Pathogen‐inducible Ta‐Lr34res expression in heterologous barley confers disease resistance without negative pleiotropic effects

Summary Plant diseases are a serious threat to crop production. The informed use of naturally occurring disease resistance in plant breeding can greatly contribute to sustainably reduce yield losses caused by plant pathogens. The Ta‐Lr34res gene encodes an ABC transporter protein and confers partial, durable, and broad spectrum resistance against several fungal pathogens in wheat. Transgenic barley lines expressing Ta‐Lr34res showed enhanced resistance against powdery mildew and leaf rust of barley. While Ta‐Lr34res is only active at adult stage in wheat, Ta‐Lr34res was found to be highly expressed already at the seedling stage in transgenic barley resulting in severe negative effects on growth. Here, we expressed Ta‐Lr34res under the control of the pathogen‐inducible Hv‐Ger4c promoter in barley. Sixteen independent barley transformants showed strong resistance against leaf rust and powdery mildew. Infection assays and growth parameter measurements were performed under standard glasshouse and near‐field conditions using a convertible glasshouse. Two Hv‐Ger4c::Ta‐Lr34res transgenic events were analysed in detail. Plants of one transformation event had similar grain production compared to wild‐type under glasshouse and near‐field conditions. Our results showed that negative effects caused by constitutive high expression of Ta‐Lr34res driven by the endogenous wheat promoter in barley can be eliminated by inducible expression without compromising disease resistance. These data demonstrate that Ta‐Lr34res is agronomically useful in barley. We conclude that the generation of a large number of transformants in different barley cultivars followed by early field testing will allow identifying barley lines suitable for breeding.


Introduction
To ensure global food security is one of the top challenges in this century. The ever-growing world population is the main driver for increasing demand for agricultural products until 2050 (Gerland et al., 2014). Loss of agricultural land and climate change are additional factors that require a higher productivity (Godfray et al., 2010). Plant pathogens are responsible for worldwide crop losses of 10%-16% on average (Chakraborty and Newton, 2011;Oerke, 2006). A major strategy to improve the efficiency of crop production is to enhance plant resistance against diseases by taking advantage of the large diversity of naturally existing resistance genes. To do so, identified resistance genes can be transferred to other crop cultivars by classical breeding or introduced into other plant species by stable genetic transformation. Most resistance genes are specific to one pathosystem, but in a few cases resistance genes were shown to be functional in several plant-pathogen interactions. Maize Rxo1 is an example of a resistance gene that was functionally transferred to the heterologous grass species rice where it mediates resistance against Xanthomonas oryzae pv. oryzicola (Zhao et al., 2005).
Pto, a resistance gene originating from tomato and conferring resistance to Pseudomonas syringae pv. tomato, was functional against strains of P.s. pv. tabaci expressing the effector avrPto after stable transformation into Nicotiana benthamiana (Rommens et al., 1995).
Ta-Lr34 has been successfully used in wheat resistance breeding since the beginning of the last century (Kolmer et al., 2008). Ta-Lr34 has two predominant alleles in the wheat gene pool, referred to as Ta-Lr34res (resistant) and Ta-Lr34sus (susceptible) (Dakouri et al., 2010;Lagudah, 2011). Ta-LR34res and Ta-LR34sus differ only by two amino acid polymorphisms, and Ta-Lr34res evolved from the ancestral Ta-Lr34sus version after domestication . Orthologues of Ta-Lr34res were detected in rice and sorghum, but not in maize and barley. Based on the two critical amino acid polymorphisms of Ta-LR34res in wheat, all of these orthologues were identified as susceptible haplotypes (Krattinger et al., , 2011. Ta-Lr34res was transformed to rice , barley (Risk et al., 2013), and maize . In all three grass species, Ta-LR34res was found to be functional: In rice it mediated resistance against rice blast (Magnaporthe oryzae), accompanied by a late LTN phenotype . In barley, Ta-Lr34res, but not Ta-Lr34sus, provided strong resistance against barley leaf rust (Puccinia hordei) and barley powdery mildew (Blumeria graminis f. sp hordei, Bgh). In contrast to wheat, Ta-Lr34res under control of its native promoter was highly expressed already at seedling stage in barley, resulting in an early strong LTN phenotype. Thus, Ta-Lr34res activity is accompanied by drastic pleiotropic effects with high fitness costs in barley.
Reducing or eliminating negative effects of a transgene can be achieved by using a different genetic background or by altering the expression level. As the negative effects of Ta-Lr34res were described to be expression level-dependent (Chauhan et al., 2015), we tested whether altered Ta-Lr34res expression can result in barley plants with good levels of disease resistance but no negative effects on growth vigour. We hypothesize that to minimize negative effects on barley growth, Ta-Lr34res should only be active when resistance is needed, i.e. at the early stage of infection. In this case, the promoter of choice would be pathogen-inducible. Pathogen-inducible promoters are activated early after infection (Hernandez-Garcia and Finer, 2014). A welldescribed pathogen-induced barley (Hordeum vulgare) gene is Hv-Ger4c, which belongs to a cluster of nine genes on barley chromosome 4H . Himmelbach et al. (2010) fused the Hv-Ger4c promoter to GUS and performed transient expression assays as well as experiments using stably transformed barley plants. Plants inoculated with Bgh and Puccinia hordei as well as the non-host pathogens Bgt and Puccinia triticina showed increased expression levels compared to noninfected controls, regardless if the interaction was compatible or not. Expression was measured in epidermal peels as well as whole leaf tissue. The results revealed an epidermis-specific expression pattern of the Hv-Ger4c promoter. Further, Himmelbach et al. (2010) reported that Hv-Ger4c is not activated upon abiotic stress treatment such as ozone, cold, wounding, UV irradiation or wind. This made the Hv-Ger4c promoter an excellent candidate for our study. Here, we provide proof of concept that induced, epidermis-specific expression of Ta-Lr34res in barley results in broad spectrum resistance without compromising fitness and this can provide much needed inbuilt resistance against fungal pathogens.

Characterization of the barley lines transgenic for
Hv-Ger4c::Ta-Lr34res We stably transformed barley cv. Golden Promise with Ta-Lr34res under the control of the Hv-Ger4c promoter and characterized the resulting transgenic lines (later called Hv-Ger4c::Ta-Lr34res lines) for transgene expression, disease resistance, and growth parameters. Nineteen primary transformants (T0) were obtained and grown to maturity. Eighteen T0 plants produced grains and T1 plants were genotyped by PCR. Plants were analysed regarding the development of LTN, resistance to the barley powdery mildew isolate K1 and the expression of the full-length Ta-Lr34res coding sequence (cDNA) ( Table S1).
To identify potential transformants with only slightly reduced or unaffected fitness but good disease resistance, 2-week-old T1 plants were analysed in a first step for the presence/absence of LTN. In addition, resistance to powdery mildew was analysed on 3-week-old plants 7 days after infection (dpi). Progeny of sixteen events showed either strong LTN at seedling stage (six T1 progeny, Table S1) similar to line BG 9 with the native Ta-Lr34res promoter (Chauhan et al., 2015;Risk et al., 2013) or were susceptible (eight T1 progeny) and were therefore not of interest for further investigation. T1 progeny of events 8 and 11 showed no or reduced LTN compared to line BG 9. Both lines showed resistance to powdery mildew 7 dpi at a comparable level as line BG 9. For further analyses, progeny of events 8 and 11 was selected for uniform resistance to powdery mildew in T2. To do so, at least 24 T2 seedlings of different individual T1 plants were grown to 2-week-old plantlets and infected with barley powdery mildew. T2 families showing uniform resistance and no segregation pattern in the genotype based on PCR were considered as uniform lines. Southern blot analysis (based on a probe derived from the marker gene used in transformation) of these T2 lines, uniform for resistance and LTN phenotype, detected three nonsegregating insertions and one segregating insertion in line 8 and two insertions of which one segregated in line 11 ( Figure S1). In both lines 8 and 11, the segregating insertion had no influence on the resistance phenotype, indicating that uniform lines are uniform for the functional insertion(s) represented by the nonsegregating bands in the Southern blot ( Figure S1). Therefore, seeds of uniform lines were pooled within transformation events and used for further investigation. Similarly, the corresponding azygous sister lines (null-segregants) were determined.
Hv-Ger4c::Ta-Lr34res lines showed enhanced resistance to barley leaf rust and powdery mildew To further characterize pathogen resistance, fourth leaves of T2 plants of the lines 8, 11, BG 9 and their corresponding azygous sister lines were taken for macroscopic and biochemical analyses of Ta-Lr34res-mediated resistance. We performed infection assays with the barley leaf rust isolate BRG1.2.1 and the barley powdery mildew isolate K1 as described in Risk et al. (2013). Line BG 9 developed no visual symptoms upon infection with barley leaf rust, while the corresponding azygous sister line BG 9 sib was fully covered by uredia (Figure 1a). In line 8, a much lower amount of uredia developed compared to the azygous sister line 8 sib. In addition, uredia were smaller and surrounded by chlorotic flecks (Figure 1a), mimicking the partial resistance in wheat. In line 11, mainly chlorotic spots and a few very small uredia were observed, whereas the corresponding azygous sister line 11 sib was fully covered by uredia, indicating full susceptibility ( Figure 1a). In the infection assay using barley powdery mildew, line BG 9 showed only a few chlorotic spots and both lines 8 and 11 developed only a few mildew pustules, while all azygous sister lines were completely covered by the fungus (Figure 1a). To assess whether resistance of the barley lines transgenic for Ta-Lr34res under the control of the Hv-Ger4c promoter might be of interest for agricultural application, an infection assay was performed under near-field conditions. To do so, the same lines tested in the glasshouse were grown in a convertible glasshouse (Romeis et al., 2007) and artificially infected with barley powdery mildew. Under these near-field conditions, similar macroscopical results were observed regarding resistance to powdery mildew as recorded in the glasshouse, indicating the functionality of the Hv-Ger4c::Ta-Lr34res-mediated resistance under near-field conditions ( Figure 1a). To quantify Ta-Lr34res-mediated resistance, the amount of the fungal cell wall component chitin was measured using the biochemical WGA-FITC method (Ayliffe et al., 2014. The fourth leaves of plants at the five-leaf stage were taken in the glasshouse experiment, whereas for the experiment under near-field conditions fourth leaves at the six-leaf stage were used for chitin measurements. In the infection assay in the glasshouse using barley leaf rust, a Kruskal-Wallis test analysis indicates that overall differences in the level of chitin exist among lines (X 2 (3, N = 42) = 31.2, P < 0.001). Post hoc comparison tests showed that there was a significantly reduced chitin amount in all Hv-Ger4c::Ta-Lr34res lines compared to the pooled azygous sister lines (all pair-wise P-values between transgenic lines and the azygous sister lines <0.008, Figure 1b). Further, chitin levels of lines 8 and 11 were not different from the one in line BG 9 (pair-wise P-values >0.5, Figure 1b). After infection with barley powdery mildew in the glasshouse, a Kruskal-Wallis test analysis revealed overall differences among lines (X 2 (3, N = 42) = 29.8, P < 0.0001). Here again, post hoc pair-wise comparison tests revealed that chitin amounts were clearly reduced compared to the pooled azygous sister lines (all pair-wise P-values <0.02) and chitin amounts of lines 8 and 11 were similar to the one in BG 9 (all pair-wise P-values >0.5).
Under near-field conditions (Kruskal-Wallis test, (X 2 (3, N = 42) = 26.5, P < 0.0001), all Hv-Ger4c::Ta-Lr34res lines showed a significant reduction in chitin amounts compared to the azygous sister lines (all pair-wise P-values <0.02) and reached similar chitin levels as the highly expressing line BG 9 (all pair-wise P-values >0.9, Figure 1b). Taken together, these results show that all Hv-Ger4c::Ta-Lr34res lines reach similar resistance levels as the highly expressing line BG 9 where the Ta-Lr34res gene is under control of its own promoter.
Hv-Ger4c::Ta-Lr34res lines show reduced LTN and reduced impact on growth parameters To investigate whether the inducible Hv-Ger4c::Ta-Lr34res lines show less negative effects on growth parameters compared to BG 9, seedlings were investigated for the presence and intensity of LTN. Whereas line 8 showed no LTN in the first leaf at the five-leaf stage, line 11 showed LTN, but at a reduced level compared to BG 9 under glasshouse conditions ( Figure 2a). The reduced LTN level at the five-leaf stage might indicate an overall reduction in fitness costs. To determine possible negative effects of Hv-Ger4c:: Ta-Lr34res at later growth stages, plants were grown until maturity and growth parameters were measured. Growth parameters included gram of grains per plant, dry weight, number of spikes per plant, number of grains per spike, grains per plant and the 1000-grain weight. To test the influence of environmental conditions on growth parameters, data were collected from plants grown in the glasshouse and under nearfield conditions, respectively (Figure 2 and Figure S2).
Under glasshouse conditions, lines 8 and 11 reached similar values in gram of grains per plant as their corresponding azygous sister lines (ANOVA (F(5,54) = 41.6, P-value <2e À16 ), all pair-wise P-values >0.2, Figure 2b). Furthermore, for both lines similar values were measured compared to their azygous sister lines for all growth parameters tested ( Figure S2 and Table S2). Lines 8 and 11 showed considerably higher values than BG 9 in all measurements, indicating an improvement of growth vigour by the use of the Hv-Ger4c promoter to control Ta-Lr34res expression (Figure 2b and Figure S2).
We also wanted to test whether altered expression of Ta-Lr34res by the Hv-Ger4c promoter would lead to improved growth vigour under near-field conditions. All Hv-Ger4c::Ta-Lr34res lines showed no LTN phenotype compared to BG 9 at the seedling stage (Figure 2a). Line 8 performed the same as the azygous sister line for all growth parameters, indicating that Ta-Lr34res was not linked to negative effects in this particular line (Figure 2b, Figure S2 and Table S2). At later stages, the negative effects on growth developed continuously in lines BG 9, 11 and 11 sib, resulting in lower values for gram grains (Figure 2), number of spikes per plant and grains per plant ( Figure S2). Lines BG 9, 11 and 11 sib produced less dry weight biomass than line 8, line 8 sib and line BG 9 sib. This resulted in a lower number of spikes per plant and consequently a lower amount of grains per plant ( Figure S2 and Table S2). In addition, lines BG 9, 11 and 11 sib lodged at the flowering stage. They produced grains on ground-lying tillers, which lead to 1000-grain weights which were mostly not different compared to their corresponding azygous sister lines ( Figure S2 and Table S2). Because of the broken tillers, lines BG 9, 11 and 11 sib turned out not to have appropriate growth vigour under near-field conditions. In most cases, growth values of line 11 did not differ from line 11 sib. This indicates that the negative effects observed for this line under near-field conditions must have been caused by somaclonal or epigenetic variation due to the tissue culture process rather than by the presence of Ta-Lr34res.
Induction of Ta-Lr34res expression by the Hv-Ger4c promoter ANOVA analysis of expression data revealed significant differences of Ta-Lr34res expression among the tested lines (rust glasshouse: (F(5,32) = 47.9, P-value <0.0001, Figure 3a), mildew glasshouse: (F(4,28) = 93.2, P-value <0.0001, Figure 3b) nearfield: (F(2,18) = 42.6, P-value <0.0001, Figure S4). Expression analysis revealed that the Ta-Lr34res expression levels of the uninfected Hv-Ger4c::Ta-Lr34res lines were 12-to 39-fold lower compared to the uninfected line BG 9 carrying the Ta-Lr34res promoter (Figure 3) under glasshouse conditions. A low level of basal expression in uninfected plants of line 8 was detectable by RT-qPCR in the rust assay but not in the mildew experiment. This is possibly due to the different plant growth conditions used in the mildew and rust experiments. Expression levels under nearfield conditions of uninfected Hv-Ger4c::Ta-Lr34res plants were 5-to 11-fold lower compared to BG 9 ( Figure S4).
To assess whether Ta-Lr34res expression was pathogeninducible after fusion to the Hv-Ger4c promoter, infection assays with powdery mildew and barley leaf rust were performed. Fourteen-day-old seedlings were inoculated with barley powdery mildew and barley leaf rust, respectively. After 7 dpi (mildew) and 10 dpi (rust), third leaves were harvested for expression analysis. Infected leaves were compared to mock-infected leaves grown at the same time and under the same conditions. Interestingly, no induction upon rust infection was detected (all pair-wise comparison between mock-infected and infected lines P-values >0.7, Figure 3a). In contrast, Ta-Lr34res expression was induced upon infection with powdery mildew in all Hv-Ger4c::Ta-Lr34res lines (line 11: pair-wise P-value <0.0001). Expression in line 11 was fourfold higher upon infection. In line 8, the expression level upon mock infection was below detection level and therefore no statistical analysis was possible. Yet, expression of Ta-Lr34res was detectable in infected line 8 and reached similar expression levels as infected line 11 (pair-wise comparison P-value = 0.86), indicating that Ta-Lr34res was also induced upon infection in line 8 (Figure 3b).

Discussion
Enhancing resistance by alteration of expression levels is frequently accompanied by increased fitness costs. The overexpression of tomato Pto (Tang et al., 1999) or of the NLR gene BAL in Arabidopsis (Stokes et al., 2002) led to microscopic lesions or dwarfing and twisted leaves, respectively. Furthermore, wheat lines over-expressing the NLR gene Lr10 showed reduced grain weight compared to the Lr10 donor line under near-field conditions (Feuillet et al., 2003;Romeis et al., 2007). Interestingly, different pleiotropic effects caused by the same transgene have been observed among different transformation events. In one of four transgenic lines over-expressing the wheat powdery mildew disease resistance gene Pm3b, no pleiotropic effects were observed under field conditions. In the three lines Pm3b#2-4, plants showed different level of chlorosis and reduced fertility, whereas this phenomenon was not observed in line Pm3b#1 (Brunner et al., 2011). The different phenotypes of the two genotypes Pm3b#2 and Pm3b#1 could be explained by higher Pm3b expression levels in Pm3b#2 compared to Pm3b#1. However, Pm3b#1 and Pm3b#4 showed similar expression levels of Pm3b transcripts, but no pleiotropic effects were detected in Pm3b#1, indicating that not only expression levels but also position effects with subtle changes in expression over time and space might influence the phenotype (Brunner et al., 2011).
In the present study, we found that one of two lines transgenic for Hv-Ger4c::Ta-Lr34res analysed in detail (line 8) showed enhanced disease resistance without negative pleiotropic effects under glasshouse and near-field conditions. This demonstrates that the Ta-Lr34res gene can be used as a resistance source in barley, very similarly to wheat.
The only pathogen-inducible barley promoter with demonstrated functionality in driving conditional expression of a transgene is Hv-Ger4c (Himmelbach et al., 2010). Although we found this promoter to be highly useful in our work, it might be worth exploring additional inducible promoters for fine-tuning expression levels. It has been shown that pathogen-and defence hormone-inducible promoters can be functional in heterologous species, as shown for Os-PR10a (induced by Xanthomonas oryzae pv oryzae and salicylic acid, SA), which was activated by SA in stably transformed Arabidopsis (Hwang et al., 2008). Together with the rapid accumulation of RNAseq data that allow a fast identification of pathogen-inducible genes, there are an increasing number of candidate promoters that can be used for analysis of inducible expression.
In this study, two lines expressing Ta-Lr34res under control of the Hv-Ger4c promoter were selected for detailed analyses. There, we detected some basal expression of Ta-Lr34res in the Hv-Ger4c::Ta-Lr34res lines and Ta-Lr34res was specifically induced by the Hv-Ger4c promoter after infection with powdery mildew, but not with barley leaf rust. Similarly, Himmelbach et al.
(2010) measured a lower induction of Hv-Ger4c promoter-driven GUS expression after infection with barley leaf rust compared to powdery mildew. This could be due to the fact that barley powdery mildew infects epidermal cells, whereas barley leaf rust grows mainly in the mesophyll. Possibly, this indicates differences in inducibility of the promoter by pathogens in different tissues of barley. The fact that Hv-Ger4c::Ta-Lr34res lines showed resistance against barley leaf rust although an induction of Ta-Lr34res expression was not confirmed indicates that the detected basal expression of Ta-Lr34res was sufficient to mediate disease resistance.
The combination of different resistance genes is generally considered to be a successful strategy to enhance resistance. In barley, a considerable number of resistance genes have been described (Andersen et al., 2016;Seeholzer et al., 2010). The most well-known and cloned major barley resistance genes that act efficiently as single genes are Rpg1 and the Mla alleles (Brueggeman et al., 2002;Chelkowski et al., 2003). It has been shown that the use of quantitatively acting resistance genes can extend the life expectancy of R genes and the combination of these two types of resistance is a very promising approach to achieve durable resistance in agricultural ecosystems (Brun et al., 2010;McDonald, 2010). Thus, a combination of, e.g. Mla genes with the quantitatively acting Ta-Lr34res might be an effective strategy to achieve durable resistance against mildew in barley. Another quantitatively acting gene in barley mildew resistance is the modulator of defence and cell death Mlo gene, which encodes a seven-transmembrane domain protein and is a negative regulator of pathogen defence response. Homozygous loss-of-function mutant (mlo) barley plants show enhanced and durable resistance against barley powdery mildew (Jorgensen, 1992). So far, several mutant alleles (mlo) were described, which resulted in inhibited fungal growth (Buschges et al., 1997;Piffanelli et al., 2002).
The Rpg1 gene encodes a protein with two tandem kinase domains and has been used in the northern parts of the USA and Canada to protect barley from stem rust for more than 65 years (Brueggeman et al., 2002). Therefore, Rpg1 confers durable resistance, but in contrast to Ta-Lr34res this is a resistance active specifically against stem rust. It would be interesting to cross the Hv-Ger4c::Ta-Lr34res lines with Morex, the donor of Rpg1, to test whether the Rpg1/Ta-Lr34res gene combination would result in still improved resistance compared to the corresponding parental lines. Pyramiding different resistance genes or alleles seems promising to enlarge spectrum and durability of pathogen resistance. However, stacking of certain different Pm3 alleles resulted in suppression of the Pm3-mediated resistance response, indicating that combining of resistance genes and/or alleles is not always useful (Stirnweis et al., 2014).
Earlier studies demonstrated that Ta-Lr34res is transferrable to different crop species and was shown to confer durable, partial resistance also in these heterologous species (Chauhan et al., 2015;Rinaldo et al., 2016;Risk et al., 2013;Sucher et al., 2016). Interestingly, the expression of Ta-Lr34res was accompanied by negative pleiotropic effects, leading to increased fitness costs in barley and certain rice lines but not in durum wheat and maize. The data of this study and the results presented for barley by Chauhan et al. (2015) show a correlation of Ta-Lr34res expression and the level of disease resistance as well as negative pleiotropic effects in barley. Similar results have been reported in rice . In all rice lines transgenic for Ta-Lr34res, expression of the gene was controlled by its native promoter, and  identified one lowexpressing transgenic rice line with no negative pleiotropic effects. This indicates that the genetic backgrounds of rice, maize and durum are all suitable to express Ta-Lr34res under the control of its own promoter to give good resistance without fitness costs, at least in some carefully selected genotypes. In contrast, in barley no Ta-Lr34res transgenic line under control of the native wheat promoter showed expression levels of Ta-Lr34res which were low enough to avoid negative pleiotropic effects. The genetic background of barley seems not to be suitable to allow optimal Ta-Lr34res expression levels when controlled by the native wheat promoter. Our results show that the negative pleiotropic effects of Ta-Lr34res in barley can be overcome by induced expression of Ta-Lr34res using the Hv-Ger4c promoter. Therefore, barley plants transgenic for Ta-Lr34res are potentially useful for agricultural applications. It seems rather straightforward to generate a large number of Hv-Ger4c::Ta-Lr34res lines or cross line 8 to elite barley cultivars and to select from them the lines with good resistance levels but no pleiotropic effects as determined under field conditions. Such lines would then represent ideal parents in breeding programmes.

Experimental procedures
Cloning of Hv-Ger4c promoter upstream of genomic fragments of Ta-Lr34res and stable transformation in barley The Gateway binary vector p6UGLPΔSwaI containing the Hv-Ger4c promoter was used as a source for the promoter sequence (Himmelbach et al., 2010). The native promoter of Ta-Lr34res in constructs p6U:Lr34-ResGenomic (Risk et al., 2013) was replaced by a two way cloning strategy. First the full-length cDNA of Ta-Lr34res was cloned in the p6UGLPΔSwaI by Gateway cloning method resulting in the p6UGLPΔSwaI:Lr34rescDNA plasmid. There is a unique NruI restriction site within the very first exon of Ta-Lr34res, and a MluI site is present just upstream of Ta-Lr34res native promoter in p6U:Lr34-ResGenomic. The fragment containing Hv-Ger4c promoter and part of Ta-Lr34res cDNA till the NruI site was amplified with primers MluI-proGLP-Fwd (5 0 -ACGCGTCTGCAGGAATTCGATCAGC-3 0 ) and NruI-Lr34-Rev (5 0 -CGCTGTCGCGATGCCAATTCTAACTCGG-3 0 ) and cloned in PCR cloning vector pSCB (Agilent Technologies, Santa Clara (CA), USA) resulting in the pSCB:MluIProGLP-Lr34NruI plasmid. After confirmation by sequencing, the promoter Hv-Ger4c flanked by the MluI site and Ta-Lr34res NruI fragment was replaced in the original p6U:Lr34-ResGenomic with MluI and NruI restriction sites thereby exchanging the native Ta-Lr34res promoter with pathogen-inducible Hv-Ger4c promoter.
Stable transformation of barley cv. Golden Promise was performed using Agrobacterium-mediated stable transformation according (Hensel et al., 2009) with the exception that the callusforming media contained 1.25 mg/L CuSO 4 pentahydrate and the A. tumefaciens strain AGL-1 was used.
Selection of transformants uniform for resistance to powdery mildew T0 plants were grown to adult stage, and T1 grains were harvested. Two-week-old T1 seedlings were infected with powdery mildew isolate K1 and analysed macroscopically for Ta-Lr34res-mediated resistance (7 dpi). To do so, the barley line BG 9, transgenic for Ta-Lr34res under the native promoter and its corresponding sister line BG 9 sib (described earlier, (Risk et al., 2013)) were used as positive and negative control, respectively. Resistant T1 progeny was then checked for presence of Hv-Ger4c::Ta-Lr34res on genomic DNA level by PCR and analysed for full-length cDNA expression (Table S1). Finally, copy number was determined by Southern blot analysis.

RNA extraction
Total RNA was extracted from third leaves using SV Total RNA Isolation system (Promega, Madison (WI), USA), and RNA integrity was checked by electrophoresis on a 1.2% agarose gel in 1x sodium-borate buffer. For cDNA synthesis, RNA concentration was measured with a Nanodrop ND-1000 spectrophotometer and 500 ng of total RNA was used for cDNA synthesis with the iScriptTM advanced cDNA synthesis kit (Bio Rad, Hercules (CA), USA).

Detection of transgenic barley and determination of transgene copy number and full-length cDNA amplification
Total genomic DNA was extracted from leaves using the CTAB method (Stein et al., 2001), and presence of Ta-Lr34res was assessed using the marker csffr1 . Southern blot was performed by digesting 10 lg of genomic DNA with EcoRI and using a 32 P-labelled probe covering the HPT gene of the p6U vector as described previously (Risk et al., 2013). Segregants of the T1 generation negative in PCR were propagated to T2 and checked by Southern blot analysis. If absence of Ta-Lr34res was confirmed, they were considered to be azygous sister lines (sib).

Growth conditions and infection assays
Grains were sown in 96-well Jiffy pots filled with soil (Einheitserde Profi Substrat, Einheitserde Werkverband e.v., Germany) and put over night at 4°C in the dark for synchronized germination. Afterwards, plants were transferred to a growth chamber (16 h, 20°C, light and 8 h, 17°C, darkness) until indicated time points of the respective experiments. Seedling propagation was performed by repotting plants to 2 L pots (one plant per pot) and growing them in glasshouse at standard growth conditions. The convertible glasshouse system was described in (Romeis et al., 2007). Its roof is automatically opened during dry and windless weather conditions. In addition, one side is permanently open which allows outdoor temperatures ( Figure S5). Twenty plants each of lines BG 9, BG 9 sib, lines 8, 8 sib, 11 and 11 sib were sown on 8th of March 2016 in central cylinders (42.2 L) within big plastic containers filled with fresh farm soil which was taken freshly from a neighbouring field. Plants were grown until the five-leaf stage and reduced to ten plants per central cylinder.
Five cylinders were used per line. The surrounding area was planted with Golden Promise (GP) as buffering plants. Plastic containers were arranged in two blocks, and within each block each line was assigned randomly to a cylinder. The growth parameter measurements under glasshouse conditions were performed in 15 L pots instead of cylinders in a similar manner with the exception that due to a lack of space no plastic containers with surrounding buffering plants were grown.
Barley leaf rust infections were performed as described in Risk et al. (2012)

Chitin measurement
Chitin measurement was performed as described in Ayliffe et al. (2013Ayliffe et al. ( , 2014 with the following exceptions. Samples were diluted 1 : 10 for measurement; samples were treated for 20 min a steam cooker instead of autoclaving; the fluorometer Synergy H1 Hybrid 311 Reader (BioTek Instruments GmbH, Switzerland) was used for measuring. Four technical replicates were used for each sample. The readouts were analysed using the software Gen5, version 2.03.1 (BioTek Instruments GmbH, Switzerland). Chitin amounts were calculated using standard curves ( Figure S3) generated by defined amounts of Chitin from shrimp shells (Sigma-Aldrich, St. Louis (MO), USA).

Statistical analysis
Differences in lg chitin per mg fresh weight and differences growth parameter values per plant across lines were tested using an ANOVA. If necessary, data were square-root-transformed to ensure normal distribution of residuals. When the normal distribution could not be reached, Kruskal-Wallis test was used. Multiple comparison P-values were then computed using the post hoc Tukey-Kramer HSD or Kruskal-Nemenyi tests. Expression values were log 10 -transformed to correct the exponential character of RT-qPCR. Differences in expression were tested with an ANOVA, following the same procedure as described above. All statistical analyses were performed in R v.3.2.1. Critical P-values were used for the decision for significance to create the letter codes. Critical P-values, P-values for all graphs shown and the transformation of raw data are given in Table S2.

Supporting information
Additional Supporting Information may be found online in the supporting information tab for this article: Figure S1. Southern blot of Hv-Ger4c::Ta-Lr34res barley lines. Figure S2. Additional growth parameters to Figure 2. Figure S3. Standard curves for chitin measurements shown in Figure 1. Figure S4. Normalized relative expression of Ta-Lr34res in plants grown in the convertible glasshouse under near-field conditions. Figure S5. The convertible glasshouse enables near-field growth conditions. Table S1. Phenotypical analysis of T1 progeny from different transformation events. Table S2. P-values, transformation of raw data and critical P-values for statistical analysis.