• Open Access

Combination of reversible male sterility and doubled haploid production by targeted inactivation of cytoplasmic glutamine synthetase in developing anthers and pollen


  • Alexandra Ribarits,

    1. Max F. Perutz Laboratories, University Departments at the Vienna Biocenter, Department of Plant Molecular Biology, Dr Bohrgasse 9/4, 1030 Vienna, Austria
    Search for more papers by this author
  • A. N. K. Mamun,

    1. Max F. Perutz Laboratories, University Departments at the Vienna Biocenter, Department of Plant Molecular Biology, Dr Bohrgasse 9/4, 1030 Vienna, Austria
    Search for more papers by this author
    • Present address: Plant Biotechnology Division, IFRB, AERE Bangladesh Atomic Energy Commission, GPO Box No. 3787, Dhaka-1000, Bangladesh

  • Shipeng Li,

    1. Plant Research International, PO Box 16, 6700 AA, Wageningen, the Netherlands
    Search for more papers by this author
  • Tatiana Resch,

    1. Max F. Perutz Laboratories, University Departments at the Vienna Biocenter, Department of Plant Molecular Biology, Dr Bohrgasse 9/4, 1030 Vienna, Austria
    Search for more papers by this author
  • Martijn Fiers,

    1. Plant Research International, PO Box 16, 6700 AA, Wageningen, the Netherlands
    Search for more papers by this author
  • Erwin Heberle-Bors,

    1. Max F. Perutz Laboratories, University Departments at the Vienna Biocenter, Department of Plant Molecular Biology, Dr Bohrgasse 9/4, 1030 Vienna, Austria
    Search for more papers by this author
  • Chun-Ming Liu,

    1. Plant Research International, PO Box 16, 6700 AA, Wageningen, the Netherlands
    Search for more papers by this author
  • Alisher Touraev

    Corresponding author
    1. Max F. Perutz Laboratories, University Departments at the Vienna Biocenter, Department of Plant Molecular Biology, Dr Bohrgasse 9/4, 1030 Vienna, Austria
    Search for more papers by this author

* Correspondence (fax +43 14277 9546; e-mail alisher.touraev@univie.ac.at)


Reversible male sterility and doubled haploid plant production are two valuable technologies in F1-hybrid breeding. F1-hybrids combine uniformity with high yield and improved agronomic traits, and provide self-acting intellectual property protection. We have developed an F1-hybrid seed technology based on the metabolic engineering of glutamine in developing tobacco anthers and pollen. Cytosolic glutamine synthetase (GS1) was inactivated in tobacco by introducing mutated tobacco GS genes fused to the tapetum-specific TA29 and microspore-specific NTM19 promoters. Pollen in primary transformants aborted close to the first pollen mitosis, resulting in male sterility. A non-segregating population of homozygous doubled haploid male-sterile plants was generated through microspore embryogenesis. Fertility restoration was achieved by spraying plants with glutamine, or by pollination with pollen matured in vitro in glutamine-containing medium. The combination of reversible male sterility with doubled haploid production results in an innovative environmentally friendly breeding technology. Tapetum-mediated sporophytic male sterility is of use in foliage crops, whereas microspore-specific gametophytic male sterility can be applied to any field crop. Both types of sterility preclude the release of transgenic pollen into the environment.


Today, many important seed (corn, rape), vegetable (tomato, pepper) and foliage (tobacco, forage grasses) crops are available as F1-hybrid cultivars. They have great commercial value as a result of heterosis (Birchler et al., 2003) and the protection of breeder's rights. Two inbred lines are crossed to produce F1-hybrid seeds, and a reliable system of pollination control is mandatory to enforce cross-pollination (Perez-Prat and van Lookeren Campagne, 2002). However, the lack of a universal technology for reversible male sterility is a major barrier to the exploitation of the advantages of F1-hybrid seed production in a wider range of plant species (Perez-Prat and van Lookeren Campagne, 2002). Male sterility avoids the labour of manual emasculation, and serves as a molecular strategy for transgene containment by preventing pollen release to the environment (Daniell, 2002; Perez-Prat and van Lookeren Campagne, 2002). In practice, tight sterility and efficient fertility restoration to maintain the male-sterile lines are musts for the successful incorporation of the sterility trait into F1-hybrid breeding (Budar and Pelletier, 2001; Perez-Prat and van Lookeren Campagne, 2002).

The most commonly used male sterility system is cytoplasmic male sterility (CMS) (Kaul, 1988). However, it is not available in any given crop, and requires tedious breeding procedures. Near-isogenic male-fertile plants carrying the appropriate restorer genes need to be available for the nuclear restoration of fertility, and, unless using naturally occurring CMS, potential parental inbred lines must be transformed into male-sterile lines, which is usually achieved through a series of backcrosses (Hanson and Bentolila, 2004). Genetically engineered reversible CMS has recently been presented as a promising alternative (Ruiz and Daniell, 2005), and potentially makes CMS available in any species in which plastid transformation is established. Furthermore, it avoids the segregation that typically occurs when a plant carrying a nuclear sterility gene is crossed with a male-fertile line, which hampers the generation of uniform populations of male-sterile plants (Chase, 2006).

Nuclear transformation, manipulating developmental or metabolic pathways in anthers or pollen, has resulted in male sterility, which has been reversed in several instances (Mariani et al., 1990, 1992; Goetz et al., 2001; Park et al., 2002; Huang et al., 2003; Yui et al., 2003). However, the controlled restoration of fertility remains difficult. Methods developed to restore fertility may be inefficient or impracticable on a large scale (Perez-Prat and van Lookeren Campagne, 2002).

Glutamine has a central position in plant amino acid metabolism (Cren and Hirel, 1999; Miflin and Habash, 2002). It plays an essential role during pollen development, as isolated and in vitro-cultured microspores are unable to develop into functional pollen grains in a medium lacking glutamine (Kyo and Harada, 1986). Glutamine synthetase (GS, E.C. catalyses the ATP-dependent conversion of glutamate to glutamine, utilizing ammonia as the substrate, and is the pivotal player in nitrogen metabolism (Cren and Hirel, 1999; Miflin and Habash, 2002). GS is an octameric enzyme, and is found in two iso-enzymatic forms, cytoplasmic GS1 and chloroplastic GS2, which are tightly regulated during male reproductive development (Becker et al., 1992; Dubois et al., 1996; Cren and Hirel, 1999). Two recent publications have shown the importance of cytosolic GS in the reproductive development of rice (Tabuchi et al., 2005) and maize (Martin et al., 2006), and chemical inhibition of GS in developing rice anthers causes male sterility and depletes plants of crucial amino acids, including glutamine and glutamate (Kimura et al., 1994; Cren and Hirel, 1999). Exploiting this knowledge, we used a dominant-negative mutant (DNM) approach to inhibit GS activity, specifically in the tapetum and microspores of tobacco. Foliage sprays and in vitro maturation restored the fertility of male-sterile plants, and completely homozygous male-sterile lines were rapidly produced by microspore embryogenesis. Thus, two different methods of fertility restoration and one new method of maintenance and propagation of male-sterile lines were made available.


Dominant-negative inhibition of GS in the tapetum and microspores of tobacco

Genome survey and reverse transcriptase-polymerase chain reaction (RT-PCR) analysis have shown that there are at least five copies of GS1 and one copy of GS2 in the tobacco genome (data not shown). GS1 and GS2 cDNAs were isolated from Nicotiana tabacum cv. Petit Havana SR1 (Maliga et al., 1975) and confirmed by sequencing (data not shown). A DNM GS2 under the control of the tapetum-specific TA29 promoter (Koltunow et al., 1990) (pTA29-ΔGS2) was created by deleting the C-terminal 45 amino acids, which are critical for GS activity but not for subunit binding (Eisenberg et al., 2000). Pollen fertility was not affected when this construct was transformed into Arabidopsis (data not shown). Therefore, in addition to the activity domain, the N-terminal chloroplast-targeting signal was deleted in the pTA29-ΔGS2-L construct (Figure 1). The severe decrease in pollen fertility and the very low seed set observed in Arabidopsis transformants (data not shown) showed that the deletion of the chloroplast-targeting signal was essential to inactivate the enzyme. The DNM of GS1 was achieved by introducing point mutations in two critical sites identified in plants and Escherichia coli (Clemente and Marquez, 1999; Eisenberg et al., 2000). To obtain maximum detrimental effects on GS activity and minimum interruption of subunit interaction, asparagine at position 56 was replaced by alanine, and arginine at position 291 by leucine, which was expected to result in severe structural changes. The mutated GS1 gene was fused to the tapetum-specific TA29 (Koltunow et al., 1990) or the microspore-specific NTM19 (Oldenhof et al., 1996; Custers et al., 1997) promoter (Figure 1). The constructs pTA29-ΔGS2-L, pTA29-GS1A56L291 and pNTM19-GS1A56L291 were transformed into tobacco via Agrobacterium-mediated leaf disc transformation (Horsch et al., 1985), and 19–50 independent, kanamycin-resistant and PCR-positive transgenic plants were obtained per construct. Stable integration of the T-DNAs was confirmed by Southern blotting (Figure 2), and supported by segregation analysis of seeds produced by cross-pollination on a kanamycin-containing medium to determine the transgene copy number (Table 1). Southern blot hybridization revealed two common bands in addition to the variable ones (Figure 2), most probably representing the endogenous GS1 and GS2 genes. A similar pattern has been reported previously (Temple et al., 1993). A 1 : 1 segregation ratio, indicative of a single insertion, was found in seven lines, whereas another seven lines carried two or more copies of the transgene (Table 1).

Figure 1.

Schematic representation of the pBINPLUS vectors used for transformation. (a) pTA29-ΔGS2-L: 45 amino acids from the N-terminal sequence and 59 amino acids from the C-terminal sequence were deleted. (b) pTA29-GS1A56L291: in GS1, aspartate at position 56 (As) was replaced by alanine (Al), and arginine at position 291 (Ar) was replaced by leucine (L), and fused to the tapetum-specific promoter TA29. (c) pNTM19-GS1A56L291: mutated GS1 was fused to the microspore-specific promoter NTM19. LB, left border; NPTII, neomycin phosphotransferase type II (nptII) conferring kanamycin resistance; NTM19, microspore-specific promoter; P, nopaline synthetase (nos) promoter; RB, right border; T, nos terminator; TA29, tapetum-specific promoter.

Figure 2.

Southern blot analysis of wild-type and pTA29-ΔGS2-L transformed plants. Genomic DNA (10 µg) was isolated from leaves of wild-type (wt) and four independent transgenic lines (22, 25, 29, 32).

Table 1.  Pollen viability, in vitro pollen germination and seed set of male-sterile pTA29-ΔGS2-L (pTA29-GS2), pTA29-GS1A56L291 (pTA29-GS1) and pNTM19-GS1A56L291 (pNTM19-GS1) transformants after backcrosses with wild-type or selfing, and segregation on kanamycin-containing medium (100 mg/L). Seed weight data are means from at least six seed pods per line
ConstructLineViability mature pollen T0 (%)In vitro pollen germination T0 (%)Average seed weight/pod (g) selfingAverage seed weight/pod (g) crossSegregation ratio KanR : KanS
  1. KanR : KanS, kanamycin resistant : kanamycin susceptible; n.a., not applicable.

pTA29-GS2L8 14.0 ± 0.9000.0610 : 1
L12 16.2 ± 1.1000.11 3 : 1
L22  5.4 ± 0.3000.07 3 : 1
L25 14.6 ± 0.7000.06 3 : 1
L32  7.8 ± 0.6000.07 1 : 1
pTA29-GS1L110000.13 1 : 1
L140000.09 3 : 1
L410000.08 3 : 1
L450000.08 3 : 1
pNTM19-GS1L3 52.3 ± 4.446.49 ± 1 : 1
L8 52.6 ± 0.646.02 ± 1 : 1
L39 50.7 ± 0.746.47 ± 1 : 1
L61 49.6 ± 2.543.67 ± 1 : 1
L64 50.2 ± 1.546.37 ± 1 : 1
Wild-type95.45 ± 1.395.32 ±

The developmental defects are confined to male reproductive organs

Vegetative development of primary T0 transformants was similar to that in wild-type plants. Five lines carrying pTA29-ΔGS2-L and four lines harbouring pTA29-GS1A56L291 (Table 1) were completely male sterile. Eight more lines displayed highly reduced fertility (data not shown). After dehiscence, almost all pollen grains (91–99.5%) were found to be aborted, and no single pollen was able to germinate in vitro (Table 1). Ten of 19 pNTM19-GS1A56L291 lines exhibited 50% pollen viability, as expected for single insertions of the transgene, and five representative lines are displayed in Table 1. Light and fluorescence microscopy demonstrated that pollen started to die close to, or immediately after, the first pollen mitosis stage (Figure 3). This result was independent of the construct and unexpected, as the activity of both promoters sets in well before this stage (Koltunow et al., 1990; Oldenhof et al., 1996).

Figure 3.

Cytology of microspores and pollen grains from transgenic pTA29-ΔGS2-L (pTA29-GS2; L12), pNTM19-GS1A56L291 (pNTM19-GS1; L11) and wild-type (wt) tobacco plants. Developing pollen grains were examined under the light microscope or stained with 4′,6-diamidino-2-phenylindole (DAPI). At the tetrad stage (a), no differences were visible between transgenic and wt cells; at the microspore stage (b), the number of viable microspores started to decrease in the transgenic plants compared with the wt plants; at the mature pollen stage (c), most of the pollen grains in the transformed plants were aborted. Bar, 20 µm.

Tapetum-specific expression of the mutated GS1 or GS2 resulted in anthers which changed colour from green to olive at about the first pollen mitosis stage, dried up prior to dehiscence and released very small amounts of mostly aborted pollen grains (Figure 4). As expected, similar to rice panicles (Kimura et al., 1994), slightly increased ammonia contents were found in anthers of pTA29-ΔGS2-L plants compared with the wild-type (data not shown), which might explain the browning of anthers (Figure 4). The tapetum began to degenerate at the early microspore stage, accompanied by precocious deposition of wall thickenings in the endothecium (Figure 4). Wild-type and pNTM19-GS1A56L291 anthers looked similar until the dehiscence stage. However, the transgenic anthers clearly released less pollen (Figure 4).

Figure 4.

Phenotypic alterations in transgenic vs. wild-type tobacco flowers and anthers. (a) Anthers in male-sterile plants turned olive to brown after the first pollen mitosis (pTA29-GS2, L22). pNTM19-GS1 (L83), from left to right: fertile flower of a cross between a homozygous doubled haploid (DH) and a wild-type plant (DH × wt); restored fertile flower of a DH line after spraying with glutamine (DHr); male-sterile flower of a homozygous DH plant (DHst); fertile flower of a wild-type plant (wt). (b) Anther development in wild-type (1–3) and three independent transgenic pTA29-ΔGS2-L lines (L8, L22, L32). Transverse sections (5 µm) were stained with toluidine blue and examined by bright-field microscopy. The tapetum (arrowhead) appeared as a smooth layer in the anther locule at the microspore stage (1), started to degenerate from the late unicellular microspore stage (2), and was fully degraded at the mature pollen stage (3), accompanied by wall thickenings in the endothecium. In transgenic anthers at the late unicellular microspore stage (L8, L22, L32), degeneration of the tapetum (arrowhead) was clearly advanced and wall thickenings (double arrowhead) were deposited precociously. Bar, 50 µm.

Primary pNTM19-GS1A56L291 plants set as many seeds as wild-type plants after self- and cross-pollination, showing that female fertility was not impaired (Table 1). A variable decrease in average seed set as a result of pod-to-pod variation was observed in pTA29-GS2 and pTA29-GS1 plants (Table 1).

Production of homozygous male-sterile plants via microspore embryogenesis

For complete sterility and to maintain the male-sterile lines as a non-segregating population, the male sterility trait must be homozygous. It is known that, on stress treatment, isolated microspores become embryogenic and form haploid embryos when cultured in appropriate medium (Touraev et al., 2001). Doubled haploid (DH) production is a rapid method to achieve complete homozygosity and is established in many important crops, including dicots (tobacco, pepper, rapeseed) and monocots (barley, wheat, corn) (Maluszinski et al., 2003). Exploiting the finding that, in all lines, microspores were viable in sufficient numbers, DH lines homozygous for the transgenic male sterility trait were regenerated via in vitro microspore embryogenesis (Touraev et al., 2001). Microspores were isolated from anthers of heterozygous plants, re-programmed towards embryogenesis by a starvation treatment and cultured in a medium adapted for embryo formation (Touraev and Heberle-Bors, 1999). This procedure is very efficient in wild-type tobacco microspores, as typically 70% of isolated and cultured microspores start to divide and eventually form embryos. Microspore viability after isolation from male-sterile plants was in the region of 30%, 10% of which developed into embryos. Torpedo-shaped embryos (Figure 5) formed after 1 month, were treated with colchicine for chromosome doubling and were selected on a kanamycin-containing medium. Completely male-sterile and homozygous T1 DH progeny were obtained from all lines (data not shown). Homozygosity is particularly crucial to achieve 100% male sterility in pNTM19-GS1A56L291 lines (Table 1). In a simulation of F1-hybrid seed production (see also Figure 7), homozygous pNTM19-GS1A56L291 T1 DH plants were used as female parents, and pollinated with wild-type pollen (male donor line). The T2-hybrid progeny were kanamycin-resistant, produced 50% viable, non-transgenic pollen, and self-pollination resulted in almost full seed set (Table 1).

Figure 5.

Embryogenesis in microspores from pTA29-GS1A56L291 (pTA29-GS1, L11) and wild-type (wt) plants cultured in vitro. (a) Freshly isolated tobacco microspores. (b) Globular embryos. (c) Torpedo- and cotyledon-stage embryos after 7 weeks of culture. Bars, 20 µm (a), 100 µm (b and c).

Figure 7.

Chart showing the generation and application of sporophytic (TA29) and gametophytic (NTM19) reversible male sterility induced by inactivation of glutamine synthetase. Transformation with pTA29-ΔGS2-L (pTA29-GS2) or pTA29-GS1A56L291 (pTA29-GS1) yields heterozygous, completely male-sterile plants, whereas heterozygous pNTM19-GS1A56L291 (pNTM19-GS1) transformants release 50% non-transgenic pollen and therefore are fertile. Homozygous, 100% male-sterile plants are produced as doubled haploids via microspore embryogenesis. These plants are used as female parents in a cross with a male-fertile inbred line. To maintain the homozygous male-sterile plants and to increase their number, three methods are available: (1) self-pollination after spraying plants with glutamine; (2) pollination with in vitro-matured pollen; (3) microspore embryogenesis.

Pollen fertility in the male-sterile lines can be restored by spraying plants with glutamine or by in vitro pollen maturation

Male sterility caused by chemical inhibition of GS in developing rice anthers can be overcome by the exogenous application of glutamine (Kimura et al., 1994). We likewise succeeded in restoring pollen fertility in DNM GS male-sterile plants, in both the heterozygous (Table 2) and homozygous (Table 3) state. Six male-sterile pTA29-GS1 (lines L11, L14, L41) and pTA29-GS2 (lines L22, L25, L32) plants per line, grown from seeds after cross-pollination, and male-sterile pNTM19 plants (lines L8-DH, L39-DH, L61-DH, L64-DH), propagated by microspore embryogenesis, were sprayed with an aqueous solution of glutamine. Sprays were applied on to leaves and developing inflorescences twice a day for 7–10 days from the meiosis stage. Self-pollination resulted in seed pods which contained sufficient amounts of viable seeds for further propagation (Tables 2 and 3).

Table 2.  Restoration of fertility by spraying leaves and inflorescences of male-sterile heterozygous T1 plants pTA29-GS1A56L291 (pTA29-GS1) and pTA29-ΔGS2-L (pTA-GS2), obtained after pollination of male-sterile T0 plants with wild-type pollen, with an aqueous solution of 0.05 mm glutamine. In vitro pollen germination frequencies are indicated by means ± standard deviation. Seed pods were obtained after manual self-pollination with restored pollen
ConstructLineIn vitro pollen germination after restoration (%)Number of seed pods after self-pollinationAverage seeds per seed pod
  1. n.a., not applicable.

pTA29-GS1L1112.11 ± 2.412   181
L14  7.3 ± 0.5 5   123
L41  0.3 ± 0.1 0     0
pTA29-GS2L22  8.4 ± 1.6 7   120
L25  7.8 ± 1.4 5   102
L32 10.1 ± 2.410   151
Wild-type 94.5 ± 2.0n.a.> 1000
Table 3.  Viability of developing pollen, germination frequencies and seed set in doubled haploid (T1 DH) homozygous male-sterile pNTM19-GS1A56L291 T1 lines, before and after spraying with 0.05 mm glutamine. Pollen viability and germination were determined four times, and seed set was evaluated in 10–12 seed pods
LineViability of microspores (%)Viability of bicellular pollen (%)In vitro pollen germination (%)In vitro pollen germination after restoration (%)Number of seeds/pod after selfingSeed weight/ pod after backcrossing (g)
L8-DH24.2 ± 1.3 4.6 ± 0.4011.3 ± 0.6   1880.11
L39-DH23.9 ± 1.6 4.9 ± 0.2010.8 ± 1.4   1290.09
L61-DH24.8 ± 2.5 2.6 ± 0.3011.8 ± 1.3   1890.10
L64-DH26.2 ± 1.7 3.5 ± 0.2011.1 ± 1.2   2010.11
Wild-type96.2 ± 2.195.4 ± 1.395.3 ± 0.194.5 ± 2.0> 12000.15

Tobacco microspores can be matured in vitro in medium supplemented with the necessary nutrient sources, including glutamine, and mature pollen may be used to pollinate flowers in vivo (Touraev and Heberle-Bors, 1999). Viable microspores were isolated from male-sterile pTA29-GS1 (lines L11, L14, L41) and pTA29-GS2 (lines L8, L12, L25) plants and, within 6 days, were cultured in vitro to maturity in the presence of glutamine (Table 4; Figure 6). Both wild-type and transgenic in vitro-matured pollen grains produced pollen tubes (Figure 6) when transferred to appropriate germination medium (Touraev and Heberle-Bors, 1999). Pollination of receptive wild-type stigmas with an enriched suspension of viable, in vitro-matured pollen resulted in fruit set (Table 4). The segregation ratios of the seeds obtained corresponded to those after conventional backcrosses (data not shown), proving that the pollen of male-sterile plants can indeed be rescued. Self-pollination, by pipetting in vitro-matured pollen on to stigmas of male-sterile plants, also resulted in seed set (data not shown). Hence, the application of glutamine on to male-sterile plants, or in the culture medium to isolated microspores, is sufficient to restore pollen viability, and indicates that a lack of glutamine leads to pollen abortion in the DNM GS transgenic plants.

Table 4.  Restoration of fertility by in vitro maturation. Microspores were isolated from pTA29-GS1A56L291 (pTA29-GS1) and pTA-ΔGS2-L (pTA29-GS2) lines, matured in vitro and used for in situ pollination of wild-type plants. Viability, maturation and germination frequencies of pollen are means from three experiments and are given ± standard deviation
ConstructLineMicrospore viability after isolation (%)Mature pollen after in vitro maturation (%)Germination of in vitro-matured pollen (%)Number of pods formed after in situ pollinationAverage seed weight/pod (mg)
pTA29-GS1L1130.7 ± 1.421.1 ± 1.010.2 ± 0.53236
L1424.5 ± 5.615.9 ± 0.4 9.8 ± 0.83120
L4130.9 ± 2.721.7 ± 1.7 0.3 ± 0.10  0
pTA29-GS2L837.1 ± 2.822.9 ± 3.312.4 ± 0.44251
L1235.4 ± 2.321.2 ± 0.510.1 ± 0.84223
L2529.8 ± 3.216.8 ± 1.910.8 ± 0.73180
Wild-type96.2 ± 2.190.8 ± 3.057.6 ± 4.53763
Figure 6.

In vitro pollen maturation of isolated microspores from pTA29-GS1A56L291 (pTA29-GS1, L11), pTA29-ΔGS2-L (pTA29-GS2, L12) and wild-type (wt) plants. After isolation, microspores (a) were cultured in medium T1, supplemented with glutamine. Starch accumulation was observed after 6 days of incubation (b). (c) In vitro-matured pollen grains germinated when incubated in germination medium GK. Bar, 20 µm.


Plant reproduction is a complex process requiring a precisely timed series of developmental steps and coordinated action of metabolic pathways (McCormick, 1993). Consequently, male sterility can be engineered through the disruption of the developmental or metabolic pathways in anthers or pollen, or by the delivery of cytotoxic compounds to the tapetum or pollen grains (Williams, 1995; Perez-Prat and van Lookeren Campagne, 2002). We have produced 100% male-sterile tobacco plants by metabolic engineering of glutamine. Inhibition of the rate-limiting enzyme in glutamine biosynthesis, GS, was triggered by mutated GSs (GS1 or GS2) under the control of the tapetum-specific TA29 or microspore-specific NTM19 promoter. The GS2 N-terminal sequence, targeting the protein to the chloroplasts, and its C-terminal functional domain were deleted to achieve a DNM effect. Removing only the functional domain resulted in GS2-specific inhibition of enzyme activity, which did not lead to male sterility. Deleting both domains, however, suppressed pollen maturation, probably caused by the interaction of the DNM GS2 with the endogenous GS1. Hence, it can be concluded that GS1 is critical for tapetum and pollen development, whereas GS2 is less essential. In all tobacco transformants, vegetative development remained unaltered compared with wild-type plants. However, nine lines transformed with tapetum-specific DNM GS were completely male sterile. Anther sections showed early degeneration of the tapetum and precocious deposition of wall thickenings in sterile anthers, as metabolic alterations, e.g. ammonium accumulation and starvation from glutamine, probably led to premature senescence of sterile anthers. Likewise, senescing leaves were characterized by decreased activity of GS2 and increased ammonium contents (Chen et al., 1997; Masclaux et al., 2000). The fertility of male-sterile plants was restored by supplying glutamine either in the in vitro maturation medium or by spraying plants, indicating that the lack of glutamine had caused male sterility. These results are in agreement with earlier findings that pollen maturation requires glutamine (Kyo and Harada, 1986) and that the inhibition of GS in anthers leads to male sterility (Kimura et al., 1994). Pollen abortion commenced close to, or immediately after, the first pollen mitosis stage, independent of the construct. This allowed us to produce DH plants via microspore embryogenesis. These homozygous plants were male sterile, showing that the trait was transmitted to the progeny, and could be used as inbred parental lines. This step was important if the male sterility trait was to be induced by microspore-specific DNM GS. In these plants, pollen viability was reduced by 50%, indicating the insertion of an active foreign DNA in one locus. Any fertile pollen produced is the wild-type. Consequently, the risk of transgene release by uncontrolled cross-pollination is virtually ruled out. This applies not only to the male-sterile lines, but also to the F1-hybrids generated.

Although different male sterility systems have aided in the understanding of male reproductive development, they often lack efficient methods to multiply the male-sterile plants (Perez-Prat and van Lookeren Campagne, 2002). This shortcoming limits their use in commercial hybrid seed production. CMS, often tried in F1-hybrid breeding, requires appropriate restorer genes to restore male fertility, which are delivered by crossing the male-sterile with a male-fertile line (Hanson and Bentolila, 2004). CMS and appropriate restorer genes are not available for most agronomically important crops (Ruiz and Daniell, 2005; Chase, 2006), and the conversion of two potential parental inbred lines into a male-sterile and a near-isogenic restoration line by hybridization is tedious. An interesting alternative to naturally occurring CMS is engineered CMS via the chloroplast genome, taking advantage of the importance of fatty acid biosynthesis for pollen development (Ruiz and Daniell, 2005). Continuous illumination is sufficient to restore male fertility. The application of the system in agronomically important species depends on the feasibility to engineer plastid genomes, and recent advances in plastid transformation have opened up new perspectives (Daniell et al., 2005). Nuclear transformation is possible in a number of plants, including important crops. Thus tobacco, Arabidopsis and maize have been utilized to study the roles of cytokinins and gibberellins in pollen development (Huang et al., 2003). Impaired hormone degradation and signal transduction resulted in male sterility, and spraying the plant hormone kinetin restored male fertility in some male-sterile lines (Huang et al., 2003). This approach is potentially widely applicable, but the exogenous application of plant hormones may affect plant growth and development. Male sterility caused by glutamine deficiency employs the regular metabolite glutamine to restore fertility. Glutamine is likely to be innocuous in an agricultural setting. Toxic or potentially harmful substances to induce male sterility or restore fertility should be avoided.

Metabolic engineering is an effective approach to induce male sterility in diverse plant species. Flavonoid deficiency in pollen leads to male-sterile plants (van der Meer et al., 1992), and fertility can be restored by the application of flavonols to the stigmas or pollen before fertilization (Mo et al., 1992). In some species, the female tract produces flavonols to stimulate pollen tube growth (van Eldik et al., 1997), which may be disadvantageous for the tightness of sterility. Disruption of an anther-specific iso-enzyme of extracellular invertase resulted in the late abortion of pollen, which allowed an attempt to be made at in vitro maturation in order to rescue pollen from male-sterile plants, without success (Goetz et al., 2001). In contrast, in many male sterility technologies, pollen abortion sets in early (for example, Mariani et al., 1990; Huang et al., 2003; Yui et al., 2003). The early onset of pollen sterility is considered to be desirable as it prevents unwanted self-pollination. Metabolic engineering of glutamine targets pollen development in the stage just after the formation of microspores. This simultaneously ensures both tight sterility and the formation of viable microspores. Microspores cannot only be isolated and cultured for microspore embryogenesis, but also for in vitro maturation. In the breeding process, these two methods can be used to produce homozygous male-sterile plants, to restore their fertility and to propagate them.

The major disadvantage of nuclear transformation is that, typically, segregation occurs, and male-fertile plants must be eliminated (Mariani et al., 1990, 1992). Usually, the sterility trait prevents the production of homozygous plants, which are required for maintenance breeding and F1-hybrid seed production. The induction of male sterility by glutamine deficiency allows efficient fertility restoration by foliage sprays of glutamine or by in vitro maturation. Thus, plants can be maintained by selfing, and tedious backcrossing is avoided. As a novelty, we showed the use of microspore embryogenesis to rapidly generate completely homozygous male-sterile plants, which offers an efficient way to propagate the male-sterile lines.

The two different modes of genetic transmission of the sterility trait facilitate adaptations of the technology for F1-hybrid breeding in most species. Gametophytic male sterility using the constructs driven by the NTM19 promoter can be applied to seed crops, as 50% of the pollen in the resulting hybrids will be wild-type, fertile and sufficient for full seed set (Figure 7). Sporophytic male sterility using the TA29-driven constructs is of use for non-seed crops, as no fertile pollen is produced. For both types of male sterility, lines homozygous for the male sterility trait can be produced, and used as female parents in crosses with another male-fertile inbred line for F1-hybrid breeding (Figure 7). The current technology should be directly applicable in breeding nurseries, and of particular interest to producers of expensive vegetable and ornamental seeds, frequently dealing with tedious manual manipulations. Its implementation in diverse species would define the key stages in the production process, and thus facilitate adaptation for large-scale seed production under field conditions. Glutamine applications may be automated under commercial breeding conditions, and the optimization of their timing and formulation should significantly increase restoration efficiency. Having provided proof of concept in tobacco, we feel confident that the presented environmentally friendly F1-hybrid seed production technology can be adopted in many commercially important crops.

Experimental procedures

Plant material

Nicotiana tabacum L. cv. Petit Havana SR1 (Maliga et al., 1975) plants were grown in the glasshouse at 25 °C with 16 h of daylight or in vitro on Murashige and Skoog (MS) medium (Murashige and Skoog, 1962) at 25 °C for 4–6 weeks after surface sterilization of seeds with 70% ethanol and 10% sodium hypochlorite (Fluka, Buchs, Switzerland; v/v; 1% active chlorine).

Construction of plasmids with DNM in GS genes

Tobacco GS genes (GS1 and GS2) were isolated by RT-PCR from a tobacco cDNA library (N. tabacum cv. Petit Havana SR1), and confirmed by sequencing analysis. The construct pTA29-ΔGS2-L was created by deletions of the C-terminal activity domain (amino acids 944–989) and the N-terminal chloroplast-targeting signal (amino acids 1–59) under the control of the tapetum-specific TA29 promoter (Koltunow et al., 1990). The DNM of GS1 was achieved by introducing point mutations in two critical sites (Clemente and Marquez, 1999; Eisenberg et al., 2000). Asparagine at position 56 was replaced by alanine, and arginine at position 291 by leucine. The mutated GS1 gene was fused to the tapetum-specific TA29 (Koltunow et al., 1990) or the microspore-specific NTM19 (Oldenhof et al., 1996; Custers et al., 1997) promoter, and cloned into the binary vector pBINPLUS. All constructs (Figure 1) were transformed into Agrobacterium tumefaciens (LBA4404) by electroporation.


Agrobacterium-mediated leaf disc transformation was performed essentially as described by Horsch et al. (1985). Transgenic shoots were selected on MS medium containing 50 mg/L kanamycin and 500 mg/L cefotaxime sodium (Duchefa, Haarlem, the Netherlands). Rooted plantlets were transferred to soil for further growth under the same conditions as described for wild-type plants in the glasshouse until flowering.

Isolation of genomic DNA, PCR amplification and Southern hybridization

Genomic DNA for PCR and Southern blotting was isolated from leaf tissues of kanamycin-resistant plants using the DNeasy Plant Mini Kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions. The presence of the introduced constructs was assayed by PCR analyses using primers for the 3′-region of the TA29 promoter (5′-GCGACATAGGCTCGAAGTATGCAC-3′) or the NTM19 promoter (5′-GCTTCTAATTGATCAGACGAGGAC-3′), and specific for the introduced genes (5′-GACCTGTGCTAGATCCATCATAGTT-3′ for GS1 and 5′-CTCAGAAGCATGCTTGACTGGCTTTG-3′ for GS2) under standard PCR conditions. Genomic DNA (10 µg) was digested for 12–16 h with NcoI and VspI (MBI Fermentas, St. Leon-Rot, Germany) using 2 U/µg DNA. Digested DNA was fractionated on 0.8% agarose gels, and transferred on to GeneScreenPlus membranes (Perkin-Elmer, Wellesley, MA, USA). Blots were probed with a 419-bp PCR product, generated by amplification with the primer pair 5′-GGTACTAACGGAGAGGTTATGCCAGG-3′ and 5′-CCTTTGCCTTGCTTCTCAGTGTCAGC-3′, and labelled with 32P using a RadPrime DNA Labelling System (Invitrogen, Carlsbad, CA, USA). Hybridization was performed at 65 °C in 0.5 m Na2PO4 (pH 7.2), 7% sodium dodecylsulphate (SDS).

Protein isolation and immunoblotting

Frozen fresh young leaves (100 mg) and anthers collected from buds of 9–11 mm were ground in ice-cold protein extraction buffer [500 mm N-2-hydroxyethylpiperazine-N′-2-ethanesulphonic acid (HEPES), 750 mm KCl, 100 mm MgCl2, 100 mm phenylmethylsulphonyl fluoride]. The homogenate was centrifuged at 20 000 g for 40 min at 4 °C. Twenty micrograms of protein per sample were run on 10% SDS–polyacrylamide gels, and blotted on to polyvinylidene difluoride membranes (Immobilon-P; Millipore, Billerica, MA, USA). Membranes were blocked overnight at 4 °C in PBS-T (0.14 m NaCl, 2.7 mm KCl, 1.5 mm KH2PO4, 4 mm Na2HPO4, 0.05% Tween-20) containing 5% non-fat dried milk. Rabbit anti-GS (1 : 5000) was used as the primary antibody, and alkaline phosphatase-conjugated goat anti-rabbit immunoglobulin G (IgG) (Sigma-Aldrich, Taufkirchen, Germany) as the secondary antibody. The reaction was visualized by fluorography using CDP-Star (Amersham Biosciences, GE Healthcare, Munich, Germany) as a substrate.

Selection of male-sterile plants and segregation analysis of backcross-derived seeds

The primary T0 transgenic lines were self-pollinated or backcrossed with pollen from wild-type tobacco plants. Seeds were germinated on MS medium containing 100 mg/L kanamycin to select for transformed T1 progeny, and copy numbers of DNM GS insertions were estimated by segregation rates.

Cytological analysis of male gametophyte development

The viability of pollen and the stage of pollen development were determined under a fluorescent microscope (Leitz-DIAPLAN; Leica, Wetzlar, Germany) using the fluorescein isothiocyanate (FITC) or ultraviolet (UV) filter channel after staining in fluorescein diacetate solution (1 µg/µL; Sigma-Aldrich) or 4′,6-diamidino-2-phenylindole (Partec, Münster, Germany), as described by Touraev and Heberle-Bors (1999). Images were taken using Kodak Gold (Kodak, Stuttgart, Germany) ISO400 or ISO800 films.

Histological analysis

Anthers collected at different developmental stages were fixed overnight in FAA (50% ethanol, 5.0% glacial acetic acid, 3.7% formaldehyde; v/v), and kept in 70% ethanol until further processing. Fixed samples were embedded using the Kulzer Technovit 7100 Kit (Kulzer, Wehrheim, Germany). Sections of 5 µm were stained in 1% toluidine blue (w/v) for 10 min, and photographed under an Olympus BX50 microscope (Olympus, Tokyo, Japan) with a Nikon D100 digital camera (Nikon, Tokyo, Japan).

In vitro pollen germination

In vivo-matured pollen grains were collected from open tobacco flowers, germinated in medium GK for 3–6 h at 25 °C in the dark (Touraev and Heberle-Bors, 1999), and pollen tube growth and germination frequencies were observed under a light microscope.

Ammonia extraction and quantification

Protein was isolated from anthers 9–11 mm in size, as described for immunoblotting, and the phenol hypochlorite assay (Berthelot) reaction was used to determine ammonia contents, as described by Witte and Medina-Escobar (2001).

Production of homozygous T1 male-sterile plants via microspore embryogenesis

Homozygous T1 DH male-sterile lines were produced essentially as described by Touraev and Heberle-Bors (1999). Microspores were isolated from anthers aseptically excised from 11-mm buds, cultured in starvation medium B at 33 °C for 6 days and transferred to embryogenesis medium AT3, resulting in the formation of embryos. After 6–8 weeks, chromosome doubling was performed by incubating embryos in AT3 medium supplemented with 0.1% (w/v) colchicine for 6–8 h at 25 °C in the dark. Diploidized embryos were transferred on to kanamycin-containing MS medium (100 mg/L) for the selection of transgenic plants. Flow cytometry was performed as described by Indrianto et al. (1999).

Fertility restoration of male-sterile lines by spraying

Glutamine (0.01–0.05 mm) was dissolved in water, filter-sterilized and sprayed twice a day for 7–10 days on to leaves and inflorescences of young glasshouse-grown plants, starting from the meiosis stage. In vitro pollen germination frequencies were determined as described by Touraev and Heberle-Bors (1999). Transgenic and wild-type plants were manually self- and cross-pollinated with rescued pollen, and seeds were subjected to segregation analysis on kanamycin-containing (100 mg/L) medium.

Fertility restoration by in vitro pollen maturation

Microspores at the late unicellular stage were isolated from male-sterile lines and matured for 6 days as described by Touraev and Heberle-Bors (1999). After gradient centrifugation in 60% Percoll (Amersham Biosciences) and B medium with 1 m mannitol to remove dead pollen, the mature pollen grains were re-suspended in germination medium GK. Three microlitres of suspension (3500–5000 grains/µL) were applied to stigmas of emasculated flowers protected with paper bags. Seeds were collected after 3–4 weeks, and subjected to segregation analysis on kanamycin-containing medium.


We are grateful to Dr Robert Dirks (Rijk Zwaan, the Netherlands) for helpful advice, and Dr Ralf Buchner (Department of Ultrastructure Research and Palynology, University of Vienna, Vienna, Austria) for assistance with histological images and access to facilities. This work was supported by the European Commission-funded HybTech project (QLK5-CT-1999-30902) and by a fellowship from the Austrian Exchange Service (ÖAD) to M.A.