Development of late blight resistant potatoes by cisgene stacking

Genetic disease resistance is an effective tool for sustainable management of late blight, caused by Phytophthora infestans, which is economically the most important disease of potato. Breeding at the beginning of the twentieth century concentrated on major dominant late blight resistance (R) genes from the Mexican wild species Solanum demissum and eleven of these R genes were introgressed in potato [1–4]. However, rapid breakdown of resistance in potato varieties containing S. demissum R1, R2, R3, and R10[3, 5] has sparked an increased focus on the introgression of multiple broad spectrum R genes in order to impart durability to commercial varieties. It has turned out in various crops and pathosystems that stacking of multiple R genes is necessary to provide satisfactory resistance in the field [6]. Although the used R genes provide resistance to broad spectra of late blight strains, the predominant agricultural deployment of only one R gene can drive the evolution of new virulent strains. In the absence of chemical controls this might even result in the destruction of an entire harvest [7]. Therefore, the use of combinations of R genes with different spectra must be pursued to increase durability of resistance and thereby providing food security under no or little fungicide application. R gene stacking might be achieved by genetic crossings but the desired variety characteristics will never be fully recovered due to the high level of heterozygosity in potato. Sarpo Mira is an example of a durably late blight resistant potato variety which contains a stack of at least four R genes [8, 9]. Unfortunately, the variety has not acquired a large market share yet because established varieties are preferred by farmers, processors and consumers.

Addition of stacks of cloned R genes [10–17] to existing varieties (resistant or susceptible) through genetic modification (GM) technology is therefore an attractive alternative. Moreover, GM technology circumvents the problem of linkage drag and can speed up the introgression of the R gene [18, 19]. GM technology has, however, met various types of opposition and a major point of criticism concerns the introduction of “foreign” genes into the food chain and environment. However, within the framework provided by cisgenesis only natural genes from the same or crossable species are used [20, 21]. Cisgenes are, therefore, already present in the natural gene pool of the crop plant and cisgenesis only facilitates their introduction into crops. Indeed a majority of a broad panel of European consumers find cisgenic apples safe and not harmfull for the environment [22].

Recently, the transformation of three broad spectrum potato late blight resistance genes (Rpi-sto1, Rpi-vnt1.1 and Rpi-blb3) was described in potato [23]. Rpi-sto1, Rpi-vnt1.1 and Rpi-blb3 are native genes from crossable species and are therefore considered as cisgenes for potato. However, the plants in the study from Zhu et al. [23] are “transgenic” as the selectable marker gene, NPTII, was of bacterial origin. Also beyond the cisgenesis framework it is not desired to introduce antibiotic resistance genes into the environment and in this study, we established a pipeline for Agrobacterium-mediated transformation of potato in the absence of a selectable marker gene (marker-free transformation). After the absence of vector backbone integration was confirmed, these potatoes were designated as “cisgenic” because of the absence of any foreign (non-potato) genes. This is the first scientific report on the production and functional evaluation of cisgenic R gene stacking in different potato varieties.

Cisgenesis, is a new approach for traditional plant breeding that uses genetic modification technology to introduce natural genes from within a plant species or from crossable plant species, into varieties [26]. Therefore, any gene “alien” to the breeder’s gene pool can be avoided in the end product which is causal to many environmental and consumers’ concerns about GM food crops [22]. Not only can widely used susceptible varieties, like Bintje and Atlantic, be converted into resistant varieties, also resistant varieties, like Potae9, can be complemented with additional resistance genes to avoid or delay future resistance breakdown. In order to complement existing varieties with stacks of cisgenic R genes, two choices must be made: 1. The method to introduce the R gene stack and 2. The method to exclude sequences of foreign origin from transformation events. With respect to the introduction method, in this study we chose transformation by marker-free binary vectors and subsequent regeneration in medium without selective antibiotics followed by PCR-based selection of transformation events [27]. Alternatives involving the removal of a selectable marker gene by site specific recombination pose disadvantages because of remnant sequences of foreign origin [28].

The average marker-free transformation frequency was 1.3% and seems to be genotype independent. In a previous marker-free transformation study in potato [27] a T-DNA of 6 kb was transformed with a frequency of 3.5% when A. tumefaciens strain AGL0 was used, and 0.4% when A. tumefaciens strain LB4404 was used. It can not be concluded that AGL1 + virG, which was used in this study, was less efficient in transferring the T-DNA than AGL0 in the study from de Vetten et al. [27]. From unpublished experiments in our laboratory it is known that regeneration time increases with the size of the T-DNA. We, therefore, assume that the lower transformation frequency in our study is rather related to the larger T-DNA size (11 kb) of the Rpi-vnt1:Rpi-sto1 construct. Therefore, for stacking of more than two genes in cisgenic transformation, the effect of an increased insert size (e.g. >11 kb) on transformation frequency remains to be tested. It is known that marker-assisted transformation frequency is highly genotype dependent in potato [29]. Also here we found that transformation frequencies ranged from 10-71% in different varieties (Table 5). This variation was remarkably less (1-2.4%) in marker-free transformation experiments (Table 2). It must be noted that transformation frequencies can vary between different experiments and that we here only performed a limited number of experiments. However, the currently presented experiments show that marker-free transformation is less prone to varietal differences than marker assisted transformation. This could be caused by differences in antibiotic tolerance between the varieties that provides transformed cells different abilities to develop into a shoot.

In terms of vector backbone integration, marker-free transformation apparently produces a lower percentage (24%) of vector backbone integrations compared to marker-assisted transformation (40-50%). Again, the number of experiments is limited and firm conclusions cannot be drawn. The vector backbone and border sequences in pBINPLUS and pBINAW2 are highly similar and we do not expect that these differences affect vector backbone integration. A potential explanation could be that the presence of the NPTII gene directly next to the left border of the T-DNA would stimulate selection of higher levels of backbone integration. As it is known that left border recognition is inaccurate in Solanaceae, [30], especially when agrobacterium strain AGL1 is used [31], positioning of NPTII near the left border would force the integration of the complete T-DNA. So, it might also lead to higher levels of vector backbone integration. In marker-free transformation, six plants out of 14 tested cisgenic plants did not appropriately express Rpi-sto1 as observed using agroinfiltration of the corresponding Avr genes (like H43-1, -10, -12; Table 4). An obvious explanation could be that T-DNA insertion did not proceed all the way to the left border resulting in 3′ truncations of Rpi-sto1. These non-functional cisgenic events and the corresponding DNA samples were discarded in an early phase during the selection and, unfortunately, this hypothesis could not be confirmed. We observed and described some cisgenic plants differing morphologically from wild type varieties in the greenhouse (Table 4, Additional file 1). This is a generally observed phenomenon and in tissue culture-based breeding schemes it should be considered that aberrant plant phenotypes must be selected against [29].

According to the established experimental scheme, it takes less than one year to obtain potato plants with cisgenic R gene stacks from the R gene construct preparation to the functional validation of the resulting cisgenic plants by performing DLAs. As 2-3 shoots per explant can be collected and 30 independent transformed plants are required considering backbone integration and expression, it is recommended that between 1000-1500 explants are to be treated in a marker-free-transformation experiment of potato. The efficiency of PCR analysis can be improved by a factor 10 by pooling ten shoots, so that the labour intensity of the selection of marker-free transformation events is considered reasonable as compared to the marker-assisted transformations. Considering 2-3 years’ field trials, it takes totally 3-4 years to produce late blight resistant cisgenic events in established potato varieties, which can be released for seed tuber multiplication. This time span is remarkably short compared to the conventional breeding scheme. The cisgenic potatoes selected in this study will be further tested for several years to evaluate whether the transferred R genes are stably expressed over many vegetative cycles. Chimeras and epigenetic silencing are issues that could affect stability of resistance. Also agronomic performance needs to be assessed and confirmed in multiple growing seasons.

We have set up and pursued an effective cisgenic marker-free transformation strategy for commercial potato varieties. It was found that marker-free transformation frequency was much less genotype dependent than marker-assisted transformation. Also the frequency of vector backbone integration tended to be lower in the marker-free transformations as compared to the marker-assisted transformations. The susceptibility or the narrow late blight resistance spectra of the selected varieties were upgraded to broad spectrum resistance after the successful introduction of two cisgenic late blight R genes. According to the recent conclusion of the European Food Safety Authority GMO Panel, cisgenic plants have a risk level similar to conventionally bred plants [32]. The cisgenic potatoes, generated in this study, will offer a safe, environmentally friendly, alternative to the current agricultural practice which is highly dependent on the use of chemical late blight control agents. For developing countries, where chemical control agents are unaffordable, cisgenic upgrades of local potato varieties might even ensure food security.