MINELESS plants was raised in a growth room. Approximately 60% of the soil-grown microspore-derived MINELESS plants attained full fertility upon colchicine application. Fully matured siliques from DH MINELESS plants led to a good seed set, and ultimately provided sufficient DH seeds to carry out different analyses. Morphological observations were made for non-fertile and fertile flowers using a stereo-microscope. Flowers from DH MINELESS plants had fully developed anthers and long stamen filaments (Fig. 1F).

DH MINELESS seeds show negligible or very low and stable myrosinase activity

Specific myrosinase activity in wild-type MDEs ranged from 29.0 to 396.4 nmol glucose min−1 mg−1 protein (Fig. 2A). MINELESS MDEs showed lower activity over a narrower range (0–87.4 nmol glucose min−1 mg−1 protein). The specific myrosinase activity in wild-type seeds ranged from 118.6 to 426.3 nmol glucose min−1 mg−1 protein (Fig. 2B). In contrast, MINELESS seeds showed activity ranging from 0.0 to 6.9 nmol glucose min−1 mg−1 protein. A low specific myrosinase activity (0.0 – 19.4 nmol glucose min−1 mg−1 protein) was also observed for the 5-d COTY MINELESS (Fig. 2C). This was in contrast to the much higher specific myrosinase activity in 5-d COTY wild type, varying from 77.1 to 345.8 nmol glucose min−1 mg−1 protein. Wild-type and MINELESS plants differed significantly in terms of the specific myrosinase activity in MDEs (P <0.05), single seeds, and single 5-d COTY (both P <0.001).

Fig. 2.

Distribution of specific myrosinase activity in wild-type and MINELESS single microspore-derived embryos (MDEs) (n=15), single seeds (n=15), and 5-day-old single cotyledons (5-d COTY) (n=15–30).

Myrosinase and ESP isoforms in MINELESS

Immunoblot analysis revealed the major bands of myrosinase polypeptide classes of B. napus (denoted as 65, 70, and 75 kDa) in wild-type samples of seeds (Fig. 3A). In contrast, none of these myrosinase isoforms was present in MINELESS single seeds (Fig. 3A). Immunoblot analysis using the anti-ESP polyclonal antibody detected two bands of ESP isoforms of 35 kDa and 39 kDa in all wild-type protein extracts (Fig. 3B) (Foo ; Borgen ). On the other hand, as has been reported previously (Borgen ), the 35 kDa ESP isoform was absent in MINELESS single seeds (Fig. 3B).

Fig. 3.

Immunoblot analysis of single seeds of the wild type and MINELESS. (A) Expression of myrosinase proteins was detected with anti-myrosinase 3D7 antibody. (B) Expression of ESP was detected with the anti-ESP antibody as described by Foo . Three different protein extracts of the wild type are designated W1, W2, and W3, and three different protein extracts of MINELESS are designated M1, M2, and M3. A 10 μg aliquot of total protein was loaded in each well.

In-gel myrosinase assay

Sulphate ions liberated after hydrolysis of sinigrin react with barium and precipitate in the gel as barium sulphate. Crude protein extracts of wild-type and MINELESS seeds separated by IEF showed the presence of three isoforms in wild-type and none in MINELESS seeds (Fig. 4).

Fig. 4.

In-gel barium sulphate assay after isoelectric focusing (IEF) of crude native protein extracts of seeds. Two crude protein extracts of wild-type seeds are designated W1 and W2, while two crude protein extracts of MINELESS seeds are designated M1 and M2. W1 and W2 show three expected bands for myrosinase isoforms while in MINELESS no band was observed.

Enzyme kinetic analysis shows very low soluble and insoluble myrosinase activity in MINELESS seeds

The soluble and insoluble myrosinase activities were assayed in extracts of seeds in a continuous mode by photometric quantification of the released glucose (Travers-Martin ). The average soluble myrosinase activity in MINELESS seeds was <10% of the activity found in wild-type seeds (P <0.001) (Fig. 5). Myrosinase activity in the insoluble fraction was lower in MINELESS compared with wild-type extracts (Fig. 5).

Fig. 5.

Mean myrosinase activity (soluble and insoluble) in wild-type and MINELESS seeds (n=12). Wild-type and MINELESS seeds differed significantly (**P <0.001) as determined by t-test. Error bars represent the SE.

Structural analysis of DH MINELESS seeds through light and confocal microscopy shows empty and degraded myrosin cells

The targeted myrosin cells from semi-thin sections of DH MINELESS seeds stained with toluidine blue appeared empty and degraded when observed under light microscopy (Fig. 6B–F). In contrast, the wild-type tissues showed the expected distribution of myrosin cells (Fig. 6A, C, D) (Bones ; Borgen ). Light microscopic observations revealed the myrosin cells of the wild type to be densely stained with toluidine blue and spatially dispersed in seed tissues (Fig. 6A). In contrast, the targeted myrosin cells of MINELESS seeds were not stained and appeared as empty holes in the seed tissue (Fig. 6B). The individual myrosin cells of wild-type seed displayed distinct nuclei surrounded by myrosin grains (Fig. 6C, D). In contrast, the individual myrosin cells of MINELESS seed appeared transparent or empty, indicating the degeneration of myrosin grains (Fig. 6E, F).

Fig. 6.

Structural analysis of myrosin cells from semi-thin sections of wild-type and MINELESS seeds stained with toluidine blue and observed under a light microscope. (A) Section of a wild-type seed. (B) Section of a MINELESS seed. (A and B) White circles show the densely stained and ablated myrosin cells in seed tissues of wild-type and MINELESS, respectively. (C and D) Individual myrosin cells of wild-type seeds. (C) Myrosin cell with a distinct central nucleus surrounded by six lightly stained myrosin grains visible as globular green vacuoles. (D) Myrosin cell with densely stained myrosin grains and a distinct central nucleus. (E and F) Individual ablated myrosin cells of MINELESS seeds. The ablated myrosin cells display an empty central vacuole and a few small peripheral hyaline vacuoles. Bars=2 mm (A and B), 10 μm (C and D), and 5 μm (E and F).

Confocal microscopic analysis of semi-thin sections with antibody K089 against myrosinase, followed by FITC-conjugated secondary antibody, provided specific labelling for the myrosin cells of wild-type seed (Fig. 7A). The wild-type seed showed intense labelling for myrosinase inside the myrosin cell. No specific labelling could be seen in the myrosinase-negative section of MINELESS seed (Fig. 7B). The empty myrosin cells of MINELESS seed were visible as black holes in the seed embryogenic tissue.

Fig. 7.

Structural analysis of myrosin cells from semi-thin sections of wild-type and MINELESS seeds labelled with the anti-myrosinase polyclonal antibody (K089) and FITC-conjugated secondary antibodies as observed under a confocal microscope. (A) Section of a wild-type seed showing the specific localization of densely labelled myrosinase in three myrosin grains of a highly fluorescent myrosin cell. (B) Section of a MINELESS seed showing no labelling of myrosinase in three ablated myrosin cells. (B1–B3) Ablated and empty myrosin cells are marked by the white star bursts. The exposure time in B was increased to visualize the semi- and fully ablated myrosin cells. Bars = 200 μm (A and B).

Analysis of glucosinolate concentrations and profile

The obtained glucosinolate profile of three major glucosinolates from defatted seed meal of the wild-type and MINELESS seeds is shown in Fig. 8. The qualitative analysis detected three major glucosinolates: 2-hydroxybut-3-enyl glucosinolate (progoitrin), 4-hydroxyindol-3-ylmethyl glucosinolate (4-hydroxyglucobrassicin), and indolyl-3-methyl glucosinolate (glucobrassicin) in both the wild-type and MINELESS seeds. Wild-type and MINELESS seeds differed significantly in terms of 2-hydroxybut-3-enyl glucosinolate (P <0.001), 4-hydroxy-3-indol-3-ylmethyl glucosinolate (P <0.001), and indolyl-3-methyl glucosinolate (P <0.05) (Fig. 8). All the detected glucosinolates were higher in MINELESS seeds than in those of the wild type. This also resulted in higher total glucosinolate concentrations in MINELESS seeds.

Fig. 8.

Glucosinolates from single seeds of the wild type and MINELESS (n=10) detected by HPLC analysis. Wild-type and MINELESS seeds differed significantly for glucosinolates, progoitrin, 4-hydroxyglucobrassicin, total glucosinolate content (**P <0.001), and glucobrassicin (*P <0.05) as determined by t-test. Error bars represent the SE.

Analyses of glucosinolate products after autolysis

In total, 16 compounds could be detected (seven ITCs and nine NITs) by solvent extraction of single seeds of wild-type and MINELESS plants (Fig. 9A, B). Out of the seven detected ITCs, 3-butenyl-ITC, 2-methylbutyl-ITC, 4-methylthiobutyl-ITC, 2-phenylethyl-ITC, and 5-vinyl-OZT differed significantly between wild-type and MINELESS single seeds (P <0.001) and 4-pentenyl-ITC with P <0.05 (Fig. 9A). Among the nine detected NITs, only 3-hydroxy-4-pentene-NIT, 3-hydroxy-4,5-epithiopentyl-NIT (isomer 1+2), benzenepropane-NIT, 1-methoxy-indoleaceto-NIT, and indoleaceto-NIT showed significant differences between wild-type and MINELESS single seeds (Fig. 9B). All of the detected NITs showed a lower abundance in MINELESS seeds, except 1-methoxy-indoleaceto-NIT, which was 34.1% higher in MINELESS seeds.

Fig. 9.

Glucosinolate–myrosinase hydrolysis products [isothiocyanates (ITCs), nitriles (NITs), and oxazolidinethione (OZT)] from single seeds of the wild type and MINELESS (n=10) detected by GC-MS analysis. (A) Solvent extraction of ITCs: 3B, 3-butenyl-ITC; 2MB, 2-methylbutyl-ITC; 4P, 4-pentenyl-ITC; 4MTB, 4-methylthiobutyl-ITC; 2PE, 2-phenylethyl-ITC; 5V OZT, 5-vinyl oxazolidinethione. (B) Solvent extraction of NITs: 2H3B, 3H4P, 3-hydroxy-4-pentene-NIT; 45ETP, 4,5-epithiopentyl-NIT; 5MTP, 5-methylthiopentane-NIT 3H45ETP, 3-hydroxy-4,5-epithiopenytl-NIT (isomer 1+2); BP, Benzenepropane-NIT; 6MTH, 6-methylthiohexane-NIT 1MIA, 1-methoxy-indoleaceto-NIT; IA, indoleaceto-NIT The wild type and MINELESS differed significantly for ITCs and NITs: 3B ITC, 2MB ITC, 4MTB ITC, 2PE ITC, 5V OZT, 3H45ETP-NIT1 and NIT2, BP-NIT (**P <0.001), 4P ITC, 3H4P-NIT, 1MIA-NIT and IA-NIT (*P <0.05) as determined by t-test. Error bars represent the SE.

Growth and development parameters of MINELESS plants

Wild-type and DH MINELESS plants were raised in a greenhouse under environmentally controlled conditions. The emergence of seedlings and the first appearance of true leaves were similar for wild-type and MINELESS plants. Fifteen days after sowing, five and four leaves were unfolded in wild-type and MINELESS plants, respectively. Opening of the first flower was also delayed by 3 d in MINELESS plants. The statistical analysis of growth parameters by ANOVA showed significant differences on the basis of genotype (wild type versus MINELESS) means (n=12) from the four developmental stages, as described below. The four developmental stages were: before initiation of flowering, after initiation of flowering, >50% flowering, and the end of flowering.

The comparative data on fresh and dry matter accumulation comprised plant parts such as the main stem, leaves, branches, and root at four stages of development (Fig. 10). MINELESS plants accumulated significantly less fresh and dry matter for the main stem compared with the wild type (Fig. 10D). Moisture content was found to be significantly different between the main stem of wild-type and MINELESS plants (P <0.05). The fresh and dry matter accumulation for branches was 23.0% and 28.0% higher for wild-type than for MINELESS plants (Fig. 10D). A significant interaction between genotype and developmental stage was observed for leaf fresh matter accumulation (P <0.05). Fresh and dry matter accumulation in root showed almost the same trend for both wild-type and MINELESS plants (Fig. 10D).

Fig. 10.

Growth and development parameters of wild-type and MINELESS plants on an average basis of four developmental stages (3 plants×4 developmental stages=12): stage I, before initiation of flowering; stage II, after initiation of flowering; stage III, >50% flowering; and stage IV, end of flowering. (A) Height of wild-type and MINELESS plants. (B) Leaf number. (C) Leaf area. (D) Fresh and dry matter accumulation in different plant parts: main stem, branches, leaves, and root, with their percentage moisture content. Wild-type and MINELESS plants differed significantly in terms of plant height, fresh and dry matter accumulation in the main stem (**P <0.001), main stem percentage moisture, and dry matter accumulation in branches (*P <0.05) as determined by ANOVA. Error bars represent the SE.

Comparative yield parameters of wild-type and MINELESS plants

Wild-type plants attained complete maturity in ∼105 d, while the MINELESS plants reached the same developmental stage 25 d later. For yield parameters, wild-type and MINELESS plants were compared in terms of plant height, number of primary branches per plant, number of siliques per plant, silique length and diameter, undeveloped/shrivelled seeds per silique, total number of seeds per silique, above-ground biomass, root weight, total plant biomass, seed yield per plant, 1000-seed weight, and HI per plant (Fig. 11).

Fig. 11.

Yield attributes of wild-type and MINELESS plants (n=15) at final harvest after plants attained final maturity. (A) Plant height. (B) Primary branches per plant. (C) Siliques per plant. (D) Silique length and diameter. (E) Seeds per silique. (F) Biomass of plant parts and the whole plant. (G) Seed yield, 1000-seed weight, and harvest index per plant. Wild-type and MINELESS plants differed significantly in terms of the number of siliques per plant, silique length, developed and total number of seeds per silique, biomass of the main stem and root, seed yield, 1000-seed weight, harvest index (**P <0.001), plant height, primary branches per plant, silique diameter, undeveloped siliques, total above-ground biomass, and total plant biomass (*P <0.05) as determined by ANOVA. Error bars represent the SE.

The average plant height for wild-type plants was 161 cm, while for MINELESS it was 152 cm (Fig. 11A). The average number of primary branches was higher for MINELESS (eight) relative to the wild type (seven) (Fig. 11B). The average number of total siliques per plant for the wild type was 331, while for MINELESS plants it was nearly doubled (611) (Fig. 11C). The enhanced number of siliques for MINELESS plants was contributed by the higher number of immature and mature siliques, as these caused 78.5% and 87.1% increases over the wild-type siliques. The average silique length and diameter of the wild-type plants was 6.7 cm and 0.67 cm (Fig. 11D). The average silique length and diameter of MINELESS plants was only 4.8 cm and 0.59 cm. While wild-type plants on average contained 28 developed seeds per silique, MINELESS plants had only seven developed seeds per silique (Fig. 11E). The most striking observation was the severe reduction (42.6%) in the biomass of the main stem in MINELESS plants as compared with the wild type. This also manifested for the overall reduction in above-ground biomass of MINELESS plants (Fig. 11F). The average root biomass was 57.1% less for MINELESS plants than for the wild type. The seed yield of MINELESS plants also showed a severe reduction (35.3%) as compared with the wild-type plants (Fig. 11G). Correspondingly, the HI was lower (28.0%) in MINELESS plants. In contrast, the 1000-seed weight of fully developed seeds presented a 34.6% increase for MINELESS plants compared with the wild type. The morphology of wild-type and MINELESS siliques and seeds as observed under a stereo-microscope is shown in Fig. 12D–F.

Fig. 12.

Comparison of wildtype (W) and MINELESS (M) plants in terms of morphological, growth, development, and yield parameters. (A) Two-week-old plants. (B) Plants at stage I: before initiation of flowering. (C) Plants at stage III: >50% flowering. (D) Siliques of the wild type and MINELESS after plants attained final maturity. Fully mature seeds of the wild type (E) and MINELESS (F) as observed under stereo-microscope. Bars=1 mm in E and F.

Pearson's correlation analysis showed many significant relationships between different yield attributes of the wild-type and MINELESS plants (Supplementary Table S1 available at JXB online). For instance, the total biomass was positively correlated with plant height (r=0.559). The total number of siliques was positively correlated with the number of mature siliques (r=0.949). Seed yield was positively correlated with the number of developed seeds (r=0.842). In contrast, the silique length and seed yield were negatively correlated with the number of mature siliques (r= –0.761 and r= –0.758, respectively). Moreover, the 1000-seed weight was negatively correlated with the number of developed seeds (r= –0.825) (Supplementary Table S1).

Discussion

To obtain homozygous seed of the transgenic MINELESS line, DHs were produced via microspore culture. The production, growth characteristics, and yield parameters have been investigated and compared with wild-type B. napus cv. Westar plants cultured under the same conditions.

Microspore culture and production of DH MINELESS seed

The successful production of DH seeds for the transgenic B. napus MINELESS plants showed that microspore embryogenesis can be carried out efficiently by using the protocol of Hansen (2003). While performing microspore culture, as detailed in the protocol, donor plants and microspores were exposed to various stresses (cold and heat) at the required steps. Exposing B. napus to temperature stress might produce unintended side effects (Young ). However, exposing plant material to a moderate heat shock of 32 °C is not likely to produce unintended side effects in later generations as temperatures of 32 °C are common on summer days where Brassica plants are cultivated.

It has been postulated that exposure to various stresses is necessary to induce microspores to undergo embryogenesis and for the conversion from the gametophytic to sporophytic pathway (Touraev ). The basis of this technology lies in the direct conversion of pollen microspores into plants, which is an important step in DH production (Zhou ). This also ensures minimal occurrence of cytogenetic abnormalities (Fletcher ). In the past, several methods have been used to convert haploids for the production of fertile, homozygous DHs (Zhao and Simmonds, 1995; Zhou ; Weber ). Several authors have stated that colchicine treatment of isolated microspores increases the doubling efficiency of regenerated plants in B. napus (Mollers ; Hansen and Andersen, 1996; Zhao ). This is due to the fact that colchicine is a microtubule-depolymerizing agent, which disrupts microtubule formation during cell division and hinders division-related activities (Kasha, 2005). To produce seed, DHs are necessary because the microspore-derived haploid plants are sterile/non-fertile. Colchicine can be applied at any stage during the process of microspore culture, from isolated microspores to the regenerated plants. In the present experiments, an overnight colchicine application (0.1%) on internodal regions of soil-transferred regenerated sterile MINELESS plants resulted in fully fertile flowers and successful production of DH plants with a good seed set.

Intended and unintended effects of transgenic ablation of myrosin cells

Most in vitro manipulations of plant cells and tissues can result in unintended effects such as somaclonal variation. MINELESS plants have been produced through two main steps of an in vitro manipulation: (i) transgenic modification and introduction of an active promoter/gene construct controlling expression of barnase and barstar; and (ii) production of DHs through microspore cultures.

Intended and targeted effects are the specific genetic ablation of myrosin cells and their cellular content. Previous studies have shown that myrosinase is localized in myrosin cells, and the present study confirmed, as expected, a strong reduction of myrosinase expression levels and activities. As myrosinases hydrolyse glucosinolates, a corresponding modification in the capacity to degrade glucosinolates would also be expected. It was also hypothesized that some other component of the myrosinase–glucosinolate system compartmentalized in myrosin cells could be affected. One such candidate would be the ESP isoform shown to be missing in the MINELESS seeds. Previous experiments with tissue culture and somaclonal variation (Bones, 1990) showed few or no observable changes as a function of regeneration processes from Brassicaceae protoplasts and explants. Myrosinase activity was remarkably unaffected, and regenerated plants had normal myrosinase activities. It is therefore believed that most of the observed variation is in fact due to the direct effect of reduced myrosinase, the capacity to hydrolyse glucosinolates, and potentially to the effect on modified production of bioactive products such as indolics from glucosinolates.

Myrosinase activity and expression of myrosinase and ESP isoforms

The investigations on myrosinase activity in seeds coupled with the expression of myrosinase isoforms in DH MINELESS seeds provide confirmatory information for previous analyses with the transgenic MINELESS seeds (Borgen ). The results emphasize that the MINELESS seeds have barely detectable myrosinase expression levels and activity. The myrosinase activity data from the 5-d COTY provide further evidence for the very low amounts of myrosinase in DH MINELESS seeds. Classical myrosinases are glycosylated dimeric proteins with subunit molecular weights in the range of 62–75 kDa in plants (Bones and Rossiter, 1996; Thangstad ). On the basis of cDNA sequencing, B. napus genes have been grouped into three subfamilies, MA, MB, and MC, encoding proteins with molecular weights of ∼75, 65, and 70 kDa, respectively (Rask ). The MA and MC classes of myrosinase genes are expressed particularly in developing seeds, and MB genes are expressed throughout development (Lenman ). MA myrosinases occur as free soluble dimers, while MB and MC families of myrosinases are found in complexes with several other proteins (Rask ; Eriksson , 2002). The constant and/or very low amount of soluble and insoluble myrosinase activity in MINELESS DH seeds, along with the absence of three isoforms, provides adequate evidence for the strongly reduced levels of myrosinase in MINELESS seeds. Faint bands of 78 kDa and 62 kDa were detected particularly in MINELESS seeds. The band of 78 kDa has been observed earlier in the purified seed myrosinase pool of Sinapis alba (Eriksson ). The same study also found peptides of low molecular mass seed myrosinases (59 kDa and 62 kDa) in S. alba seeds. In the wild type, the presence of three bands of 65, 70, and 75 kDa myrosinases is consistent with the occurrence of three different myrosinase isoforms in B. napus seeds (Lenman ; Falk ; Rask ; Borgen ). Additionally, the complete absence of three bands from two independent native protein extracts of DH MINELESS seeds on an IEF gel provides overwhelming evidence for the DH MINELESS seeds to be virtually myrosinase negative.

As expected, one of two ESP isoforms, the 35 kDa form, detected in wild-type seeds was absent in the MINELESS seeds. This coincides with previous information on ESP detection in transgenic MINELESS seeds (Borgen ). In contrast, in the wild type, the anti-ESP antibody revealed two strong bands of 39 kDa and 35 kDa. In B. napus, two isoforms of ESP with molecular masses of 39 kDa and 35 kDa have been shown (Bernardi ; Foo ; Borgen ). This again provides information about the possible co-localization of myrosinase and ESP in seed and cotyledon myrosin cells of brassicas. In A. thaliana, the ESP has been reported to be localized to the stem S-cells and the epidermis of all the above-ground plant parts except the anthers (Burow ). The myrosinases have also been shown to have a different localization in Arabidopsis relative to the brassicas (Andréasson ; Husebye ; Thangstad ).

Structural analysis of DH MINELESS shows empty myrosin cells

Confocal and light microscopic observations showed the distribution and presence of myrosin cells, exhibiting immunofluorescent and densely stained myrosin grains in wild-type seeds. This has been documented previously by several authors for different plant parts of Arabidopsis and Brassica by using various approaches (Thangstad ; Bones ; Höglund ; Andréasson ; Eriksson ; Husebye ; Borgen ). Histological and immunocytochemical confocal microscopic analysis of myrosin cells from DH MINELESS seeds show that myrosin cells in seeds are ablated, as observed previously in transgenic MINELESS seeds (Borgen ).

DH MINELESS seeds show high glucosinolate concentrations but low concentrations of glucosinolate hydrolysis products

From the glucosinolate analysis of wild-type and MINELESS single seeds, three major glucosinolates were found: 2-hydroxybut-3-enyl glucosinolate, 4-hydroxy-3-indol-3-ylmethyl glucosinolate, and indolyl-3-methyl glucosinolate. The detected glucosinolate profile from the wild-type and MINELESS single seeds resembled similar patterns obtained in previous studies conducted with oilseed rape (Fenwick and Heaney, 1983; Sang ; Clossais-Besnard and Larher, 1991; Daun and McGregor, 1991; Brown and Morra, 1995; Shahidi ; Matthaus and Luftmann, 2000). 2-Hydroxybut-3-enyl glucosinolate (progoitrin) and 4-hydroxyindol-3-ylmethyl glucosinolate (4-hydroxyglucobrassicin) were found to be the major glucosinolates in seeds of oilseed rape (Sang ; Clossais-Besnard and Larher, 1991). As observed in this study, indol-3-ylmethyl glucosinolate has been reported to be the minor constituent in seed meal of oilseed rape (Sang ). There was a significant variation in the quantity of individual glucosinolates and the total glucosinolate concentration between the wild-type and MINELESS seeds. All the detected glucosinolates were of a significantly higher concentration in MINELESS seeds. As expected, the glucosinolate hydrolysis products were substantially reduced in DH MINELESS seeds (Fig. 9). The hydrolysis of the glucosinolate progoitrin (2-hydroxybut-3-enyl glucosinolate) from B. napus seed meal releases the important goitrogenic compound 5-vinyl-OZT (Fenwick and Heaney, 1983; Mabon ), and in MINELESS seeds this goitrogenic substance was markedly reduced.

MINELESS plants show delayed development and modified yield attributes

The growth and development data show that plant growth was positive and steady throughout the growth cycle, for both the wild-type and MINELESS plants. However, the MINELESS plants exhibited a relatively slower growth and developmental pattern, which was evident from the growth and yield attributes analysed (Figs 10–12). The MINELESS plants were significantly shorter than the wild-type plants at all growth stages, calculated on an average of four developmental stages (Fig. 10). This difference in plant height of MINELESS plants during the developmental period was associated with the shorter MINELESS plants at final maturity (Fig. 11). The plant height at final maturity has been proposed to be related to the growth behaviour of a crop (Sana ). Bolting or opening of the first flower was delayed in MINELESS plants by 3 d as compared with the wild type. In oilseed rape the most important stage, which influences seed yield, is the onset of flowering. It has been suggested that the most limiting process starts shortly after the onset of flowering when decreases in the total leaf area are expedited due to shading (Gabrielle ). After an initial contribution by leaves during early plant growth, the main stem becomes the major organ for photosynthetic supply to the seeds (Major and Charnetski, 1976). Another parameter which is possibly linked to the slower growth rate of MINELESS plants is their longer duration of vegetative growth. The MINELESS plants possessed a greater number of leaves and leaf area, which was added mainly by the leaves from branches. Even when the wild-type plants had >50% flowers, MINELESS plants had fewer flowers and a higher number of leaves (i.e. a delay in the maturity of MINELESS plants was expected; Figs 10–12). In addition, the statistical analysis by ANOVA indicated that the interaction between developmental stage and genotype (wild type/MINELESS) is significant for the total leaf area and leaf matter accumulation. This means that the MINELESS plants behave slightly differently compared with the wild-type plants during all developmental stages. As well as being relatively slow in development, MINELESS plants disseminated a large reduction in the main stem fresh and dry matter, which was evident on an average basis of all the developmental stages (Fig. 10). In a study with the production of DH plants for cv. Westar, the alterations in morphology and architecture have been shown for DHs in comparison with its parental Westar (Malik ). The authors considered that Westar-derived DH lines might result from related genetic (or perhaps non-genetic) heritable changes, such as common rearrangements or phenotypic exposure of recessive gene alleles due to fixed homozygosity at all loci.

Furthermore, the yield data show that the growth and developmental differences between the wild-type and MINELESS plants for developmental stages resulted in the significant differences in plant yield attributes. It has been proposed that many different plant processes, which are operating within a plant during the growing season, contribute towards crop yield (Thurling, 1993). The silique size was smaller for MINELESS plants, with a reduced number of seeds in comparison with the wild type (Fig. 11). The reduced number of developed seeds per silique highlights that the ablation of myrosin cells occurs exactly during silique/seed development, as was designed (Borgen ). This has resulted in a relatively different canopy and yield structure of MINELESS plants. In oilseed brassicas, the numbers of secondary and tertiary branches, silique-bearing branches, siliques per plant, seeds per silique, and 1000-seed weight have been considered as the important morphological components of seed yield (Thurling, 1974; Katiyar ; Sana ). Individual plant seed yield is considered to be linked to the number of siliques per plant, which is conclusively determined through reduction in the number of branches, buds, flowers, and young pods, and supply of nutrients and water (Allen and Morgan, 1972; Tayo and Morgan, 1979; Rood and Major, 1984). Also, the seed number per silique is associated with the silique length (Diepenbrock, 2000). Evidence has been provided that long siliques generally contain more seeds, which is mainly due to the response of seed growth towards seed content and the genetic background of the tested material (Chay and Thurling, 1989; Léon and Becker, 1995).

Yield is classified as biological yield (total biomass) and economic yield (the economically useful part of the plant), and the relationship between the two, the HI, is the proportion of seed dry matter to above-ground biomass (i.e. economic yield=biological yield×HI; Fageria, 1992; Diepenbrock, 2000). It is evident from the data presented here that the seed yield and HI were significantly affected in MINELESS plants. In contrast, the comparatively higher 1000-seed weight attained by fully developed seeds of MINELESS may be due to the lower number of seeds per silique, which resulted in better utilization of resources and development of seeds, as previously proposed (Sana ).

Conclusions

Homozygous DH MINELESS seeds have been produced through microspore culture of transgenic MINELESS plants. MINELESS seeds have been characterized and compared with those of the wild-type cultivar Westar. Additionally, growth, development, and yield data have been obtained and compared with those of the wild type. Analyses of tissues, proteins, glucosinolates, and degradation profiles verify that DH seeds with strongly reduced and constant myrosinase levels, increased glucosinolate concentrations, and reduced endogenous potential for glucosinolate degradation have been obtained. The DH seeds produced will be used in plant–insect interaction studies as well as in animal (including farmed fish) feeding experiments. The seed material will also be useful for further characterization of the glucosinolate–myrosinase systems and its function in Brassica plants.

Supplementary data

Supplementary data are available at JXB online.

. Correlation coefficients between the yield attributes for Brassica napus cv. Westar (wild type) and MINELESS plants.