• Open Access

Transgenic Pm3b wheat lines show resistance to powdery mildew in the field


(Tel +41 44 634 8230; fax +41 44 634 82 04;
email bkeller@botinst.uzh.ch)


Plant resistance (R) genes are highly effective in protecting plants against diseases, but pathogens can overcome such genes relatively easily by adaptation. Consequently, in many cases R genes do not confer durable resistance in agricultural environments. One possible strategy to make the use of R genes more sustainable depends on the modification of R genes followed by transformation. To test a possible transgenic use of R genes, we overexpressed in wheat the Pm3b resistance gene against powdery mildew under control of the maize ubiquitin promoter. Four independent transgenic lines were tested in the greenhouse and the field during 3 years. The four lines showed a five- to 600-fold transgene overexpression compared with the expression of the endogenous Pm3b gene in the landrace ‘Chul’. Powdery mildew resistance was significantly improved in all lines in the greenhouse and the field, both with naturally occurring infection or after artificial inoculation. Under controlled environmental conditions, the line with the strongest overexpression of the Pm3b gene showed a dramatic increase in resistance to powdery mildew isolates that are virulent on the endogenous Pm3b. Under a variety of field conditions, but never in the greenhouse, three of the four transgenic lines showed pleiotropic effects on spike and leaf morphology. The highest overexpressing line had the strongest side effects, suggesting a correlation between expression level and phenotypic changes. These results demonstrate that the successful transgenic use of R genes critically depends on achieving an optimal level of their expression, possibly in a tissue-specific way.


Wheat is the most important crop species after rice for human nutrition and has been constantly improved by breeding efforts (Feuillet et al., 2008). Important traits for wheat breeders are yield and grain quality, but the chance that wheat plants can fully develop these traits depends on their resistance/tolerance to biotic and abiotic factors. Therefore, wheat lines are intensively selected for disease resistance. Major disease resistance (R) genes have been widely used in wheat breeding for many decades because they are genetically simple and very effective. They act as immune sensors that recognize directly or indirectly pathogen-derived molecules, the effectors (Chisholm et al., 2006; Jones and Dangl, 2006). This recognition induces defence responses that limit pathogen growth (Lam et al., 2001; Hückelhoven, 2007). The majority of R genes confer resistance to only a subset of all races of a pathogen species, resulting in the so-called race-specific resistance. However, single major R genes are often rapidly overcome in lines with widespread cultivation. Therefore, breeders are constantly challenged to broaden the resistance resources in their breeding programmes.

In addition to classical breeding, several approaches have been used to modify plants for enhanced disease resistance by recombinant gene technology. Collinge et al. (2008) identified three main strategies: (i) the direct interference with pathogenicity or inhibition of pathogen physiology; (ii) the regulation of natural induced host resistance by altering the pathogen recognition or the downstream signalling; and (iii) the pathogen mimicry through pathogen-derived sequences. The latter strategy only applies to viral resistance and was not yet studied extensively in wheat (Sharp et al., 2002; Fahim et al., 2010). Schlaich et al. (2006) provided the first example for an application of the first strategy in wheat that was successful under greenhouse and field conditions. Examples for the second strategy are more numerous (reviewed in Blechl and Jones, 2009), but only a few of these transgenic lines were tested in the field, and only in some of them reductions in disease symptoms were observed (Zhao et al., 2006; Mackintosh et al., 2007; Shin et al., 2008). Transgenic wheat lines overexpressing the resistance gene Lr10 showed increased resistance to Puccinia triticina compared to lines carrying the endogenous copy of Lr10 in growth chamber tests and on detached leaf culture bioassays (Feuillet et al., 2003; Romeis et al., 2007). Under semi-field conditions, these transgenic plants showed a reduction in the 1000-kernel weight, suggesting that fitness costs arise as a result of the Lr10 overexpression (Romeis et al., 2007).

R gene overexpression was also studied in other crops and in model organisms. Overexpression of Prf (Oldroyd and Staskawicz, 1998) and Pto in tomato (Tang et al., 1999), Xa3 in rice (Cao et al., 2007) and BAL (At4g16890) in Arabidopsis (Stokes et al., 2002) resulted in enhanced disease resistance. However, it was also found that overexpression of R genes can cause negative effects by triggering more frequent cell death and thus might result in a yield penalty (Tang et al., 1999). In addition, the constitutive induction of defence responses might lead to plant damage (Stokes et al., 2002). So far, none of these R gene overexpressor lines were tested in the field.

In wheat, only six R genes have been cloned so far, Lr10 (Feuillet et al., 2003), Lr21 (Huang et al., 2003), Pm3 (Yahiaoui et al., 2004), Lr1 (Cloutier et al., 2007), Yr36 (Fu et al., 2009) and Lr34 (Krattinger et al., 2009). The Pm3 powdery mildew resistance locus occurs in seven distinct alleles (Pm3a–g) in the bread wheat gene pool commonly used by wheat-breeding programmes. Pm3a contributes significantly to the powdery mildew resistance in currently grown wheat cultivars of the southern Great Plains in the United States (Chen et al., 2009), while Pm3b, Pm3d and Pm3f are present in current Scandinavian wheat cultivars and breeding lines (Lillemo et al., 2010). The Pm3b allele was first identified in the landrace Chul from Uzbekistan (Briggle, 1966) and is the most frequent functional Pm3 allele among 2325 tested landraces and non-elite cultivars (Kaur et al., 2008; Bhullar et al., 2010a). All seven Pm3 alleles confer resistance to a specific set of powdery mildew races and were molecularly isolated (Yahiaoui et al., 2004, 2006; Srichumpa et al., 2005). A Pm3 allele-mining strategy allowed the cloning of ten additional alleles (Pm3kt) (Bhullar et al., 2009, 2010b; Yahiaoui et al., 2009).

It was previously described that Pm3c and Pm3f confer residual resistance to virulent isolates (Nass et al., 1981), and we hypothesized that this effect might be enhanced in Pm3 overexpressor lines. Here, we describe that strongly overexpressing Pm3b plants showed partial resistance against isolates virulent on lines carrying an endogenous copy of Pm3b. Under field conditions, powdery mildew infection was strongly reduced in four tested Pm3b overexpressor lines compared to the susceptible control lines and to the Pm3b donor line Chul. One of these transgenic lines showed no side effects, while the remaining three exhibited pleiotropic effects in the field.

Our work presented here is part of a joint project of several research groups (wheat consortium; http://www.NRP59.ch). In addition to the experiments described here, the transgenic Pm3b lines were also analysed in agronomical field performance tests adapted from classical variety trials in which disease resistance against several fungi, yield, yield components and kernel quality were determined. An important part of the consortium was devoted to studies of the ecological performance. For example, the transgene x environment interactions were analysed in glasshouse experiments and in the field season 2008 (Zeller et al., 2010), and the genetic and ecological consequences of introgression of transgenic wheat in a wild relative, Aegilops cylindrica, were studied. A set of further biosafety research experiments studied the impact of these transgenic lines on other organisms (von Burg et al., 2010, 2011; Peter et al., 2010; Song-Wilson et al., 2010; Lindfeld et al., 2011).


Generation of transgenic wheat lines expressing Pm3b

We have transformed the powdery mildew resistance gene Pm3b into the susceptible wheat cv. Bobwhite SH 98 26, hence abbreviated Bobwhite, by microprojectile bombardment. T1 offspring were tested for powdery mildew resistance by leaf segment infection tests using an isolate avirulent on Pm3b. Fifteen independent (i.e. deriving from different calli) T1 families segregated for resistance and were characterized by Southern blot analysis. Four families showed the presence of a single copy of Pm3b, five a full length plus one or more truncated copies of Pm3b, and the remaining lines had more complex transgene insertion patterns. In three families, not all transgenic plants were resistant, probably due to transgene silencing. However, in eleven of the T1 families, presence of the transgene co-segregated with the powdery mildew resistance. All resistant plants and one susceptible, non-transgenic (null segregant) plant of these families were grown to T2. As the null segregants (sister lines) are carrying possible (epi)genetic alterations independent of the transgene that potentially occur during the tissue culture and regeneration process, they were used as direct negative controls. Southern blot and PCR analysis showed the absence of the selectable marker gene (Pmi) in the null segregants. For further analyses, we focused on four families and named them Pm3b#1, Pm3b#2, Pm3b#3 and Pm3b#4. Southern blot analysis suggested that Pm3b#1 to Pm3b#3 carry a single full-length copy of the Pm3b transgene and Pm3b#4 one full-length and one truncated copy (Figure S1). Their corresponding sister lines were named Sb#1, Sb#2, Sb#3 and Sb#4, respectively.

Segregation analysis was performed on the T2 families of the events Pm3b#1 to #4 and plants homozygous for the transgene loci were identified. Resistance was tested on leaf segments of seedlings (Figure 1a), and the presence of the transgene was determined by PCR analysis using Pm3-specific primer pairs. In heterozygous (hemizygous) families of Pm3b#1, Pm3b#2 and Pm3b#4, transgenes co-segregated with the resistance phenotype, and the segregation was consistent with a 3 : 1 resistant:susceptible ratio (chi-square test; = 0.75 for Pm3b#1, = 0.37 for Pm3b#2 and = 0.95 for Pm3b#4), indicating a Mendelian inheritance of Pm3b. In contrast, heterozygous families of event Pm3b#3 did not segregate in a 3 : 1 resistant:susceptible ratio (chi-square test, < 0.001), but showed a higher proportion of susceptible plants. The progeny of selfed homozygous T2 plants of Pm3b#3 showed a phenotypical range between full susceptibility to full resistance. In total, 28% of these T3 plants were susceptible (Table S1), although they still carried the Pm3b transgene as determined by PCR analysis. Southern blot analysis showed that these susceptible plants inherited the full-length transgene to at least the sixth generation (data not shown). For further propagations, only resistant plants of Pm3b#3 were used, but in all further generations a fraction of the offspring showed powdery mildew susceptibility (Table S1). When this selection of resistant parents was omitted in the T5 and T6 generation, which were propagated in the field, the proportion of susceptible plants increased to 44% in T6 and to 65% in T7. Thus, the frequency of susceptible progeny from resistant Pm3b#3 plants was similar in the greenhouse and the field.

Figure 1.

 Transgenic Pm3b lines Pm3b#1 to Pm3b#4 confer resistance to powdery mildew. Segments of primary leaves were infected with powdery mildew strain 07296 that is avirulent (a) or 96.10 that is virulent (b) on the Pm3b donor lines Chul and Chul/8*Chancellor. Cultivars Bobwhite and Chancellor were used as susceptible controls.

Pm3b#2 plants show broad-spectrum resistance in the leaf segment assay

Transgenic Pm3b lines were infected with eight different powdery mildew isolates that are virulent on the Pm3b donor line Chul and on its backcross line Chul/8*Chancellor. Infection tests with these isolates on leaf segments of seedlings revealed that Pm3b#1 and Pm3b#4 were as susceptible as their corresponding sister lines and Bobwhite (Table 1 and Figure 1b). In contrast, Pm3b#2 showed lower infection levels to all tested isolates. Individual Pm3b#3 plants differed in their reactions to the same isolate, some being susceptible and others being partially resistant. Owing to the high variance in disease scores, the difference between Pm3b#3 and Sb#3 was only significant for two isolates, 97011 and 96.10. This suggests that the race-specific resistance in wheat lines with the endogenous Pm3b gene is broadened in Pm3b#2 as well as in a fraction of the Pm3b#3 plants to a partial resistance against isolates virulent on lines with the native Pm3b gene.

Table 1.   Infection of control and transgenic Pm3b seedlings with powdery mildew isolates virulent on wheat lines with an endogenous Pm3b gene. Disease was scored as percentage of the surface area of tested leaf segments that was infected with powdery mildew (mean of three replicates ± SE)
  1. Chul/8*CC was compared to Chancellor and the transgenic lines Pm3b#1–4 to their respective sister lines; statistical significance was tested using Student’s t-tests (*< 0.05; **< 0.01; ***< 0.001).

  2. SE, standard error.

  3. Isolate 07296 was used as avirulent control.

Bobwhite75 ± 577 ± 385 ± 587 ± 370 ± 687 ± 7n.d.73 ± 370 ± 10
Chul73 ± 373 ± 783 ± 783 ± 373 ± 983 ± 960 ± 1280 ± 65 ± 3
Chul/8*CC83 ± 377 ± 383 ± 783 ± 377 ± 397 ± 370 ± 680 ± 67 ± 3**
Chancellor80 ± 670 ± 687 ± 387 ± 3n.d.93 ± 773 ± 777 ± 363 ± 7
Pm3b#170 ± 660 ± 1087 ± 373 ± 3*63 ± 767 ± 340 ± 1573 ± 3 0***
Sb#173 ± 777 ± 980 ± 687 ± 360 ± 1083 ± 757 ± 1357 ± 1583 ± 7
Pm3b#247 ± 7*18 ± 2***43 ± 9*42 ± 14*15 ± 3***12 ± 4**7 ± 2***18 ± 7** 0***
Sb#277 ± 373 ± 380 ± 683 ± 377 ± 390 ± 1070 ± 077 ± 387 ± 3
Pm3b#373 ± 1773 ± 360 ± 1283 ± 353 ± 3**70 ± 2137 ± 1537 ± 12*8 ± 3***
Sb#383 ± 783 ± 387 ± 380 ± 683 ± 393 ± 770 ± 673 ± 387 ± 3
Pm3 #483 ± 777 ± 387 ± 383 ± 377 ± 393 ± 767 ± 377 ± 3 0***
Sb#477 ± 387 ± 387 ± 370 ± 1273 ± 393 ± 763 ± 367 ± 978 ± 2

Transgenic Pm3b lines show powdery mildew resistance in the field

We tested whether the powdery mildew resistance observed in leaf segment infection tests is also functional under field conditions. Previous powdery mildew virulence profiling that was performed in Switzerland in 2007 (Brunner et al., 2010) revealed that for all the Pm3a to Pm3g alleles, virulent strains are present. The virulence rate for Pm3b was 4% overall in Switzerland and 7% in the Eastern part of Switzerland where the field site was situated. The powdery mildew resistance trials with the transgenic Pm3b lines were performed during the field seasons 2008, 2009 and 2010 in microplots, and in 2008 also in macroplots. Microplots were flanked on both sides by a row of highly susceptible wheat plants to increase powdery mildew pressure. The test lines in the microplot experiments were subjected to natural powdery mildew infection, to artificial infection, or, as control, were treated with fungicide. Cultivar Bobwhite was highly susceptible in our environment and, therefore, represents an ideal genetic background for powdery mildew resistance studies. In all 3 years and under natural as well as artificial infection, the transgenic Pm3b lines were significantly more resistant than their corresponding sister lines and the non-transformed Bobwhite parent (< 0.001 for all experiments; Table 2). Interestingly, they were also more resistant than the Pm3b donor line Chul (< 0.001 for 2009 and 2010). In the microplots 2008 and in the natural infection in 2010, Pm3b#2 was more resistant than the other transgenic Pm3b lines (= 0.027 in 2008; = 0.008 in 2010), but this effect was only marginally significant in 2009 (= 0.087). In 2008, Pm3b#3 was as resistant as Pm3b#1 and Pm3b#4 (= 0.194). In fact, some plants inside the plots planted with Pm3b#3 were highly infected, but overall these plots showed low infection levels. This is in agreement with the segregation of the resistance phenotypes observed in seedling infection tests (Table S1). In 2009, we used Pm3b#3 seeds derived from T5 plants that were not selected for resistance and therefore showed a higher susceptibility rate (44%). Consequently, Pm3b#3 was significantly less resistant than Pm3b#1 and Pm3b#2 (= 0.002), but still more resistant than the sister line Sb#3 (< 0.001; Table 2).

Table 2.   Field evaluation of Pm3b transgenic and control lines for powdery mildew resistance
YearWheat line*NaturalArtificialNatural
  1. Values report the area under the disease progress curve (AUDPC) (mean of four replicates ± standard error).

  2. *In transgenic and sister lines, generation (T) is indicated.

  3. Natural powdery mildew infection.

  4. Artificial powdery mildew infection.

2008Bobwhite161 ± 7188 ± 353 ± 23
T4 Pm3b#140 ± 911 ± 50
T4 Pm3b#215 ± 56 ± 20
T5 Pm3b#336 ± 1021 ± 140.1 ± 0.1
T4 Pm3b#446 ± 919 ± 80.5 ± 0.5
T4 Sb#1168 ± 5195 ± 2105 ± 5
T4 Sb#2166 ± 4190 ± 584 ± 16
T4 Sb#3171 ± 4191 ± 286 ± 29
T4 Sb#4171 ± 3179 ± 9124 ± 12
2009Bobwhite227 ± 8229 ± 11 
Chul (Pm3b)157 ± 8161 ± 14 
T5 Pm3b#184 ± 1996 ± 11 
T5 Pm3b#281 ± 1780 ± 16 
T6 Pm3b#3139 ± 18110 ± 15 
T5 Sb#1240 ± 9239 ± 5 
T5 Sb#2230 ± 16232 ± 11 
T5 Sb#3238 ± 10229 ± 10 
2010Bobwhite221 ± 6210 ± 8 
Chul (Pm3b)186 ± 4146 ± 14 
T5 Pm3b#139 ± 180 
T5 Pm3b#200 
T5 Sb#1205 ± 11189 ± 14 
T5 Sb#2211 ± 6197 ± 25 

Expression analysis of Pm3b in the field

We developed a reverse transcription, quantitative real-time polymerase chain reaction (RT-qPCR) assay to quantify expression levels of the Pm3b gene. The specific detection of Pm3 transcripts in this sensitive assay was challenging for several reasons. First, Pm3 belongs to a large family of highly similar genes (Wicker et al., 2007; Yahiaoui et al., 2004 and our own unpublished data). The overall sequence homology of the members of this cluster made it difficult to identify specific primer/probe sets. The problem of a large number of highly similar genes is even more pronounced in wheat because it is hexaploid. Second, we designed primers and probes in the leucine-rich repeat (LRR)-encoding region of Pm3b because transcript quantification assays targeting the 3′ end of genes is less susceptible to artefacts of RNA degradation and incomplete cDNA synthesis. As LRR domains are present in a number of protein classes, more sequences could potentially be co-targeted by our assay. Finally, because the wheat genome is not sequenced, in silico analyses on potential sequence identities are not very powerful. As none of the qPCR assays designed by commercial software were specific or sufficiently efficient, we manually designed an assay.

We analysed the Pm3b expression levels of Pm3b#1 to #4 from the powdery mildew field infection tests in 2008 and 2009. RNA samples were taken first from leaves at the stem elongation stage (BBCH 32–37), when infection with powdery mildew started, and second around flowering time (BBCH 61–71), when powdery mildew infection reached its peak level. Four or three plots per line were analysed in 2008 and 2009, respectively. To obtain representative samples, leaves of three different plants from each plot were pooled. From Pm3b#3, which segregated for powdery mildew resistance, we pooled material from six different plants. In 2008, lines Pm3b#1 to #4 were analysed under natural powdery mildew infection pressure. Pm3b expression levels were higher at the end of flowering than at stem elongation stage (= 0.009; Figure 2a). At both time points, they were similar among Pm3b#1, Pm3b#3 and Pm3b#4 (= 0.515), while Pm3b#2 showed around five times higher levels (< 0.001). These experiments were repeated in the following field season with the lines Pm3b#1, Pm3b#2, Pm3b#3 and the Pm3b donor line Chul under three different environments: natural infection, artificial powdery mildew infection and fungicide treatment. RT-qPCR results demonstrated that there was no overall effect of the treatments on the Pm3b expression levels (= 0.141), suggesting that Pm3b expression was not influenced by powdery mildew infection. Pm3b transcript levels of the transgenic lines were significantly different from Chul (< 0.001), being on average 11, 55 and 5 times higher in Pm3b#1, Pm3b#2 and Pm3b#3, respectively, compared to Chul (Figure 2b). These results were consistently found at both sampling time points with one exception: Pm3b expression was about ten times reduced in Chul at flowering time compared with the stem elongation stage, leading to a 130-, 617- and 112-fold expression level difference to Pm3b#1, Pm3b#2 and Pm3b#3, respectively (Figure S2). This indicates a differential expression of the native Pm3b and the ubiquitin promoters during plant development.

Figure 2.

 Relative Pm3b expression level was determined by RT-qPCR. (a) Relative Pm3b mRNA levels in transgenic lines Pm3b#1 to #4 in macroplots 2008. Results are plotted as the ratio to Pm3b#1 expression at the stem elongation stage. Values report the average of four biological replicates and error bars the standard error of the mean (SEM). (b) Pm3b expression level in the Pm3b donor line Chul and the transgenic lines Pm3b#1 to #3 in microplots 2009. Leaves were sampled at the stem elongation stage (BBCH 32/33). Results are plotted as the ratio to Pm3b#1 with natural infection (mean of three biological replicates ± SEM).

To test whether the variability of resistance phenotypes between individual Pm3b#3 plants homozygous for the transgene locus correlates with differences in expression levels, we determined the Pm3b transcript levels of 20 individual Pm3b#3 plants and compared them to those of seven individual Pm3b#1 plants (Figure 3a). While the T6 Pm3b#3 plants showed a wide range of Pm3b expression levels (between 0.01 and 3.42), expression was homogenous in Pm3b#1 (between 0.81 and 1.27; F-ratio = 27.8, < 0.001). These data show that the majority of Pm3b#3 individuals show transgene silencing at different intensity and that three of the tested plants are stronger Pm3b expressors than Pm3b#1. To analyse whether transgene silencing causes the susceptibility observed in Pm3b#3 plants, leaf segments of seedlings were infected with powdery mildew. Resistant (no disease symptoms), intermediate resistant (20–30% of diseased leaf surface) and susceptible (60–100% of diseased leaf surface) plants were used for expression analysis. As control line, Chul/8*Chancellor was used, which carries an endogenous copy of Pm3b and showed no disease symptoms upon powdery mildew infection. RT-qPCR analysis showed that the resistant Pm3b#3 plants overexpressed Pm3b compared to Chul/8*Chancellor, while susceptible plants barely expressed the transgene, revealing transgene silencing (Figure 3b). The intermediate resistant plants showed similar Pm3b expression levels as Chul/8*Chancellor plants (= 0.470). This suggests that similar amounts of Pm3b transcripts lead to a less complete protection against powdery mildew in the Bobwhite genetic background than in the Chancellor genetic background.

Figure 3.

 A fraction of the Pm3b#3 plants show transgene silencing. (a) Pm3b transgene expression levels of field-grown plants at the stem elongation stage (BBCH 32) were determined by RT-qPCR in seven individual Pm3b#1 and 20 Pm3b#3 plants. Values represent the mean of three technical replicates and are plotted as the ratio to the mean of Pm3b#1. (b) Pm3b mRNA abundance in Pm3b#3 plants is highly variable and correlates with the level of powdery mildew resistance. Pm3b#3 seedlings were grown in a climate chamber, and the first leaf was infected with powdery mildew. Seedlings were classified as resistant (R), intermediate resistant (IR) and susceptible (S). Relative expression values depict the mean of five plants of each class plotted as ratio to Chul/8*Chancellor and error bars the SEM.

Pm3b#2, Pm3b#3 and Pm3b#4 show pleiotropic phenotypes under field conditions

When the Pm3b#1 to #4 lines were grown in growth chambers or in the greenhouse, no pleiotropic phenotypes were observed. However, in all three field seasons, leaves of Pm3b#2 and Pm3b#3 turned light green or even yellow (Figure 4a). The chlorosis was not observed in developing leaves, but increased with age in fully grown leaves. All Pm3b#2 plants were affected, while Pm3b#3 plants segregated for this phenotype. Leaf chlorophyll content measured at the end of flowering (BBCH 69) in flag leaves was reduced by 15% in Pm3b#2 and by 8% in Pm3b#3 compared to their sister lines (Table 3), but was most likely even lower at later time points. The stomatal conductance (gs) was also reduced by 21% in Pm3b#2 (245 ± 15.8 mmol/m2s) compared to Bobwhite (312 ± 17.4 mmol/m2s) and Pm3b#1 (298 ± 12.3 mmol/m2s; = 0.002). In Pm3b#2 and Pm3b#3, leaf thickness (as deduced from the measurement of the surface area and fresh weight of single flag leaves) was not altered, indicating that the brighter leaf colour resulted from a lower chlorophyll concentration (data not shown). Considering the high expression level of Pm3b#2 and the positive correlation between the powdery mildew resistance level and the Pm3b transgene expression level in Pm3b#3, strong Pm3b overexpression in Pm3b#2 and in a fraction of Pm3b#3 possibly caused this leaf damage. Interestingly, we observed that chlorotic Pm3b#3 plants showed less growth of powdery mildew, supporting this hypothesis.

Figure 4.

 Some of the transgenic Pm3b lines showed pleiotropic phenotypes under field conditions. (a) Pm3b#2 and a fraction of the Pm3b#3 plants showed chlorotic leaves when grown in the field. Representative pictures of the first leaf below the flag leaf were taken at the ear emergence stage (BBCH 59). (b) Pm3b#2 and Pm3b#4 showed open florets in some environments.

Table 3.   Morphological traits of transgenic Pm3b lines (mean ± SE)
  Plant height (cm) 
LineChlorophyll contentNaturalArtificialFungicidePooled§Seed set (%)
  1. The transgenic Pm3b lines were compared with their corresponding sister lines, and differences within each pair were tested for significance by Student’s t-tests (chlorophyll content) or multiple linear regressions (*< 0.05; **< 0.01; ***< 0.001).

  2. SE, standard error; n.d., not determined.

  3. Data from microplots 2008, fungicide treatment.

  4. Data from microplots 2009.

  5. §Overall mean ± SE was calculated from pooled data of the three treatments.

  6. Data from microplots 2008; data from artificial infection and fungicide treatment were pooled.

Bobwhite41.6 ± 0.593 ± 292 ± 197 ± 194 ± 193 ± 1
Pm3b#141.3 ± 1.188 ± 0*93 ± 290 ± 290 ± 195 ± 1
Sb#142.6 ± 0.392 ± 292 ± 294 ± 193 ± 194 ± 1
Pm3b#234.4 ± 0.8***88 ± 1**87 ± 188 ± 1**87 ± 1***87 ± 3
Sb#240.2 ± 0.494 ± 291 ± 196 ± 294 ± 194 ± 1
Pm3b#336.3 ± 1.1*92 ± 289 ± 192 ± 291 ± 195 ± 1
Sb#339.6 ± 0.693 ± 191 ± 194 ± 293 ± 193 ± 1
Pm3b#438.0 ± 0.3n.d.n.d.n.d.n.d.81 ± 2*
Sb#438.8 ± 0.7n.d.n.d.n.d.n.d.94 ± 1

Pm3b#2 had a reduced plant height compared with its corresponding sister line. This difference occurred in all 3 years (< 0.001; Table 3 and Table S2). Besides the genotype, also the treatment influenced the plant height, as fungicide-treated plants were taller than those without fungicide in 2009 (= 0.004). The same trend was also observed in 2008 (< 0.001), but this difference could also be caused by a block effect. In 2010, plant height was not different between the treatments (= 0.143). Inspection of the individual wheat lines indicated that the overall plant height difference between the treatments in 2009 was actually because of a reduced plant height of the susceptible lines (sister lines and Bobwhite) in the infected plots compared to the fungicide-treated plots (= 0.001). The resistant Pm3b lines had the same height in all three treatments (= 0.478).

A fraction of the Pm3b#2 and Pm3b#4 plants had flowers that remained open after flowering, a phenotype known from male sterile plants (Figure 4b). Indeed, we detected many flowers that did not produce seeds. The overall seed set rates of Pm3b#2 and Pm3b#4 were reduced compared to their corresponding sister lines in microplots in 2008 (Table 3). For Pm3b#4, this effect was significant (= 0.048), and for Pm3b#2 it was marginally significant (= 0.070). In the macroplots 2009 and 2010, this reduction was again observed in Pm3b#4 (< 0.001), but not in Pm3b#2 (= 0.573 in 2009, = 0.298 in 2010; Table S2), indicating that seed set reduction depends on subtle environmental differences. This is also supported by the fact that in 2009, the flower phenotype was not observed in the microplots of the powdery mildew infection trials, but in the propagation plots that were on the same field (data not shown).

In the first two field seasons, we monitored the phenological stages at seven or more time points throughout the growing season. We could not detect any consistent difference in developmental stages between the transgenic and the corresponding sister lines (data not shown).

Among the traits described earlier, we found no consistent difference between the sister lines and Bobwhite. For example, Sb#3 was shorter and Sb#4 was taller compared with the other sister lines (= 0.038 for both; Table S2). However, this difference was not found in 2009 (Table 3). Possibly, these effects are caused by differences in seed properties, as the seeds for 2008 were grown in different greenhouse chambers in different seasons, while the seeds of 2009 were all propagated in the same field in 2008. Together, this indicated that tissue culture-induced effects played a minor role compared to the effects of the transgenes.


The Pm3b transgene confers resistance under field conditions

The four transgenic lines Pm3b#1 to #4 were more resistant than the sister lines and the non-transformed Bobwhite plants in all leaf segment infection tests using isolates avirulent on Pm3b (Figure 1a). This increased resistance was also observed in field environments that have complex mixtures of avirulent and virulent powdery mildew races (Table 2). Interestingly, all transgenic Pm3b lines were more resistant than the Pm3b-containing landrace Chul. The high powdery mildew susceptibility of the parental cultivar Bobwhite and the sister lines suggests that they have no seedling or adult resistance that would provide protection from powdery mildew. Nevertheless, it is also known that the effectiveness of R gene function can depend on the genetic background of the breeding material (Cao et al., 2007). Thus, we cannot exclude that cultivar Bobwhite simply has a more efficient signal transduction mechanism that is triggered by Pm3b-mediated recognition. We can also not fully exclude that the selectable marker gene Pmi, which co-segregated with Pm3b in all four lines, increased powdery mildew resistance. However, such an effect is very unlikely because the responses to powdery mildew were race specific. In addition, there is no evidence from literature that the encoded enzyme phosphomannose isomerase, which catalayses the reversible isomerization of mannose 6-phosphate and fructose 6-phosphate, has an impact on the wheat–powdery mildew interaction. Furthermore, transgenic Lr10 plants that also carry the Pmi marker gene (Feuillet et al., 2003) were as susceptible to powdery mildew as their control lines (our unpublished data). The increased resistance levels of the transgenic lines compared to Chul are explained best by the increased Pm3b expression levels.

In the field resistance tests, Pm3b#2 showed the lowest infection levels compared to the other lines, most likely due to the additional partial resistance against races virulent on the native Pm3b observed in leaf segment infection tests (Table 1, Figure 1b). This phenotype is similar to the ‘residual resistance’ mediated by defeated (overcome) R genes (Poland et al., 2009) that was described in wheat, rice and bell pepper (Nass et al., 1981; Brodny et al., 1986; Kousik and Ritchie, 1999; Li et al., 1999). Again, the Pm3b expression levels that are highest in Pm3b#2 explain best the increased powdery mildew resistance. Several studies have reported that R gene overexpression induces enhanced resistance (Oldroyd and Staskawicz, 1998; Tang et al., 1999; Stokes et al., 2002; Feuillet et al., 2003; Cao et al., 2007). Field trials with potato transformed with the late blight resistance gene RB under control of the native promoter revealed a positive correlation between RB transcript abundance and the resistance level (Bradeen et al., 2009). Bieri et al. (2004) have shown that the level of R gene product is a critical parameter for resistance gene function. In addition, the gene-dosage-dependent action of many R genes with different reaction patterns in homozygous versus heterozygous plants indicates that the amount of R gene product is critical. The mechanism by which R gene overexpression leads to enhanced resistance is not elucidated yet. It was hypothesized that R proteins may have some intrinsic activity, and their overexpression could result in protein levels that suffice to induce detectable defence responses even in the absence of the matching avirulence genes. Abundant R proteins could also titrate proteins that normally repress R protein activation (Oldroyd and Staskawicz, 1998; Stokes et al., 2002; Frost et al., 2004).

Transgene expression and silencing are not affected in field environments

Quantitative transgene expression analysis of field-grown T4 and T5 plants showed that Pm3b expression levels were very similar in 2008 and 2009 and with or without powdery mildew infection (Figure 2). This demonstrates that transgene expression shows little dependence on the environment. Our data suggest that the frequency of transgene silencing in line Pm3b#3 under field conditions followed the same trend as in the greenhouse. The random silencing of only individual, but not all plants of a Pm3b#3 generation, has also been reported for two other transgenic wheat lines where the occurrence of transgene silencing was positively correlated with increased levels of transgene methylation (Anand et al., 2003; Howarth et al., 2005). Transgene silencing is a well-known phenomenon in the transformation of plants (Vaucheret et al., 1998), including wheat (Müller et al., 1996; Chen et al., 1998; Demeke et al., 1999; Alvarez et al., 2000; Iyer et al., 2000; Bourdon et al., 2002; Anand et al., 2003; Howarth et al., 2005; Li et al., 2005). Unstable transgene expression is known to occur more frequently in transformants carrying complex insertion patterns with multiple transgene copies (Vaucheret et al., 1998). However, there are also reports of silenced transgenes that occur as single copy (Elmayan and Vaucheret, 1996), as was the case for Pm3b.

One of four lines showed no pleiotropic phenotypes

We did not observe any pleiotropic phenotypes under the controlled conditions of greenhouses or growth chambers during seed propagation. Under field conditions, however, an altered leaf phenotype and fertility problems occurred. Overexpression of the tomato CC-NB-LRR-type of resistance gene Prf (Oldroyd and Staskawicz, 1998) and the rice Xa3, an LRR-receptor kinase-like R gene (Cao et al., 2007), led to broad-spectrum or enhanced resistance, respectively, without any deleterious side effects. Wheat lines overexpressing the CC-NB-LRR-encoding Lr10 gene were less diseased than the Lr10 donor line, and no pleiotropic phenotypes were observed except for a reduced grain weight under semi-field conditions (Feuillet et al., 2003; Romeis et al., 2007). The overexpression of tomato Pto, which encodes a serine/threonine kinase (Tang et al., 1999) and BAL, an NBS-LRR R protein from Arabidopsis (Stokes et al., 2002), also conferred enhanced resistance. However, these lines exhibited microscopic lesions, or dwarfing and twisted leaves, respectively. These phenotypes are different from the pleiotropic effects in the Pm3b#2, Pm3b#3 and Pm3b#4 lines. Interestingly, however, the phenotypes observed in our study have often been observed in wheat-breeding programmes. For example, chlorosis and reduced fertility were observed in the wheat pre-breeding programme of ICARDA using wild wheat progenitors (Valkoun, 2001).

The molecular basis of the pleiotropic phenotypes in Pm3b#2, Pm3b#3 and Pm3b#4 is not known. As the corresponding sister lines were not affected, the observed effects must be caused by the transgenes. The co-occurrence of strong Pm3b expression, strong powdery mildew resistance and chlorotic leaves in a segregating subset of Pm3b#3 plants strongly suggests that the chlorosis is a consequence of the Pm3b gene product and not of transgene position effects. This is consistent with the finding that the chlorotic leaves were also present in the strongest Pm3b expressor, Pm3b#2. The fact that this chlorosis occurred in two independent events further supports that the transgene products, and not position effects, cause this pleiotropic effect. We cannot fully exclude that the selectable marker gene Pmi is causing pleiotropic effects. However, this is very unlikely because for at least 20 different plant species protocols for a Pmi-based selection were established (Song et al., 2010), but no pleiotropic effects were reported. In field trials with transgenic wheat and rice, no side effects were observed as well (Privalle et al., 2000; Reed et al., 2001; Datta et al., 2007), and nine events of maize carrying Pmi were approved by the regulatory authorities of at least one country (http://cera-gmc.org/index.php?action=gm_crop_database).

The Pm3b transcript dose effect explains well the chlorosis phenotype and the fact that the strongest Pm3b expressor, Pm3b#2, showed all three side effects (chlorosis, reduced fertility and reduced plant height). However, it cannot explain why Pm3b#1 showed no side effects and Pm3b#4 had reduced fertility, while both had similar Pm3b expression levels. The reduced fertility in Pm3b#4 could be caused by position effects via the disruption of endogenous genes or modification of the ubiquitin promoter-driven transgene expression. It was reported that in a minority of independent transgenic wheat lines, the ubiquitin promoter-driven expression was restricted to only a few tissues (Rooke et al., 2000). This suggests that also differences in Pm3b expression levels between different tissues may exist. Since reduced fertility can be caused by many conditions [high and low temperatures, water stress, high humidity, boron deficiency and low radiation (Subedi et al., 1998 and references therein)], a very high PM3B protein content in any of the tissues might interfere with any of a number of cellular processes. Transcriptome and metabolome profiling of specific tissues may help to identify candidate pathways that are affected by the overexpression or ectopic expression of Pm3b. Such analyses have shown to be very sensitive and are used for the assessment of possible adverse effects of transgene expression (Barros et al., 2010; Kogel et al., 2010).

The ubiquitin promoter was reported to be strongly and constitutively active in all tissues (leaves, roots, flowers, immature inflorescence, endosperm and embryo) (Rooke et al., 2000; Jones and Sparks, 2009). Thus, it is possible that the ectopic expression of Pm3b caused some of the pleiotropic phenotypes. Although new natural and synthetic promoters have been described (Rushton et al., 2002; Furtado et al., 2009), they still need to be validated under field conditions. In fact, the ubiquitin promoter is the only promoter that was widely used in transgenic wheat grown in the field. Since powdery mildew attacks only the epidermis of the aerial parts of wheat plants, a promoter that is only active there would be optimal. In fact, the wheat GstA1 promoter fulfils theses criteria and was studied intensively in wheat plants (Altpeter et al., 2005). Therefore, this promoter represents an ideal tool to test whether tissue-specific expression of Pm3b will prevent pleiotropic effects while still conferring powdery mildew resistance.

The sustainable use of Pm3 alleles as transgenes

Since elite wheat cultivars represent only a small fraction of the wheat gene pool, favourable alleles, genes or gene complexes from landraces and wild relatives are used in breeding (Baum et al., 1992; Feuillet et al., 2008). It was suggested that the natural diversity at R gene loci should be better exploited (Hammond-Kosack and Parker, 2003). However, undesirable gene linkages (linkage drag), and in the case of the use of wheat progenitors also the occurrence of fertility problems and strong hybrid necrosis, are limiting the use of valuable genes (Valkoun, 2001). Our study shows that the transgenic (or cisgenic if the native promoter is used) use of an R gene offers a possibility to circumvent this problem. A Pm3 allele-mining strategy showed that the genetic resources indeed can be accessed at the molecular level (Bhullar et al., 2009).

To improve the durability of major R genes, several strategies in breeding and agriculture were developed (McDonald and Linde, 2002). Based on the characteristics of the pathogen, it was suggested that the multiline strategy is most promising for the wheat–powdery mildew pathosystem (McDonald and Linde, 2002). Multilines are mixtures of different lines of ideally identical genetic background, each carrying a different R gene introgressed by backcrossing into the recurrent parent (Mundt, 2002; for reviews see Wolfe, 1985). They have been proven effective in disease reduction in small grain crops (reviewed in Mundt, 2002; Zhu et al., 2000). The transformation of wheat with different, single Pm3 alleles each will allow the production of Pm3 multilines that will be genetically identical except for the Pm3 allele.

A second frequently discussed and partially applied strategy is the pyramiding of resistance genes. In this strategy, several different R genes are combined in the same plant, making the adaptation of the pathogen more difficult (McDonald and Linde, 2002). The transgenic use of Pm3 alleles would allow the stable combination of different Pm3 alleles in the same line which is not possible by classical breeding.

It is commonly known among breeders that greenhouse studies do not unravel the agronomical potential of a line. Not surprisingly, our results are showing the necessity of field trials to evaluate agronomically relevant field resistance. Field trials are also important for basic research, as field conditions might reveal aspects on the gene function that are not detectable under controlled conditions. By only using four independent transgenic Pm3b lines, we detected different pleiotropic effects. Blechl and Jones (2009) suggested that 8–10 events may be required for functional genomics research, 20 or more for applied research and hundreds of different transgenic events for commercial application. In the frame of the current legislation in Switzerland and the European Union, it is very challenging to obtain the required legal permits for field releases, and it is currently not feasible within regular research projects to work on large numbers of events.

Our work shows a promising first step in the transgenic use of an R gene. The further development of this work for integration into breeding would require the testing of a large number of events, the use of more specific promoters and the creation of multilines with different Pm3 alleles.

Experimental procedures

Plasmid constructs used for wheat transformation

A 4.5-kb fragment containing the entire coding region of Pm3b (Yahiaoui et al., 2004) was amplified from genomic DNA by PCR using primers UP6 (5′-GGCACAGACAAAGCTCTG-3′) and N3′SP3R (5′-ACAATCAGGGATCAGGCC-3′) (Srichumpa et al., 2005) and nested primers BamHI-1 (5′-TTAATTGGATCCCAATGGCAGAGCGGGTGGTC-3′) and BamHI-2 (5′-TATATAGGATCCGCTTCAGCTCCGGCAGGCCTG-3′). The BamHI fragment of the amplicon was cloned under the control of the maize ubiquitin promoter (ubi) with the nopaline synthase terminator (nos) into the BamHI site of vector pAHC17 (Christensen and Quail, 1996) and verified by sequencing. For transformation, only the Pm3b gene cassette (ubi:Pm3b) was utilized, which was released from the vector backbone by enzymatic digestion using NotI restriction sites previously introduced into pAHC17 5′ of ubi and 3′ of nos (S. Travella, unpublished). Similarly, the selectable marker gene phosphomannose isomerase (Pmi; Reed et al., 2001) was cloned under the control of the maize ubiquitin promoter into pAHC17 (S. Travella, unpublished). Prior to bombardment, the gene cassette (ubi:Pmi) was enzymatically released from the vector backbone using a HindIII and a NotI site, the latter being previously introduced 3′ of nos.

Biolistic transformation of the susceptible wheat Bobwhite SH 98 26

The hexaploid spring wheat cultivar Bobwhite SH 98 26, which was developed by CIMMYT in Mexico, was transformed via particle bombardment essentially as described (Pellegrineschi et al., 2002). In summary, 3000 immature wheat embryos were isolated from freshly harvested wheat seeds and co-transformed in seven separate experiments with the Pmi and the Pm3b gene cassettes by particle bombardment. Primary transformants (T0 plants) were regenerated in tissue culture and selected on mannose-containing media (Wright et al., 2001). Twenty-five of them carried at least one full-length copy of Pm3b as inferred from Southern blot analysis and were allowed to self.

Genotypic characterization of wheat plants transgenic for the Pm3b gene

Transgenic T0 and T1 plants were analyzed by DNA gel blots. Isolation of genomic DNA from leaves and Southern hybridization were performed as described (Stein et al., 2001; Travella et al., 2006). DNA was digested with restriction enzymes BamHI and SacI. Cleavage with BamHI produces a 4.5-kb fragment containing the complete coding region of Pm3b and SacI cuts once inside the gene cassette, allowing determination of insert copy number. Membranes were hybridized with probes R081 or SuB3 that bind to PM3B-encoding sequences. R081 and SuB3 were amplified by PCR from the Pm3b-containing plasmid used for transformation with the primer pair R081A (5′-TGGGCTCCAGAAACCAGA-3′) and R081B (5′-TTTGATGCTGCC-CAGTTG-3′) or Cint18F (5′-CCAATACTCAAGCTTAGCTACAA-3′) and Cint17R (5′-GTGGCAGAGGGAAGTCAT-3′), respectively. Absence of co-transformed contaminations of the gene cassettes with plasmid backbone containing beta-lactamase (bla) was confirmed in events Pm3b#1 to #4 by PCR using primers Amp-F1 (5′-TTTCCGTGTCGCCCTTATTC-3′) and Amp-R1 (5′-CAGTGAGGCACCTATCTCAG-3′) with the parameters: 34 cycles of 30 s at 94 °C, 30 s at 53 °C, 1 min at 72 °C, and by Southern blot analysis using the full-length bla gene as probe (816 bp).

For segregation analysis, presence of the transgene was determined by PCR with the primer pairs SuB13 (5′-TGCCTAGAAGATCTATGCTTATCAG-3′) and SuB8 (5′-CCGCTCACGGACTAGCCTC-3′) or sbi342 (5′-TGGGCAGCATCAAACGC-3′) and sbi143 (5′-CAAGACCGGCAACAGGATTC-3′). PCR using the primer pair SuB13/SuB8 was performed with the following parameters: 30 cycles of 30 s at 94 °C, 30 s at 65 °C and 50 s at 72 °C. The parameters for the primer pair sbi342/sbi143 were 30 cycles of 30 s at 94 °C, 30 s at 60 °C and 45 s at 72 °C.

RNA isolation and cDNA synthesis

For expression analysis of Pm3b in field-grown plants, a 4-cm leaf segment of the penultimate fully developed leaf from three plants per plot at stem elongation stage (BBCH 32–37) or of the flag leaf from three plants per plot at flowering stage or a few days later (BBCH 61–71) were pooled. For Pm3b#3, either leaf material from six plants per plot was pooled or each leaf piece was analysed individually. At each time point, leaf material was harvested from four (2008) and three (2009) plots per line. From greenhouse grown Pm3b#3 plants, 3-cm segments of the second and third leaf were pooled. Leaf samples were frozen immediately after harvesting and stored at −80 °C.

Leaf material was ground in custom-made polyethylene bags in the presence of 2 mL of RNA lysis buffer (4 m guanidine isothiocyanate, 10 mm Tris pH 7.5, 0.97%β-mercaptoethanol) using a hand homogenizer as described for DNA extraction (Stein et al., 2001). Per sample, 190 μL of the lysate was purified with the SV Total RNA Isolation System kit (Z3100; Promega, Dübendorf, Switzerland) including DNase I treatment. RNA quantity and purity were determined using a NanoDrop ND-1000 spectrophotometer (NanoDrop products, Wilmington, DE). Only RNA samples with A260/A280 ratios between 2.1 and 2.2 were further processed. RNA integrity was checked by electrophoretic separation on a 1% agarose gel in 1× TAE running buffer (40 mm Tris–acetate, 2 mm EDTA). First-strand cDNA was synthesized from 500 ng of total RNA using 40 units of Moloney Murine Leukemia Virus (M-MLV) Reverse Transcriptase RNAse H Point Mutant (9PIM368; Promega), 2.5 μm Oligo(dT)20, 0.5 mm dNTPs, 1× M-MLV-RT reaction buffer in a total volume of 12 μL, and incubation times and temperature were adopted from the manufacturer’s protocol. After reverse transcription, cDNA was treated with 2.5 units of RNAse H (M0297; New England Biolabs Inc., Ipswich, MA) at 37 °C for 20 min and inactivated at 65 °C for 20 min. To check for genomic DNA contamination, RT-minus controls were incorporated in the RT-qPCR experiments for the target gene Pm3b and the reference gene GAPDH, but not for Ta2776 and Ta54227 because the corresponding probes only hybridize on cDNA. All quantification cycle (Cq) values of the RT-minus controls were at least five cycles above the corresponding RT-plus, thus above the lower limit recommended by Bustin and Nolan (2004).

Pm3b expression analyses by RT-qPCR

Primers and probes were designed using Primer Express® Software v2.0 (Applied Biosystems, Rotkreuz, Switzerland) or the online tool GenScript Real-time PCR (TaqMan) Primer Design (https://www.genscript.com/ssl-bin/app/primer) or manually. Primer pairs were in silico validated for specificity using BLAST (http://blast.ncbi.nlm.nih.gov). The selected primer/probe pair for the target gene Pm3b was shown to be specific to all known Pm3 alleles. PCR products of the four primer pairs (Table 4) were cloned and sequenced, confirming that only the targeted sequences were amplified. Primers and probe concentrations were optimized as described in the TaqMan® Universal PCR Master Mix protocol (430449; Applied Biosystems) and are indicated in Table 4. Absence of primer dimers was confirmed by examination of dissociation curves with the Applied Biosystems 7500 Software (version 2.0.2.). PCR efficiency was determined on the sample Pm3b#1 using calibration curves based on six serial dilutions log10 for each amplicon. All primer/probe sets used for the reference genes had 97–102% PCR efficiencies. For the target gene Pm3b, efficiency was determined for each experiment and time point separately (macroplots 2008 – 93% (stem elongation) and 94% (flowering), microplots 2009 – 92% (stem elongation) and 90% (flowering), Pm3b#3 greenhouse 99% and Pm3b#3/Pm3b#1 field 102%).

Table 4.   Sequence and concentration of primers and probes used for RT-qPCR experiments
Target gene (UniGene)GenBank accession numberAmplicon lengthPrimer/probeSequence (5′–3′; modifications)Conc (nM)
  1. Primer/probe concentration in RT-qPCR reaction.

Ta.31015 (Pm3b)AY325736108 bpSbi342TGGGCAGCATCAAACGC600

As reference gene candidates, we used GAPDH (Travella et al., 2006), Ta.2776 and Ta.54227, which were shown to be stably expressed in wheat (Paolacci et al., 2009). Reference gene validation was performed with a subset of eight samples (from microplots 2009) using geNorm, a Visual Basic Application for Microsoft Excel (Vandesompele et al., 2002). As pairwise variation was below the cut-off value of 0.15 for all samples, no further candidate reference genes were tested. In the complete dataset, V2/3 were equal to 0.138 (macroplots field 2008, Figure 2a), 0.107 (microplots field 2009, Figure 2b and Figure S2), 0.198 (Pm3b#1 and Pm3b#3, field; Figure 3a) and 0.120 (Pm3b#3 greenhouse, Figure 3b). Data normalization was performed as described in Hellemans et al. (2007).

MGB hydrolysis probes (TaqMan®; Applied Biosystems) were used for Pm3b (Ta.31015), Ta.2776 and Ta.54227 (Table 4). For GAPDH (Ta.30768), CYBR® green (4309155; Applied Biosystems) was used. The RT-qPCR reaction was performed with 4 μL of eightfold diluted cDNA in a reaction volume of 16 μL containing 8 μL of PCR Master Mix (4309155 for CYBR® green, 4304437 for probes; Applied Biosystems) in optical 96-well plates (4346906; Applied Biosystems). RT-qPCR was carried out in an Applied Biosystems 7500 Real-Time PCR System using manufacturer’s default cycling conditions (50 °C for 20 s, 95 °C for 10 min followed by 40 cycles of 95 °C for 15 s and 60 °C for 1 min). In all experiments, at least three biological replicates of each sample were tested, and reactions were carried out with three (Pm3b) or two (reference genes) technical replicates. Technical replicates with standard deviation of Cq values above 0.5 were excluded from further analysis. As not all samples could be analysed on a single plate, samples were arranged in a randomized block design maximizing the target genes per plate. Variation between inter-run calibrators from the different plates was in the range of the technical replicates within the plates. Therefore, inter-run calibration was omitted.

Leaf segment infection tests with powdery mildew

For powdery mildew infection tests on leaf segments, plants were raised in a growth chamber at 20 °C/16 °C day/night temperature with 70% relative humidity for 10 days. Powdery mildew isolates 96236, 96224, 07230 or 07296, which are avirulent on the Pm3b donor line Chul and its backcross line Chul/8*Chancellor, were used for screening for resistant lines and for segregation analysis. Segments of the primary leaf were placed on 0.5% phytoagar (supplemented with 30 ppm benzimidazole) plates, infected with freshly propagated conidiospores (Winzeler et al., 1991), incubated at 20 °C with 16 h light per day and 80% relative humidity, and scored after 8 days as previously described (Kaur et al., 2008). The powdery mildew strains used in this study originate from the former collection of French isolates at INRA Rennes (France) or are Swiss isolates from the former isolate collection of ART Reckenholz (Switzerland) and our own collection (Brunner et al., 2010; Ph. Streckeisen, unpublished data).

Statistical analysis

We used the statistical software GenStat (VSN International Ltd., Hemel Hempstead, UK) to fit multiple linear regression models for all variables. Residual plots were examined to identify outliers and to check whether the assumptions of normality and homoscedasticity were fulfilled. Data were analysed untransformed except for the data of expression levels in 2009 which were log transformed to meet model assumptions. The critical significance level was 0.05 in all analyses.

Field testing

Field experiments were carried out at the Agroscope Reckenholz-Tänikon Research Station (ART) in Zürich-Reckenholz, Switzerland. The trials were sown on 30.3.2008, 19.3.2009 and on 25.3.2010. Per microplot (1.0 × 1.3 m), 200 (2008) or 300 (2009, 2010) viable seeds were sown in five rows using a Hege 90 Seedmatic seeding machine (Hege Maschinen, Eging am See, Germany). In macroplots (1.08 × 3 m in 2008 and 2009, 1.08 × 4 m in 2010), 400 viable seeds per m2 were sown in six rows using an Oyjord plot drill system (Wintersteiger AG, Ried, Austria). In February, the mineralized nitrogen in the top 100 cm of soil was 49.8 kg N/ha (2008), 47.6 kg N/ha (2009) and 41.7 kg N/ha (2010). Nitrogen fertilizer (30 kg N/ha) was applied at the first leaf stage (phenological stage BBCH 11; Lancashire et al., 1991) and at the flag leaf stage (BBCH 39) in 2008, before sowing and at tillering (BBCH 22–29) in 2009 and immediately after sowing and at the beginning of tillering (BBCH 14–22) in 2010. Conventional application of herbicide and insecticide was performed. Note that in 2008, no insecticide was applied in the macroplots to allow biosafety research projects observing insect food webs.

Macroplots were arranged in a randomized complete block design (2008) or in a latin square design with incomplete sub-squares (2009, 2010), respectively, with four replications. Microplots were arranged in a randomized split plot design with four replications. The lines were randomly allocated to plots within each block, with one plot of each line per block. The three treatments, natural powdery mildew infection (in two complete replicas in 2008), artificial powdery mildew infection and fungicide treatment were arranged in blocks in 2008 and randomized in 2009 and 2010. Treatments were separated with a 5.3 m (2008), a 3.96 m (2009) and a 2.64 m (2010) border crop of spring triticale cv. Trado. To increase the powdery mildew disease pressure, all tracks of experimental microplots were flanked on both sides at a distance of 0.25 m with a spreader row, consisting of a 9 : 1 mixture of the susceptible varieties FAL94632 and Kanzler. For the fungicide treatment, experimental plots and spreader rows were treated two (2008) and three (2009, 2010) times with Prosper (500g/L Spiroxamine; Leu + Gygax AG, Birmenstorf, Switzerland) between May and June. For artificial powdery mildew infection, seedlings of wheat cv. Kanzler were grown in Jiffy pots (Jiffy Products GmbH, Mölln, Germany) in the greenhouse and infected with powdery mildew isolate 96224 which is avirulent on Pm3b. These pots were planted into the spreader rows every 1.0 m (2008) and every 0.8 m (2009, 2010) in April. In June 2009, the planting of infected pots was repeated, as the warm and dry weather inhibited proper powdery mildew disease development and spreading.

Powdery mildew disease symptoms were scored on the basis of a 1–9 scale as described by Kmecl et al. (1995). Each year, plots were scored weekly from end of May to mid of June during four subsequent weeks and a last time at the end of June. AUDPC was calculated as described (Shaner and Finney, 1977; Jeger and Viljanen-Rollinson, 2001). Leaf chlorophyll content was estimated at the end of flowering (BBCH 69) in 2008 in fungicide-treated microplots using a portable chlorophyll meter (SPAD 502; Minolta, Osaka, Japan). In each of the four plots per line, ten flag leaves were measured and mean values were taken for data analysis. Leaf area was measured with a LI-3100 Area Meter (LI-COR, Lincoln, NE). Stomatal conductance of two sun-exposed green leaves (lower side) per plot was measured at the milk stage (BBCH 75) in eight plots per line in all three treatments using a SC-1 Porometer (Decagon Devices, Pullmann, NE). The measurements were taken in random order on two consecutive sunny days. Plant height was measured from the soil surface to the top of the ear excluding the awns. Seed set was determined on the main shoot of five plants per plot.

The whole experiment was surrounded by a 2.6-m guard strip of the tall spring triticale cv. Trado. The experimental plots were netted during germination and grain filling. On 13 June 2008, the field experiment was partially damaged by vandals by cutting off the plant heads with sickles. After this incident, all samplings, scorings and measurements were restricted to undamaged rows.

Consent to release these transgenic lines (application B07002) was obtained from the Swiss Federal Office for Environment under the Release Ordinance of 25 August 1999 (Nr. 814.911 in Federal Legislation) and the Gene Technology Act (Federal Law on Non-Human Gene Technology of 21 March 2003; Nr. 814.91 in Federal Legislation) in compliance with the draft of the revised Release Ordinance of 21 November 2005 and the EU Directive 2001/18/EC. Each line (transgenic event) was described in detail and was approved individually.


The national research station Agroscope Reckenholz-Tänikon ART is thanked for setting up the field experiment. We acknowledge Karen Hilzinger and Dr Christof Sautter for their contribution to the application for the field trials. Bea Senger is thanked for taking care of greenhouse plants, for field work and scorings. We are thankful to Theres Imhof-Klimm and Gabriele Büsing for excellent technical assistance and Dr Silvia Travella for technical advice and the plasmid construct containing ubi:Pmi. Carolin Luginbühl is acknowledged for the technical management of the field trials and the help with powdery mildew scoring. We are thankful to Carolina Diaz Quijano and Philipp Streckeisen for their help with powdery mildew scoring. We thank Dr A. Pellegrineschi (Centro Internacional de Mejoramiento de Maiz y Trigo, Mexico) for seeds of Bobwhite SH 98 26 and Dr P. Quail (University of California, Berkeley and U.S. Department of Agriculture Plant Gene Expression Center, Albany, California) for the plasmid pAHC17. Syngenta (Basel, Switzerland) is acknowledged for the Pmi gene. This project was financially supported by grants from the Swiss National Science Foundation (NFP59 405940-115598 and 31003A-127061/1) and is part of the wheat consortium, a subunit of the Swiss National Research Programme NRP 59 ‘Benefits and risks of the deliberate release of genetically modified plants’ (http://www.NRP59.ch).