Introduction

In doubled haploid technology, microspores are particularly attractive as targets for genetic transformation, due to their haploid and single-cell nature. Through the application of a wide range of stressing agents, a haploid microspore can be deviated to become a haploid androgenic embryo and, upon chromosome doubling, eventually a doubled haploid plant (Zhou et al. 2002a, b; Gu et al. 2003, 2004; Zhang et al. 2006a, b; Seguí-Simarro and Nuez 2008a, b). Transformation of microspores prior to androgenesis induction has the potential to avoid not only hemizygosis but also chimerism, generating embryos and plants where all of their cells express the introduced transgene in homozygosis (Dormann et al. 2001). However, the different attempts to produce transformed microspores have yielded variable results in terms of transformation efficiency. Some authors have reported a relative success for transient and stable integration of introduced marker genes into the microspore/pollen grain genome via different methods such as electroporation (Obert et al. 2004), microinjection (Jones-Villeneuve et al. 1995), microprojectile bombardment (Jardinaud et al. 1995; Stoger et al. 1995; Fukuoka et al. 1998; Nehlian et al. 2000), and infection with Agrobacterium (Pechan 1989; Dormann et al. 1998, 2001), but a reliable, efficient, and general protocol is still to be presented.

Among the different procedures used for transformation, the most efficient and widely used is infection with Agrobacterium (agroinfiltration). Agroinfiltration has been extensively used to transform a variety of plant tissues and species (Sharma et al. 2005). Additional treatments for physical wounding by sonication, silicon carbide, sand (Singh and Chawla, 1999; reviewed in Sharma et al. 2005), or microprojectile bombardment (Bidney et al. 1992) prior to agroinfiltration have proven to enhance the efficiency of Agrobacterium-mediated transformation. However, agroinfiltration of microspores or pollen grains has been for long reported as extremely difficult if not impossible (Potrykus 1991; Sangwan et al. 1993). The thick cell wall of the microspores has been traditionally considered as the main obstacle for Agrobacterium to penetrate the cell, and published alternatives to overcome this physical barrier are limited (Dormann et al. 2001). Therefore, the development of additional or alternative strategies to effectively introduce transgenes still deserves attention.

Direct gene transfer by microprojectile bombardment (biolistics) seems the best method to surpass the microspore wall barrier and reach the cell nucleus (Klein and Fitzpatrick-Mcelligott 1993; Sharma et al. 2005). However, biolistics is known to be less efficient than Agrobacterium-mediated gene transfer (Sharma et al. 2005). Thus, it would be desirable to combine the advantages of both techniques into an integrated protocol for microspore transformation. In this study, we focus on the development of a strategy to overcome the difficulties imposed by the microspore wall to the entry of exogenous DNA for microspore transformation. We report on an efficient method to transform Brassica napus microspores using an integrated bombardment and agroinfiltration system that combines the advantages of both methods in terms of delivery and expression of exogenous DNA, and generation of microwounds to enhance the efficiency of bacterial infection.

Materials and methods

Plant material

Brassica napus cv. PF704 plants were used as donor of microspores, grown in a growth chamber with a 16 h photoperiod (300 μE m−2 s−1) and 15/10°C day/night temperature.

Vectors and preparation of bacteria

A pBGWFS7-64 binary vector (kindly provided by Dr. Kim Boutilier, Plant Research International, Wageningen, The Netherlands) carrying the GUS reporter gene under the control of the pollen-specific p28640 promoter (AT2G28640, Exo70 gene family), was used for both microprojectile bombardment and infiltration with Agrobacterium tumefaciens (strain LBA4404).

Agrobacterium tumefaciens strain LBA4404 harboring the binary vector pBGWFS7-64 was taken from a single colony and grown at 28°C for 24 h in LB medium (10 g/l tryptone, 5 g/l yeast extract, 5 g/l NaCl, pH 7.5) supplemented with 50 mg/l rifampicin and 100 mg/l streptomycin, under continuous shaking. Finally, bacteria were centrifuged and suspended in liquid NLN-13 medium at different concentrations depending on the experiment (see results).

Preparation of microspores and transformation procedure

Prior to transformation experiments, microspores were isolated, centrifuged, and resuspended in NLN-13 medium according to Seguí-Simarro et al. (2003). For bombardment, 100 μl culture droplets were deposited onto polyester filters over solid NLN medium with different concentrations of sucrose and mannitol. Bombardment was performed either with DNA-free or DNA-coated gold particles of different sizes, using a Bio-Rad PDS-1000/He biolistic particle delivery system. After bombardment, microspores were transferred to liquid NLN medium with the same sugar and mannitol concentration used for bombardment, and 0.5 ml of a suspension of A. tumefaciens at different concentrations was added. Then, they were incubated for 24 h at 28°C in darkness, treated with carbenicilin, washed, and analyzed for GUS expression.

GUS histochemical assay

Microspores were analyzed for GUS expression using a procedure modified from Jefferson et al. (1987). The substrate buffer contained 200 mM NaPO4 buffer (pH 7.0), 1 mM K3[Fe(CN)6], 1 mM K4[Fe(CN)]6, 20 mM EDTA, 13% sucrose. and 0.1% (w/v) X-GLUC (5-Bromo-4-Chloro-3-Indolyl-D-gIucuronic acid). For each sample, 0.5 ml of the substrate buffer was added to 3 ml of solid NLN-13 medium in 10 × 35 mm Petri dishes. Samples were then placed on top and incubated for 4 or 24 h at 37°C. Finally, samples were placed at 4°C until counting and analysis.

Results and discussion

First, we set up the conditions for microprojectile bombardment over rapeseed microspores. We tested three critical parameters previously described as strongly influencing the efficiency of transformation in other systems (Vain et al. 1993; Jardinaud et al. 1995; Folling and Olesen 2001), including the amount of microspores to be bombarded in each shot, the size of the gold microprojectile particles, and the effect of different concentrations of osmoticum in the bombardment medium (Table 1). Our results showed that transformation efficiency is strongly dependent upon microspore density in the bombardment medium, with an optimal density of 105 microspores in the 100 μl of the bombarded sample. It is easy to conceive that the density of microspores must have a critical influence on the rate of transformation, since only those microspores of the top layer would be exposed to the gold particles. Thus, it is essential to maximize the number of exposed microspores, but always keeping them distributed as a monolayer (Figs. 1b, c). Increasing microspore density beyond 105 microspores/100 μl had a negative effect, and similar for the assays with 2×105 microspores/100 μl and those with 3×105 microspores/100 μl. This suggests that higher densities have an additional detrimental influence on microspore viability after bombardment.

Table 1 Optimization of parameters for particle bombardment
Full size table
Fig. 1
figure 1

Effect on transient GUS expression of the use of agroinfiltration only, bombardment only with DNA-coated particles, integrated bombardment with DNA-free particles and agroinfiltration, and integrated bombardment with DNA-coated particles and agroinfiltration, at different Agrobacterium concentrations. Means with the same letter are not significantly different at P = 0.001

Full size image

Particle size was found to have an effect on the efficiency of transformation. The best results were achieved with a combination of particles of 0.6 and 1.0 μm. This combination clearly overperformed other particle sizes, either combined or alone. The combined use of two gold particle sizes has been proven useful for microspore transformation in wheat (Folling and Olesen 2001). As for wheat, it is likely that in rapeseed highly-sized particles may carry more DNA but could irreversibly damage the microspore, whereas smaller particles are capable of transferring the vector into the nucleus without compromising cell viability.

The presence of additional osmotic agents was also proven positive for transformation efficiency. Whereas different sucrose concentrations did not affect the efficiency, a moderate concentration (0.2 M) of mannitol increased the percentage of GUS-expressing microspores independently of the concentration of sucrose used. Several reports support the notion of a beneficial effect of increased osmolarity before and during bombardment, in the enhancement of transgene expression (Vain et al. 1993). It was previously hypothesized that a mild plasmolysis of the bombarded cells may prevent the cell contents from escaping through the microholes created by bombardment. In our rapeseed microspore system, this effect would be achieved by a 0.2 M concentration of mannitol, while higher concentrations would negatively affect cell viability.

After setting up the best conditions for bombardment, we compared the efficiency, in terms of transient GUS expression, of the optimized bombardment protocol with respect to agroinfiltration alone, agroinfiltration after bombardment with naked (DNA-free) gold particles, and agroinflitration with DNA-coated particles (Table 2). For all of the assays involving Agrobacterium-mediated transformation, we tested different bacterial concentrations. Figure 1 and Table 2 show the experiments using concentrations of 0 (control with no bacteria), 4 × 10−3/100 μl (OD 600nm = 0.004), 8 × 10−3/100 μl (OD 600nm = 0.008), and 16 × 10−3/100 μl (OD 600nm = 0.016). For concentrations higher than 16 × 10−3/100 μl (OD 600nm = 0.016; data not shown), the liquid nature of the medium for microspore and bacterial coculture, and the presence of high sucrose levels gave rise to excessive Agrobacterium proliferation and contamination of cultures. At these levels of bacterial concentration, contamination was impossible to control even with high concentrations of carbenicilin. Therefore, we decided to focus on a range up to OD 600nm = 0.016. In terms of GUS expression, bombardment with DNA-coated particles was clearly more effective than single agroinfiltration or agroinfiltration after bombardment with DNA-free particles, for all of the Agrobacterium densities tested (Table 2). Bombardment with DNA-bearing microprojectiles seems to be critical for a high rate of GUS expression. This indicates that just perforation of the microspore wall to allow for bacteria to easily access the microspore may not be enough for efficient transgene expression. The ability of gold particles to directly carry DNA fragments into the cell nucleus would allow for the observed high level of GUS expression. This notion can be confirmed by the comparison of the results of transformation with agroinfiltration after bombardment (Fig. 1). When DNA-coated particles were used (Fig. 2a), GUS-expressing microspores were more abundant than with DNA-free particles (Fig. 2b). Although microperforation of the microspore coat is equivalent in both methods, particle coating provided a GUS expression level ~10-fold higher.

Table 2 Assessment of different strategies for microspore transformation
Full size table
Fig. 2
figure 2

a GUS-expressing microspores transformed by DNA-free bombardment + agroinfiltration. b Microspores transformed by DNA-coated bombardment + agroinfiltration. Bars: 40 μm

Full size image

Despite of the fact that Agrobacterium-mediated transformation is a widely used method to transform a variety of plant tissues (Sharma et al. 2005), including Brassica tissues (Cardoza and Stewart 2004), its efficiency for microspore transformation is limited (this study; Stoger et al. 1995). In parallel, it is known that mechanical wounding may increase the efficiency of Agrobacterium-mediated transformation (Singh and Chawla 1999; Hoshi et al. 2004). In this study, we demonstrate that the generation of microholes with DNA-free particles is sufficient to allow for Agrobacterium penetration and to achieve an effective transformation of rapeseed microspores, otherwise very difficult if not impossible. Similar results were observed by Bidney et al. (1992) in sunflower meristems and tobacco leaves, and by Droste et al. (2000) in soybean. However, it must be noted that in our case, the difficulties imposed by the exine coat are higher than in somatic tissues. In this scenario, the generation of physical channels through the exine seems critical. Nevertheless, among all of our experiments, the best results were obtained with the combination of bombardment with DNA-coated particles, followed by infiltration with Agrobacterium (Figs. 1, 2a), in a manner dependent on the bacterial density used (Table 2). This method has the potential to combine the advantages of the Agrobacterium transformation system with the high efficiency of biolistic DNA delivery. Besides, the addition of a controlled bacterial coculture step after bombardment does not seem to contribute any additional constraint to the viability of microspores. Although the combined transformation method reported hereby only shows transient transgene expression, we show that this approach is a promising method to incorporate into the experimental protocols for eventually obtaining transgenic, non-chimeric androgenic embryos, and double haploid plants. In summary, this study sets up the conditions for an efficient transformation protocol. Based on this protocol, further improvements in both the viability, survival and development of microspores/embryos and the stability of transformation, can now be addressed.