The wheat Lr34 multipathogen resistance gene confers resistance to anthracnose and rust in sorghum

Summary The ability of the wheat Lr34 multipathogen resistance gene (Lr34res) to function across a wide taxonomic boundary was investigated in transgenic Sorghum bicolor. Increased resistance to sorghum rust and anthracnose disease symptoms following infection with the biotrophic pathogen Puccinia purpurea and the hemibiotroph Colletotrichum sublineolum, respectively, occurred in transgenic plants expressing the Lr34res ABC transporter. Transgenic sorghum lines that highly expressed the wheat Lr34res gene exhibited immunity to sorghum rust compared to the low‐expressing single copy Lr34res genotype that conferred partial resistance. Pathogen‐induced pigmentation mediated by flavonoid phytoalexins was evident on transgenic sorghum leaves following P. purpurea infection within 24–72 h, which paralleled Lr34res gene expression. Elevated expression of flavone synthase II, flavanone 4‐reductase and dihydroflavonol reductase genes which control the biosynthesis of flavonoid phytoalexins characterized the highly expressing Lr34res transgenic lines 24‐h post‐inoculation with P. purpurea. Metabolite analysis of mesocotyls infected with C. sublineolum showed increased levels of 3‐deoxyanthocyanidin metabolites were associated with Lr34res expression, concomitant with reduced symptoms of anthracnose.


Introduction
Sorghum (Sorghum bicolor) is ranked as the fifth most commonly cultivated cereal in the world (FAOSTAT, 2016). Some of its useful attributes are tolerance to dry environments, high sugar content, high yields of forage biomass per unit of cultivated area and as a rich source of distinct phytochemicals such as dhurrin, sorgoleone and 3-deoxyanthocyanidins. While sorghum provides a useful resource for industrial purposes, for example the generation of ethanol, fibre and paper, its primary use is still for feed and food especially in the semi-arid tropics. Protecting yield losses from diseases such as anthracnose (Colletotrichum sublineolum) and rust (Puccinia purpurea), which can be variable in different agro-ecological regions continues to be a goal of sorghum improvement. Grain size and yield losses over 50% have been reported with anthracnose epidemics (Thakur and Mathur, 2000). Rust is particularly problematic in late-sown crops (White et al., 2012) with yield losses up to 65% (Bandyopadhyay, 2000).
Genetic solutions to protect crop plants against pathogens are often preferable to the use of agrochemicals. Numerous studies on plant-microbe interaction have led to an increased understanding of the molecular basis of the plant defense system, depicted by multiple layers of the plants ability to resist pathogen proliferation (Dangl et al., 2013). While most resistance genes tend to be short-lived, certain forms of plant defense genes provide more durable resistance. Studies in wheat with defined races of Puccinia (rust) and Blumeria (mildew) pathogen species have resulted in over 220 catalogued resistance genes, most of which individually provide resistance to a few races of a specific pathogen (McIntosh et al., 2013). However, a small number (e.g. Lr34, Lr46 and Lr67) have been identified that confer adult plant, broad spectrum partial resistance to multiple pathogen species. Most notable among the latter class of resistance genes is the Lr34 multipathogen resistance gene (Dyck and Samborski, 1979;McIntosh, 1992;Singh, 1992;Spielmeyer et al., 2005), which has been successfully deployed in wheat cultivation and provided durable field resistance to rust pathogens for over 100 years (Kolmer et al., 2008). Significantly, the multipathogen resistance conferred by Lr34 was not due to a cluster of resistance genes, but rather by a single gene encoding an ABC transporter Risk et al., 2012). Lr34 also differs from the other cloned multipathogen resistance gene Lr67, which encodes a sugar transporter from the STP13 lineage of monosaccharide transporters (Moore et al., 2015). The Lr34 resistance allele (Lr34res) differs from the susceptible or wild-type allele, Lr34sus, by changes to two amino acids; loss of a phenylalanine residue (F546) and a tyrosine to histidine substitution (Y634H) located in two separate, predicted transmembrane helices. The mechanism of disease resistance conferred by Lr34res remains unknown as does the substrate(s) it may translocate. Nonetheless, transfer of Lr34res as a transgene to other crops such as barley and rice has demonstrated its capability in conferring resistance against other pathogens that are unadapted to wheat Risk et al., 2013).
While Lr34res in wheat confers partial adult plant resistance to predominantly biotrophic pathogens, the observation of Lr34res efficacy in transgenic rice against the hemibiotroph, Magnaporthe grisea (causal agent of rice blast; Krattinger et al., 2016), led us in this study to examine the effect of Lr34res against the hemibiotroph pathosystem of anthracnose in sorghum. Secondly, we also note that while Lr34res effectiveness has been demonstrated against Puccinia species (P. hordeii, P. striiformis, P. graminis, and P. triticina) adapted to cool season crops, it remains unknown as to its function against Puccinia species that are virulent on warm season crops such as P. purpurea in sorghum. Thirdly, we take advantage of the pathogen-inducible visible pigmentation changes resulting from the synthesis of a unique class of flavonoid phytoalexins in sorghum, in furthering our understanding of potential signalling pathways triggered by Lr34res.

Results
Transgenic Lr34res expression confers resistance to sorghum rust (Puccinia purpurea) infection We introduced the complete wheat genomic sequence of Lr34res, encompassing 2.4 kb of native promoter and 1.5 kb native terminator sequence, by stable transformation in the genetic background of sorghum cultivar (cv.) Tx430. Four independent T0 transformants with the full-length Lr34res were obtained, of which three independent genotypes were fertile (Lr34-2, Lr34-5 and Lr34-6). Subsequent genomic and phenotypic analyses at T1-T3 generations were carried out on these three independent transgenic lines. Genomic blot analysis showed that line Lr34-2 carried a single copy of the Lr34res gene, whereas multiple copies were detected in lines Lr34-5 (three copies) and Lr34-6 (seven copies; Figure S1). Analysis of individual T3 plants from the multicopy Lr34res events showed identical genomic hybridization patterns to the Lr34 probe that were unique to either Lr34-5 or Lr34-6 T3 progeny. This suggests that multicopy events in Lr34-5 and Lr34-6 were inserted at single sites or in close proximity and consequently the absence of segregation of the transgenes.
Given that Lr34 transgenic barley and rice plants exhibited the leaf tip necrosis (Ltn) phenotype at an earlier developmental stage than wheat Risk et al., 2013), we monitored the transgenic sorghum plants for similar morphological changes. Phenotypically, the control plants (sib lines without the transgene) and the Lr34res single copy line, Lr34-2, were very similar at the seedling stage until the onset of booting. Yellowing leaf margins and leaf tips, typical of the Ltn phenotype, occurred earlier in Lr34-2 compared with adult plants lacking the transgene. In contrast, Lr34-5 and Lr34-6 lines showed a progressive development of a blotchy bronze/purple leaf coloration in adult plants from about the penultimate leaf development stage onwards ( Figure S2). The penultimate leaves of adult plants had high Lr34res transcript levels in lines Lr34-5 and Lr34-6, which was 8-13 fold higher than that detected in the single copy Lr34-2 genotype ( Figure 1). Thus, the strong leaf coloration phenotypes correlated with Lr34res expression.
Sorghum rust pathogenesis on plants infected by P. purpurea urediospores at the 5-leaf stage was analysed microscopically at 7 days post-inoculation (dpi) and for sporulation at 12-14 dpi. Microscopic analysis of wheat germ agglutinin-fluorescein isothiocyanate (WGA-FITC) binding to fungal cell walls revealed extensive hyphal development in control plants and sib lines without the Lr34res transgene (Figure 2a-c, e). In contrast, hyphal growth from infection sites in all transgenic Lr34res lines was restricted (Figure 2d, f). Macroscopically, spores from uredinia developed on all the nonLr34res plants, whereas no sporulation was detected on Lr34res transgenic genotypes (Figures 3a and S3). Further quantification of the sorghum rust fungal biomass on transgenic plants showed the presence of the Lr34res transgene reduced fungal colonization by 75%-80% ( Figure 3b). Interestingly, by 28-30 dpi, uredinia had developed on the Lr34-2 transgenic line, albeit at low frequency compared to sib lines lacking Lr34res ( Figure S4). Estimation of fungal biomass at this late period showed approximately a 25% reduction in fungal colonization, which is indicative of the slow rusting phenotype that typifies the partial resistance often seen with Lr34res in wheat. In contrast, no sporulation was detected on Lr34-5 and Lr34-6 genotypes, even at this late stage.
From previous analysis of the Sorghum bicolor genome, two adjacent Lr34 orthologs, Sb01g016770 and Sb01g016775, were considered to have arisen by gene duplication, of which Sb01g016770 was deduced to be a pseudogene . Sb01g016770 and Sb01g016775 share 71% and 75% identity respectively with the protein sequence of LR34. Of the two critical amino acids that distinguish LR34RES from LR34SUS, Sb01g016775 shared the same phenylalanine and tyrosine residues found in the wild-type variant of LR34SUS. We investigated by site directed mutagenesis whether changes to Sb01g016775 involving the two critical amino acids to a modified Sb01g016775 with a deleted phenylalanine (ΔF525) and tyrosine to histidine (Y613H) was capable of conferring resistance to sorghum rust as observed with Lr34res. Five independent stable transgenic lines of Sb01g016775 ÀΔF525, Y613H were generated, all of which expressed transcripts carrying the modified gene. Plants infected with P. purpurea developed similar levels of sporulation as control plants or sib lines by 14 dpi and failed to exhibit the Pathogen-induced leaf pigmentation, expression of genes involved in the flavonoid phytoalexin synthesis pathway and metabolite analysis Within the first 2 days following P. purpurea inoculation, reddish brown pigmented spots were observed on leaves of control and transgenic plants. Leaf area coverage and size of pigmented spots were larger in transgenic plants when compared to nontransgenic sibs and the control genotype ( Figure S6). Furthermore, the Lr34res multicopy genotypes Lr34-5 and Lr34-6 consistently exhibited more pigmented areas than the single copy Lr34-2 transgenic line. To test whether the magnitude of the pathogen-induced pigmentation was associated with Lr34 expression, transcript levels of Lr34res were quantified over a 48-h period post-inoculation. An increase in the Lr34res transcript occurred within 24-h post-inoculation (hpi) and declined by 48 hpi (Figure 4). More than threefold increased expression occurred in Lr34-5 at 24 hpi compared to Lr34-2, which parallels the extent of pigmentation noted on the leaves. We also examined the expression levels of the S. bicolor orthologous Lr34 gene, Sb01g016775 under mock and rust inoculation in comparison with the introduced Lr34res transgene. Interestingly, Sb01g016775 expression was negatively responsive to P. purpurea inoculation in contrast to the pathogen responsiveness of the wheat Lr34res demonstrated through increased expression ( Figure S7).
Pathogen-inducible synthesis of flavanone derived metabolites, some of which have been implicated in plant defense, has previously been described in sorghum (Lo et al., 1996(Lo et al., , 1999Nicholson et al., 1987). Analysis of expression levels of key enzymes involved in 3-deoxyanthocyanidin and flavone biosynthesis (Figures 4b-d, 5) revealed similar trends to effects of the Lr34res transgene. Enzymatic steps encoded by SbFNSII (flavone synthase II, a cytochrome P450 pathogen-inducible gene), SbFNR (flavanone 4-reductase) and SbDFR3 (dihydroflavonol reductase) were elevated in gene expression at 24 hpi and declined at 48 hpi ( Figure 4b-d). The high expressing Lr34res lines, typified by Lr34-5 genotype, exhibited over 15-, 75-and 140-fold increases in expression of SbDFR3, SbFNSII and SbFNR, respectively, at the peak period of 24 hpi. By contrast the control sib line showed 5-, 15-and 20-fold increases for SbDFR3, SbFNSII and SbFNR, respectively. Approximately an eightfold elevation of SbDFR3 was detected in the low Lr34res expressing genotype, Lr34-2, over the same period, whereas SbFNSII and SbFNR showed similar quantitative changes in the control line and Lr34-2 (Figure 4b-d). Taken together, the early induction of this group of genes which form part of the pathway in converting naringenin flavanones to 3-deoxyanthocyanidin and flavone biosynthesis ( Figure 5) is enhanced by the introduction of the wheat Lr34res gene upon pathogen infection.
We further investigated the production of metabolites that belong to the 3-deoxyanthocyanidin class upon infection using the well-studied C. sublineolum-sorghum pathogen-host interaction. Pathogen-induced formation of purple pigments has attributed this colour change to structurally related compounds, 3-deoxyanthocyanidins (luteolinidin and apigeninidin; Dykes and Rooney, 2006). These compounds accumulate within inclusions in the epidermal cells as a defense response to pathogen attack (Snyder and Nicholson, 1990;Snyder et al., 1991). As part of the metabolite analysis we also included the flavone aglycones, luteolin and apigenin, that have also been implicated to differentially accumulate as sorghum phytoalexins in response to pathogen infection . Metabolite analysis was conducted on elongated mesocotyls inoculated with C. sublineolum at 48, 72 and 96 hpi. Significantly enhanced levels of luteolinidin were detected in genotypes carrying Lr34res at 72 hpi ( Figure 6). Differences in metabolite accumulation were not as significant at the other two time points ( Figure S8). At 48 hpi, metabolite amounts were still rather low and by 96 hpi, the 3-deoxyanthocyanidins started to be degraded. Methoxyluteolinidin and methoxyapigenidin levels were also significantly higher in most transgenic lines at 72 hpi, but their levels were not as high as luteolinidin. As expected, the flavones (luteolin and apigenin) accumulated at considerably lower levels than luteolinidin, while some elevation of luteolin levels was detected in the high expressing Lr34res genotypes, L34-5 and L34-6 ( Figure 6).
Transgenic Lr34res expression confers resistance to sorghum anthracnose (C. sublineolum) In addition to the metabolite analysis, we investigated the effect of the Lr34res transgene on disease symptoms caused by infection with C. sublineolum. Necrotic lesion phenotypes on elongated mesocotyls were examined at 7 dpi. Mild symptoms were characterized by single localized lesions, whereas strong symptoms were associated with multiple or complete lesions along the entire length of the mesocotyl ( Figure S9). Strong anthracnose symptoms developed on 65% of the control lines compared with 30% in genotypes carrying the Lr34res transgene ( Figure 7). Analysis of the total symptoms showed approximately 33% and 26% reduction in disease severity associated with the high and low expressing Lr34res lines, respectively ( Figure 7). However, mild symptoms occurred twice as much in the single copy transgenic line compared with the higher Lr34res expressing genotype. Effect of the Lr34res transgene on plant vigour As a general observation, no differences in plant growth vigour were noticed among the control sib lines and Lr34res transgenic lines during the seedling stage and even after the 5-leaf stage when rust inoculations were conducted. However, as the plants approached booting, it was evident that the high expressing Lr34res genotypes (Lr34-5 and Lr34-6) were less vigorous in growth compared to the single copy line (Lr34-2) and sib lines lacking the transgene. To quantify the growth effects and subsequent effect on reproductive development and yield, aspects of panicle morphology and yield were measured. Panicle size tended to be smaller in genotypes with increased Lr34res gene copy number and expression (Figure 8). The mean panicle weight declined by 33% and 67%, respectively, in the single copy and multicopy Lr34res lines, respectively, as compared with the negative sib lacking Lr34res (Figure 9a). The mean peduncle diameter in comparison with the negative sib lacking Lr34res (measured immediately below the node of the basal rachis), was reduced by 1.0 mm in the single copy Lr34res line and 3.2 mm in the multicopy Lr34res genotypes ( Figure S10). The grain yield component of 100-seed weight remained unchanged between the control sib and the Lr34-2 line, whereas a reduction of 0.5-1.2 g occurred in the multicopy Lr34res lines (Figure 9).

Discussion
We demonstrate in this study that the ABC transporter encoded by the wheat Lr34res gene functions in sorghum and confers resistance to sorghum-adapted rust and anthracnose causing pathogens, while Lr34res-mediated resistance to rust caused by Puccinia species has previously been confined to species in the Triticeae (Dyck and Samborski, 1982;Rinaldo et al., 2016;Risk et al., 2013), our findings together with the recently reported observations in maize  extends the efficacy of Lr34res to the warm season adapted Puccinia species with pathogenesis on Andropogoneae taxa. The successful incorporation of Lr34res-mediated resistance into sorghum suggests that the necessary components required for biosynthesis of the Lr34 putative substrate, and proteins involved in signalling and defense response, are also present in sorghum. This finding is of importance as it opens alternate avenues to explore the genetic dissection of Lr34res-mediated resistance. Indeed, the well characterized features of pathogen-induced pigmentation in  sorghum, and associated flavonoid phytoalexin biosynthesis defense response, are avenues that were further investigated in this study. While most of the pathogen-inducible pigments formed in sorghum have been reported with the C. sublineolum and Cochliobolus heterostrophus pathosystems, we show in this study that P. purpurea infection triggers similar phenotypes as part of the early host response in the sorghum cv. Tx430. Such visible phenotypes in wheat plants with Lr34res or other rust resistance genes at 24-72 hpi are yet to be reported. The correlation of Lr34res expression and strength of pathogeninduced pigmentation suggests that the Lr34 transgene interacts with the signalling response, triggering pigmentation. Because the pigments responsible for the pathogen-induced colour changes formed in sorghum are derived predominantly from the 3-deoxyanthocyanidin flavonoids luteolinidin and apigeninidin, which accumulate as a site-specific response to fungal infection (Nicholson et al., 1987;Snyder and Nicholson, 1990), it is conceivable that the presence of the Lr34res transgene may contribute to their elevated accumulation in infected plants. Accumulation of the 3-deoxyanthocyanidins, in particular luteolinidin, occurs much faster in infected cells of resistant genotypes than susceptible genotypes, implicating early phytoalexin accumulation in preventing disease spread by restricting proliferation of fungal hyphae (Basavaraju et al., 2009;Poloni and Schirawski, 2014;Wharton and Julian, 1996). In infected cells, the 3-deoxyanthocyanidins migrate to the site of attempted penetration dependent on nuclear migration, cytoplasmic streaming and intracellular pH to provide an environment for inclusion trafficking and release of the phytoalexins (Nielsen et al., 2004). Exactly how Lr34res fits into this transport processes remains to be defined. Notwithstanding, it is noteworthy that at 24 hpi by P. purpurea, the strong expressing Lr34res transgenic lines exhibited higher expression levels for FNR, DFR3 and FNSII genes that form part of the flavonoid phytoalexin biosynthesis pathway. Interestingly, the introduction of Lr34res into barley also resulted in constitutive up-regulation of genes involved in the flavonoid pathway and in the biosynthesis of barley defense compounds, such as anthranilate synthase, anthranilate N-benzoyltransferase, agmatine coumaroyl transferase and flavonoid 7-O-methyl transferase (Chauhan et al., 2015). Lr34res in hexaploid wheat typically provides partial resistance to rusts and mildew in adult plants, although under lowtemperature conditions (>10°C) seedling resistance can be detected. In the current study, we show that Lr34res functions in seedlings and obviates the need for low-temperature induction in transgenic sorghum to provide resistance to sorghum rust and anthracnose. This observation corroborates seedling resistance by Lr34res against other pathogens reported in barley, rice, maize and durum wheat. The obvious difference in sorghum to other species being the highly expressed pathogen-inducible purple coloration due to phytoalexin production. Expression levels of Lr34res in hexaploid wheat seedlings are elevated under low  temperatures in rust-infected plants (Rinaldo et al., 2016;Risk et al., 2013) and may account for the low temperature-induced resistance. In transgenic sorghum, barley, rice, maize and durum wheat seedlings, it is likely that the expression level of the Lr34res transgene upon pathogen infection reaches a threshold level that is sufficient to trigger resistance by curbing pathogen proliferation. The leaf tip necrosis/early senescence phenotype of lines carrying Lr34res suggests a common pathway confers resistance.
Orthologues of the wheat Lr34 gene are present in the sorghum and rice genomes. Targeted changes to the two amino acids in the sorghum LR34 orthologue to mimic the wheat LR34RES failed to generate a resistance phenotype similar to previous studies with the rice orthologue . Given the amino acid sequence identity of 75% between the wheat and sorghum orthologues of the LR34 ABC transporter, it is possible that other regions of the LR34RES absent in sorghum are required for resistance function. Additionally, the Lr34res in transgenic sorghum is up-regulated upon pathogen infection, but the corresponding sorghum Lr34 orthologue showed a weak negative response to pathogen challenge, which may also account for the lack of resistance phenotype associated with the modified sorghum LR34 orthologue. Thus any attempts at further modifications to the sorghum LR34 orthologue towards engineering resistance will likely require changes beyond the coding region to include pathogen responsive regulatory sequences.
Overexpression of Lr34res typified by the multicopy transgenic Lr34-5 and Lr34-6 genotypes in this study results in plants with reduced yield components, despite exhibiting immunity against P. purpurea. Conversely, the single copy low-expressing line (Lr34-2) had similar growth vigour as the non-transgenic or sib line control plants and less detrimental effects on reproductive yield. As Lr34-2 showed no rust symptoms 2 weeks postinfection and reduced rust sporulation after 4 weeks, indicative of the characteristic slow rusting response of the Lr34res, and it holds promise for the use of Lr34 as a transgene for sorghum improvement. It may also be useful to explore high expression of Lr34res for plant immunity to various pathogens using pathogeninducible promoters in an attempt to overcome detrimental reproductive yield effects associated with constitutive overexpression in adult plants. Our findings in sorghum that Lr34res confers resistance to sorghum rust and anthracnose demonstrates that the multipathogen resistance of the wheat Lr34 gene extends to biotrophic and hemibiotrophic adapted pathogens across the Triticeae, Oryzeae and Andropogoneae taxa.

Production of transgenic Lr34 sorghum
The genomic construct of Lr34res under the native promoter and terminator sequences was cloned into plasmid pWGEM-NZf as previously described (Risk et al., 2012) and subsequently transformed into the sorghum inbred line Tx430 via microprojectilemediated transformation (Liu and Godwin 2012). The presence of the transgene in T0 plants was initially assessed by PCR with Lr34res-specific primers ) and subsequently by genomic blots probed with the Lr34 3 0 UTR DNA fragment. A genomic construct containing the sorghum Lr34 ortholog  (Sb01g016775) was generated from an EagI 16.3 kb DNA fragment from a sorghum BAC clone (CUGI BAC#156N20) subcloned into pWGEM-NZf. Site directed mutagenesis using primers Sb2Quickchange 1F/1R (1F-TGGGAGCATTATATTTTTC CATCATCATTATGCTAAATGGCATACC/1R-GGTATGCCATTTAG CATAATGATGATGGAAAAATATAATGCTCCCA) and Sb2Quickchange 2F/R (2F-CATCAATCAGTAATGGCGTTCCATCGATTTG TCGCTTCTTATG/2R-CATAAGAAGCGACAAATCGATGGAACGC CATTACTGATTGATG) was used to generate the derived subclone Sb01g016775 ÀΔF525, Y613H as per protocols in Krattinger et al. (2016). Transgenic plants with the genomic construct of Sb01g016775 ÀΔF525, Y613H with its native promoter and terminator sequences were also generated by microprojectile bombardment.

Identification of transformants and Lr34 copy number
Leaf samples (2-3 g) from T1 and T3 plants were ground in liquid nitrogen using pestle and mortars and sand. Frozen leaf material was transferred into 3 mL CTAB extraction buffer (100 mM Tris-HCl pH 8.0, 20 mM EDTA, 1.4M NaCl, 0.5% Na 2 S 2 S 5 , 2% CTAB, and 1% b-mercaptoethanol) and processed for DNA isolation in accordance with Collins et al. (1998). About 12 lg of each gDNA sample was subjected to NotI and EcoRV restriction endonuclease digestions to ascertain the presence of full-length Lr34 gene constructs and the copy number of Lr34 in transgenic sorghum lines, respectively. Digested gDNA samples were loaded on 1% agarose gels and run at 60V (at approx. 50 mA) for 18-20 h, capillary transferred onto Hybond-N+ â filter using 20xSSC buffer and UV cross-linked. Filter was subjected to 5-h prehybridization in 30 mL prehybridization solution containing salmon sperm DNA at 65°C in a hybridization tube. Subsequent hybridization incorporating 50 ng of probe DNA ( 32 P-labelled Lr34-3 0 UTR probe from the amplicon generated using ABCTEX1314F-CAGAACACCTA CAGAAGAATATC and ABCR9-GGCAAGTAGCTATATCTGTAAC) was performed according to Lagudah et al. (1991).

Sorghum rust inoculation
Sorghum seed was germinated and grown in pots at 20°C in a glasshouse or in a growth chamber. Plants at the 5-7 leaf stage were placed in closed inoculation chambers and allowed to acclimatize for 18-24 h. Puccinia purpurea urediniospores collected from Hermitage Research Station, Warwick, Queensland (28.2102°S, 152.1041°E) in Queensland, Australia (White et al., 2015) were suspended in 100-150 mL distilled water with 1-2 drops of Tween 20, and sprayed onto plants using a Preval â compressed air atomizer (Preval Sprayer Division). A complementary set of plants were treated identically but without urediniospores as controls. Inoculation chambers were closed to maintain high (80%+) humidity and placed in darkness at 20°C for 24 h before being transferred to diurnal conditions (16 h light, 8 h dark) to allow rust infection to develop.

Anthracnose infection
Sorghum seeds were sown in rolls of germination paper and kept in darkness for 4 days at 28°C (Lo et al., 1996). Etiolated seedlings with elongated mesocotyls were inoculated with spore suspensions of C. sublineolum, at the concentration of~3.0 9 10 6 conidia/mL with gelatin as a wetting agent (0.25%). Inoculated plants were incubated at 100% relative humidity at room temperature for 24 h. Three independent infections were performed on elongated mesocotyls. The phenotypes were examined after 7 days following inoculation on an average of 35 plants per genotype. For each genotype the number of plants with mild symptom, strong symptom or no symptoms were recorded.

Microscopy
At 6-8 dpi, fourth or fifth leaf samples were collected and submerged in 1M KOH and incubated for 48 h at 37°C with gentle agitation. The KOH solution was replaced with fresh 1M KOH solution after 18-24 h. The KOH solution was discarded, and the leaf material was washed gently 2-3 times with 50 mM Tris-HCl, allowing material to incubate in the Tris-HCl solution for 10-20 min per wash. 1-2 mL 50 mM Tris-HCl and 10-20 lL 1 mg/mL wheat germ agglutinin conjugated to fluorescein isothiocyanate (WGA-FITC, Sigma-Aldrich, Castle Hill, NSW, Australia) were added, and samples incubated at ambient temperature for an hour or kept at 4°C before mounting on microscope slides. Stained leaf samples were mounted on slides using a few drops of 40% glycerol before covering with cover slips. GFP3 fluorescence filters were used on a Leica MZFLIII fluorescence dissecting microscope or a Zeiss Axioimager upright fluorescence microscope to score the presence of P. purpurea infection sites in the sampled leaves.

Rust biomass assays
Chitin assays were carried out as described by Ayliffe et al. (2014). Three biological replicates of the 6th leaves were sampled 14 and 28 dpi and weighed. Leaves were cut into 1.5-2.0 cm fragments and submerged in 1M KOH with 0.15 Silwet L-77 in Falcon tubes. Leaf samples are autoclaved at 121°C and 15 psi for 20 min, then washed gently three times in 50 mM Tris-HCl pH 7.5. Plant samples were suspended in 50 mM Tris-HCl pH 7.5 at the rate of 200 mg fresh weight per mL and homogenized by sonication for 1-2 min to form a fine uniform suspension. About 4 9 100 lL of each homogenate was transferred to 4 9 200 lL PCR tubes. About 10 lL of 1 mg/mL WGA-FITC (Sigma-Aldrich) was added to each homogenate in PCR tubes and left at ambient temperature for an hour. Homogenates were washed three times by centrifuging at 250 g for 3-5 min and carefully replacing the supernatants with 50 mM Tris-HCl pH 7.5 using a micropipette. The final washed suspensions were transferred to a 96-well fluorometer microtiter plate. Fluorescence values of each sample were measured in a Wallac Victor 1420 multilabel counter at 485 nm excitation and 535 nm emission wavelengths with a 1.0 s measurement time. Means of the technical replicate fluorescence values were calculated, and the standard errors were ascertained for biological replicates.

Metabolite analysis
Samples of mesocotyl tissue (~200 mg each) from uninfected and C. sublineolum inoculated plants at 48, 72 and 96 hpi were collected, cut into segments, weighed and placed in acidified (0.1%) HPLC-grade methanol. Metabolites were allowed to leach from the tissue at 4°C overnight. The composition of plant extracts was then determined by liquid chromatography-tandem mass spectrometry (LC-MS) in accordance with the protocols as described Lo et al., 1999). Authentic standards of luteolinidin, apigeninidin, luteolin and apigenin (Sigma) were used for metabolite identification and quantification.

qRT-PCR
The sixth leaf of plantlets were sampled at 0, 24 and 48 hpi, snapfrozen in liquid nitrogen and stored at À80°C. Australia) according to manufacturer's instructions. About 1-2 lg RNA samples were subjected to first-strand DNA synthesis in 20 lL reactions using Superscript â III reverse transcriptase Life Technologies (Mulgrave, VIC, Australia). About 3 lL of 1-in-10 dilutions of first-strand synthesis products were subjected to qPCR reactions using the C1000 Touch TM thermocycler with the CFX96 TM Real-Time System (Bio-Rad, Gladesville, NSW, Australia). Reaction conditions included an initial denaturization at 95°C for 3 min; 40 cycles of denaturization at 95°C for 10 s and annealing/elongation at 60°C for 30 s, followed by a melt step range of 65-95°C with increments of 0.5°C. The sorghum actin gene (Pavli et al., 2011) was used as a reference gene for each qRT-PCR experiment, and each qRT-PCR experiment was repeated using the more stable sorghum reference gene PP2A (Reddy et al., 2016). qPCR primers specific for Lr34res, SbL34 (Lr34 orthologue), SbFNR, SbFNSII and SbDFR3 were used to measure the relative gene expressions at the different postinoculation time points are listed in Table 1. Experiments included three technical replicates of each of three respective biological replicates. Means of the DCq values were calculated, and Standard Errors were determined for the data. Gene expression values were log(base 2)-transformed, and a repeated measures analysis was performed via the linear mixed model software asreml (Butler, 2009) in R (R Core Team, 2015). Means and SE bars in Figure 4 are back-transformed to the scale of the raw expression levels.

Reproductive yield components
Sorghum plants at the T2 generation and four replications were grown to physiological maturity, and intact panicles were harvested and dried at 37°C for 48 h. Individual panicles were weighed for each of the negative and positive Lr34 lines. Peduncle diameters were measured 1 mm above the last node using Vernier calipers. Kernels were separated from the panicles and 100 kernel quantities from transgenic and non-transgenic lines were weighed.

Supporting information
Additional Supporting Information may be found online in the supporting information tab for this article: Figure S1 Genomic blot of EcoRV-restricted DNA from transgenic Sorghum. Figure S2 Lr34 sib and transgenic sorghum penultimate leaves. Figure S3 Rust sporulation on sorghum leaves 5 and 6 at 15 days post-inoculation with P. purpurea. Figure S4 (a) Pustule development and (b) fungal biomass on transgenic sorghum leaves one month post-inoculation with P. purpurea. Figure S5 Rust sporulation at 14 days post-inoculation in Lr34res transgenics and altered variants of Sb01g016775 ÀΔF525, Y613H . Figure Figure S7 Relative gene expression of sorghum Lr34 ortholog (Sb01g016775) at 0 and 24 h post-inoculation. Figure S8 Flavonoid metabolites in sorghum mesocotyls after infection with C. sublinoeleum. Figure S9 Anthracnose symptoms (arrows and yellow bracket) following mesocotyl infection by C. sublinoleum. Figure S10 Comparison of peduncle diameters of transgenic sorghum lines. Data shown as mean AE SE from 4 to 6 biological replicates.