Highly efficient and cost-effective transformation technologies are essential for studying gene function in the major cereal crops, wheat and barley. Demand for efficient transformation systems to allow over-expression, or RNAi-mediated silencing of target genes, is greatly increasing. This is due to technology advances, such as rapid genome sequencing, enhancing the rate of gene discovery and thus leading to a large number of genes requiring functional analysis through transformation pipelines. Barley can be transformed at very high efficiency but the methods are genotype-dependent. Wheat is more difficult to transform, however, recent advances are also allowing the development of high-throughput transformation systems in wheat. For many gene function studies, barley can be used as a model for wheat due to its highly efficient transformation rates and smaller, less complex genome. An ideal transformation system needs to be extremely efficient, simple to perform, inexpensive, genotype-independent, and give the required expression of the transgene. Considerable progress has been made in enhancing transformation efficiencies, controlling transgene expression and in understanding and manipulating transgene insertion. However, a number of challenges still remain, one of the key ones being the development of genotype-independent transformation systems for wheat and barley.
Efficient, high-throughput, and cost-effective transformation technology is vital in order to allow the functional analysis of genes of potential value in crop improvement programmes. This application of crop transformation may not lead to the development of a commercial GM crop but rather will provide a valuable contribution in the process leading to the development of improved conventional crops. However, the direct use of GM technology in wheat and barley may also have future applications in improving the ability of these crops to withstand environmental challenges, in increasing yield, and enhancing nutritional quality.
The two most common approaches utilizing stable transformation technology for the analysis of gene function are either to over-express the gene of interest or to silence it using RNA interference (RNAi)-based silencing. RNAi-based gene silencing technology provides highly specific gene silencing that has been used extensively to investigate gene function. However, alternative transient loss of function assay systems such as viral induced gene silencing (VIGS) can also be used. VIGS has the advantage of being more rapid than stable transformation and can be particularly useful for plants that are very difficult to transform. However, stable (RNAi) gene-silencing systems have a number of advantages including the ability to obtain a range of levels of silencing, the ability to control tissue specificity, and the fact that the silencing is heritable. In this review, the focus is on stable transformation systems.
Barley is the fourth most important cereal crop in terms of production. As well as being an important crop in its own right, it is used as a diploid model for the more complex hexaploid wheat. The first report describing a large number of fertile transgenic barley plants was in 1994 (Wan and Lemaux, 1994). In this report, immature embryos were transformed using microprojectile bombardment. Agrobacterium-mediated transformation of barley followed in 1997 (Tingay et al., 1997) again using immature embryo explants. Following this first report of Agrobacterium-mediated transformation in barley, the technology was quickly adopted as the method of choice. A comparison of biolistic and Agrobacterium-mediated transformation methods in barley was made by Travella et al. (2005). It was found that Agrobacterium gave higher transformation efficiencies as well as giving transgenic plants with lower transgene copy numbers and more stable transgene expression. A recent report of routine and highly efficient Agrobacterium-mediated barley transformation describes a high-throughput method with an average transformation efficiency of 25% for the spring genotype Golden Promise. (Bartlett et al., 2008).
Wheat is considered to be the world’s most important crop but it has lagged behind the other major cereal crops in terms of development of efficient transformation methods. The first successful wheat transformation used microprojectile bombardment of embryogenic callus tissue (Vasil et al., 1992). Agrobacterium-mediated transformation techniques were first reported in 1997 (Cheng et al., 1997). However, despite these first successful reports, the efficiency of wheat transformation remained low and microprojectile bombardment remained the method of choice as it generally gave higher efficiencies than Agrobacterium. Microprojectile bombardment and Agrobacterium-mediated transformation were compared in wheat and, after large-scale experiments, higher transformation efficiencies were achieved with Agrobacterium (Hu et al., 2003). In addition, the quality of the transformation events was higher as they had lower transgene copy numbers, thus agreeing with the findings in barley (Travella et al., 2005). High quality transformation events are usually considered to be single copy insertions of the gene of interest without additional backbone sequences or other rearrangements and with stable expression of the transgene over generations. Hu et al. (2003) used immature embryos as explants for transformation and, indeed, throughout the history of the development of transformation methodology, immature embryos have been the preferred explant for both wheat and barley. Agrobacterium-mediated transformation is now able to yield efficiencies up to 30% in wheat (Risacher et al., 2009). However, Agrobacterium-mediated transformation is limited to specific wheat genotypes whereas biolistics methods are applicable to a much wider range (Sparks and Jones, 2009).
Although high transformation efficiencies are essential to minimize the effort required to produce sufficient numbers of independent transgenic plants for analysis, there are many other desirable features of transformation systems. Simplicity and low cost is desirable to allow maximum access to the technology. The ability to obtain the required level and pattern of expression of the transgene is also important and, in some cases, it is necessary to control, or at least to determine and understand, the site of transgene insertion. High-throughput transformation systems are usually developed for a single responsive genotype and are not transferrable to alternative genotypes. An ideal transformation system would be genotype independent but this remains one of the key challenges for both wheat and barley. In this review, each of these features of transformation systems is considered in turn. The aim is not to provide an exhaustive review but to focus on some recent advances leading to improved barley and wheat systems and to identify the challenges still remaining.
Improving transformation efficiency
There are three areas to consider when attempting to improve transformation efficiency, firstly, improving regeneration from the chosen target tissue, secondly, increasing the number of transformation events and, thirdly, improving the ability to select transformation events.
The first key requirement for a successful transformation system is a highly regenerable target tissue. In barley, immature embryos have been the target tissue of choice (Fig. 1A) although alternative targets such as microspores (Shim et al., 2009) and ovules (Holme et al., 2008) have been used. In wheat, immature embryos have again been the most widely used target, although alternatives such as inflorescence tissue (Amoah et al., 2001) and callus derived from mature embryos have been used (Wang et al., 2009) (Fig. 1B). A range of culture media components have been evaluated, or the levels of key nutrients manipulated, to enhance regeneration in different culture systems. For example, Thidiazuron (TDZ) has been shown to act as a powerful growth regulator in cereals (Schulze, 2007) with the potential to enhance regeneration. Picloram has also been shown to be beneficial in some cereal cultures giving better regeneration than 2,4-dichlorophenoxyacetic acid (2,4-D) (Barro et al., 1999). Lipoic acid, an antioxidant, has also been shown to be effective at increasing the number of responding embryogenic calluses in wheat transformation experiments (Dan et al., 2009). However, simple modifications to culture media components have also had a very significant effect in enhancing regeneration. Bartlett et al. (2008) developed a high-throughput transformation system for barley and, during the optimization of the system, changes were made to the level of copper in the culture media. It was found that the addition of extra copper led to the regeneration of double the number of shoots from each immature embryo and this, in turn, increased transformation efficiency.
Despite the improvements made by adding or changing the concentration of various media components, these changes only improve regeneration in certain genotypes and in certain culture systems. The regeneration of green plants in tissue culture is known to be under genetic control and a number of studies have identified qualitative trait loci (QTL) associated with the regeneration response. In barley, for example, Bregitzer and Campbell (2001) described the identification of QTL for in vitro plant regeneration. Recently, Tyagi et al. (2010) have identified a number of genes located within QTL peaks associated with green plant regeneration in barley. These genes included a ferredoxin-nitrite reductase (NIR) gene. In rice, a NIR gene has been shown to be involved in plant regeneration (Nishimura et al., 2005) and high NIR activity was linked to high regeneration ability when a range of varieties were examined. Other genes potentially involved in regeneration response include hormone biosynthesis genes and genes involved in hormonal response. Ethylene is known to influence plant regeneration in barley and Jha et al. (2007) showed that regeneration could be influenced by manipulating ethylene. For example, addition of 1-aminocyclopropane 1-carboxylic acid (ACC), an ethylene precursor, increased regeneration in the genotype Morex but did not further improve regeneration in Golden Promise. Studies such as these indicate that, in order to improve regeneration efficiency in a range of genotypes, it is first necessary to understand the genes involved in the regeneration response and their expression patterns and then to adapt or tailor the strategies to improve regeneration to the required genotype.
The second approach to improving transformation efficiencies is to increase the number of transformation events. Recent studies have almost exclusively concentrated on using Agrobacterium as the DNA delivery system and looked at ways of improving the number of transformation events. In order to do this, target tissues have been subjected to a range of treatments, for example, vacuum infiltration and sonication were used to improve transformation in barley (Shrawat et al., 2007). In rice and maize, it has been reported that treatment of immature embryos by centrifugation and heat improves transformation efficiency (Hiei et al., 2006). The presence of additional virulence genes during Agrobacterium-mediated transformation of durum wheat was found to be necessary for successful transformation (Wu et al., 2008). The addition of certain virulence genes has also shown beneficial effects in rice transformation (Vain et al., 2004). Recent studies in Arabidopsis have shown that the over-expression of some histone genes increases susceptibility to Agrobacterium and thus increases transformation efficiency (Tenea et al., 2009). It is thought that the histone proteins have a role in protecting the DNA introduced to the plant cells. Other plant genes are also known to be involved in Agrobacterium-mediated transformation and are reviewed by Gelvin (2010). It is likely that, as our understanding of the role of different plant genes in the transformation process increases, there will be new opportunities to exploit this knowledge to increase the frequency of Agrobacterium-mediated transformation events in wheat and barley.
After regeneration efficiency from the target tissue and the number of transformation events has been optimized, it is still necessary to have an efficient system to identify the resulting transformation events. Efficient, cost-effective transformation systems must be able to select transgenic plants and avoid non-transgenic ‘escape’ plants coming through the system, as this leads to time-consuming additional analysis. In a number of cereals, including wheat and barley, the hygromycin resistance gene provides an effective selection system that rarely allows escape plants to survive (Harwood et al., 2009). However, a number of other selectable marker genes, including the bar gene, conferring resistance to the glufosinate group of herbicides, are also effective (Harwood and Smedley, 2009; Wu et al., 2008). Some selectable marker genes can provide a metabolic advantage to the transformed cells and this is referred to as positive selection, for example, the phosphomannose isomerize gene (pmi). The pmi gene has been used to select transformed tissues in many species including wheat (Gadaleta et al., 2006) and has been promoted as being more desirable than herbicide or antibiotic resistance genes from a biosafety/regulatory point of view. Where transformation is used simply as a research tool to determine gene function, the presence of a selectable marker is usually not a problem. In addition, ‘clean-gene’ technology, which allows the use of the selectable marker gene of choice, and then allows the selectable marker to be segregated away from the gene of interest in the progeny, yielding plants containing only the gene of interest, is readily available (Thole et al., 2007). Marker gene elimination has been demonstrated in barley (Matthews et al., 2001) and in wheat (Permingeat et al., 2003) and thus efficient selection of transgenic plants is not now an important limitation.
Controlling transgene expression
The ability to obtain the required level and pattern of transgene expression is another key requirement for any transformation system. The first consideration for controlling transgene expression is the choice of promoter. This area has been recently reviewed by Hensel et al. (2011) where the authors provide a list of promoters shown to be effective in the cereals. A number of promoters give good constitutive expression throughout the plant, for example, the maize ubiquitin promoter that is widely used in both wheat and barley. A comparison of gus gene activity in barley under different promoters is described by Himmelbach et al. (2007) who found the highest expression levels with the maize ubiquitin promoter. A range of tissue-specific promoters have also been shown to be effective in wheat and barley. Seed-specific promoters were examined in both wheat and barley (Furtado et al., 2009) and while some showed similar activity in both species, one pericarp-specific promoter was only active in barley. This highlights the need for care when choosing an appropriate promoter to control transgene expression. The range of available promoters has now been extended to include those induced by biotic or abiotic stress. For example, a promoter from a member of the barley Germin-like GER4 gene cluster has been shown to give high levels of pathogen-induced expression (Himmelbach et al., 2010).
In addition to the choice of promoter, the inclusion of untranslated regulatory elements and of signal peptides also affects the level and pattern of transgene expression. Signal peptides can direct proteins to particular subcellular compartments and have been utilized extensively to direct transgenic proteins to appropriate locations within the cell. The signal peptides used in transgenic wheat and barley are listed in the review by Hensel et al. (2011).
Another available tool for the manipulation of transgene expression is to use intron-mediated enhancement. The inclusion of an expression-enhancing intron within the transgene coding sequence has been shown to give higher levels of expression in both wheat and barley. In wheat, the fourth intron from the maize T3/T7-like DNA-dependent RNA polymerase (RpoT) gene, enhanced expression when included at position +165 within the luciferase transgene (Bourdon et al., 2004). The same intron-containing luciferase transgene, also gave enhanced expression in barley plants compared with controls containing a luciferase gene without an additional intron. However, inclusion of an alternative intron, the first intron from the Arabidopsis polyubiquitin 10 gene (UBQ10) gave even greater enhancement of expression in barley with an almost 3-fold increase over the control plants (Bartlett et al., 2009). Therefore intron-mediated enhancement offers an additional tool to optimize the expression of transgenes.
Where gene-silencing (RNAi) constructs are used as opposed to over-expression constructs, the level of gene silencing will be affected by choices made in the design of the construct. Choice of the size and section of the gene to include in the ‘hairpin’ construct may affect the specificity and level of silencing (Helliwell and Waterhouse, 2005). This is in addition to the effect of the promoter and any other regulatory elements on the activity of the silencing construct.
Understanding and controlling transgene insertion
Knowledge of the exact genomic location of an inserted transgene is important for risk assessment, traceability, and to increase understanding of the transformation process. Transgene insertion sites have been examined using a variety of methods. Fluorescence in situ hybridization (FISH) has been used to provide an efficient method of physically mapping transgenes in barley (Harwood et al., 2004). When this method was used to map 23 transgene insertion sites, the pattern of insertion appeared to be non-random (Salvo-Garrido et al., 2004) with evidence of clustering of independent transgene insertions and insertion in gene-rich areas of the genome. Figure 2 shows the analysis of transgene insertion using FISH in barley and the distribution of insertion sites along chromosome 5H. Transgenes can also be genetically mapped in relation to known genetic markers (Salvo-Garrido et al., 2004) to provide further evidence of the location of the transgene insertion.
Despite the availability of physical and genetic methods to gain information on the site of transgene insertion, molecular analysis of transgene insertion has yielded the greatest volume of information. Analysis of the junction between the inserted transgene and the host plant DNA can yield valuable information that can inform safety assessment of the particular transformation event. For example, it can detect unexpected rearrangements at the transgene insertion site and determine whether the insertion has disrupted known endogenous genes or regulatory sequences. It can also show the presence of plasmid backbone sequences that have been inserted along with the T-DNA. Wu et al. (2006) showed that, in wheat, approximately two-thirds of lines contained some vector backbone DNA. A variety of PCR-based methods are available to isolate transgene junction sequences (Cullen et al., 2011). The region of plant DNA adjacent to the transgene is also referred to as the transgene flanking region. Isolation of transgene flanking sequences is far more straightforward in transgenic lines generated using Agrobacterium-mediated techniques compared with biolistic-derived lines, as the T-DNA borders provide a starting point for the various DNA walking techniques. T-DNAs may insert within genes or within their regulatory sequences, thus providing a source of mutations that can be utilized for the analysis of gene function. Large T-DNA insertion mutant populations have been generated in rice and Arabidopsis (Sallaud et al., 2004), and these have been extensively utilized in functional genomics studies. There is no such resource available in wheat or barley where the large genome size and the fact that the majority of the genome is made up of repetitive elements, mean that generating enough T-DNA insertions to target a useful number of genes would be an enormous task.
It is well known that if a population of independent transgenic lines is developed, each containing a single copy of the same transgene, then a range of different transgene expression levels will be observed (Bartlett et al., 2009). This is explained by ‘position effects’, a term used to describe the variation caused by the particular genomic environment of the transgene. Many attempts have been made to target transgenes to specific locations within the plant genome, in order to remove this source of variation and to deal with the regulatory concerns that arise due to the inability to control the exact site of transgene insertion. In 2009, the first report of the use of zinc-finger nucleases to target a transgene to a specific locus in a crop plant emerged (Shukla et al., 2009). In this example, a herbicide tolerance gene was targeted to a specific location within the maize genome. Zinc fingers can be used to make specific small insertions or deletions at target sites, to generate specific sequence changes using a homologous recombination-based approach, or to target a transgene to a particular genomic location (Porteus, 2009). Following the successful demonstration of this technology in maize, it should offer a method for targeted insertion in wheat and barley.
There is an increasing need to introduce multiple transgenic traits into crop plants. The grouping of a range of separately transformed transgenes into a cultivar of choice can be complicated and time-consuming if a number of independently segregating traits are involved. A solution would either be to introduce all the transgenes at once or to target new transgenes to an existing transgenic locus. Available methods for the simultaneous introduction of multiple transgenes to plants are reviewed by Dafny-Yelin and Tzfira (2007). Naqvi et al. (2010) review successes in metabolic engineering in plants by the introduction of multiple genes either using Agrobacterium or direct DNA transfer methods. For example, corn modified to have enhanced levels of three separate vitamins by the modification of three different metabolic pathways was described by Naqvi et al. (2009). However, the option to add additional transgenes to an existing transgenic locus is attractive and more flexible. Ow (2011) reviews the available options for using recombinase-mediated approaches for gene stacking. Recombinase-mediated methods for the insertion of additional transgenes at a previous transgene insertion site have been demonstrated in soybean (Li et al., 2010). Although technical challenges still remain with the technologies for achieving targeted transgene insertion, these technologies should allow the goal of targeted transgene insertion in wheat and barley to be achieved.
Genotype dependence in wheat and barley transformation
One key remaining challenge for most crop transformation systems is to overcome genotype dependence. In barley, the most successful, high-throughput transformation systems use the spring variety Golden Promise (Bartlett et al., 2008). This genotype has excellent regeneration from immature embryo target tissues as well as good susceptibility to Agrobacterium infection. An assessment of the transformation efficiency for Golden Promise compared with three UK malting barley varieties showed that, whereas Golden Promise gave transformation efficiencies up to 35%, the other three varieties failed to yield any transformation events (WA Harwood, unpublished data). Figure 3A–D shows examples of the appearance of callus, following transformation, from all four varieties. Callus resistant to the selective agent hygromycin is only seen from Golden Promise and from one immature embryo of the variety Tipple. However, the Tipple callus failed to regenerate plants. This demonstrates the challenge of trying to transform a particular commercial genotype. Golden Promise is a gamma-ray induced mutant of the cultivar Maythorpe (Forster, 2001). Thus it is possible that a mutation present in Golden Promise confers the high transformability. However, when Maythorpe was transformed alongside Golden Promise it was possible to achieve transformation efficiencies of 25% in Maythorpe, demonstrating that both varieties were highly transformable (WA Harwood, unpublished data). Figure 3E and F show the similarity of the regenerable callus generated from Maythorpe compared with callus from Golden Promise. This clearly demonstrates that transformability in barley is under genetic control.
There are examples throughout the literature of particular wheat and barley genotypes being amenable to transformation. In wheat, more varieties can be transformed using biolistic techniques than using Agrobacterium. Sparks and Jones (2009) describe a biolistics protocol that has allowed the transformation of 35 wheat genotypes while only a few genotypes can be transformed using Agrobacterium. This is probably because, although a number of genotypes have the ability to regenerate green plants from immature embryos, only a few of them are also susceptible to Agrobacterium. There has been one report of a method for the transformation of barley that is considered to be genotype-independent. This involves the infection of in vitro-cultured ovules with Agrobacterium (Holme et al., 2008). However, isolation of the ovule target tissue is a skilled procedure and not suitable for a high-throughput transformation system.
High-throughput transformation systems in wheat and barley must use a target tissue that is easily isolated and highly regenerable. For both wheat and barley the most suitable target is immature embryos. Agrobacterium is currently the preferred DNA delivery system due to the advantages offered in terms of low copy number and stability of transgene expression. As both the regeneration of green plants from immature embryo target tissues and susceptibility to Agrobacterium are under genetic control, an understanding of the genes involved is probably a prerequisite for developing strategies to overcome genotype dependence.
Transformation efficiencies in both wheat and barley continue to increase and this is allowing the demand for an evaluation of gene function using transgenic tools to be met. However, the pace of gene discovery is also increasing, with the availability of more sequenced crop genomes and improved genomics tools, meaning that even more genes will need to go through a transformation pipeline to allow the study of gene function. A range of tools are available to help achieve the level and specific pattern of transgene expression required, but there are still gaps in the range of promoters available for use in wheat and barley that need to be addressed. Transgene integration has been widely studied and it is now possible to target transgenes precisely to particular genomic locations. However, further advances in this technology are required before it can be used routinely in wheat and barley. A key remaining challenge is the genotype dependence of most wheat and barley transformation systems and this continues to restrict the application of the technology. It is, however, likely that, by understanding and manipulating plant genes important in either the plant regeneration process or in susceptibility to Agrobacterium, it will be possible to address issues of genotype dependence and to improve transformation efficiencies further.
Support by grant in aid to the John Innes Centre from the UK Biotechnology and Biological Sciences Research Council is gratefully acknowledged. Haroldo Salvo-Garrido is gratefully acknowledged for contributing Fig. 2 and Eva Medvecka for Fig. 1. Penny Sparrow is thanked for critical reading of the manuscript.