In order to identify important defense genes to S. turcica, cDNA-AFLP analysis was carried out on susceptible (S) and resistant (R) sorghum and maize genotypes following fungal infection. In our case, the Ugandan sorghum genotypes GA06/18 (R) and Sila (S) and the maize A619Ht1 (R) and A619 (S) lines were used. The sorghum material had earlier been evaluated on various agronomical traits including important fungal diseases. Apart from S. turcica responses, GA06/18 was found to be susceptible to Cercospora sorghi, and Colletotrichum sublineolum, whereas Sila was susceptible to C. sorghi and resistant to C. sublineolum.

In total, approximately 3000 transcript-derived fragments were monitored ranging from 50 to 600 bp in size using different primer combinations (Additional file 1). Unique, up- or down-regulated transcripts in the resistant genotypes compared to the susceptible, sampled at 24, 48 and 72 hours post inoculation (hpi) were excised, amplified, sequenced and analyzed for putative function. The final transcript-set comprised of 68 sorghum and 82 maize gene candidates. Among these genes, 11 and 13, respectively, were putative stress-related according to closest genes identified in other organisms using BLASTP.

One CC-NB-LRR encoding putative R-gene (GRMZM2G005347), a member of a homologous gene pair with GRMZM2G005452 in the same locus on chromosome 2, and uniquely expressed in the resistant maize genotype, was further studied (Figure 1D). Genome analysis revealed presence of 6 homologous genes in sorghum (Figure 1A). These six genes were given the prefix St referring to S. turcica and designated St1A (Sb05g008280), St1B (Sb05g008140), St2A (Sb05g008350), St2B (Sb05g008030), St3A (Sb05g008250), and St3B (Sb05g008270). Quantitative real-time PCR confirmed furthermore that five (St1A, St2A, St2B, St3A and St3B) of the six St genes showed high relative transcript levels when the sorghum resistant GA06/18 plants were challenged with S. turcica (Figure 2). One gene, St1B, was expressed to a much lower extent compared to the other St genes, outside the limit of detection. In Sila, only St2B and St3A showed a significant increase (P < 0.005) in expressions when challenged with S. turcica (Figure 2).

The six St genes in sorghum form three gene pairs in a cluster on chromosome 5 and share a common ancestor (Figure 1; Additional file 2; Additional file 3). St gene orthologs were also found in clusters when searching the rice, maize, foxtail millet and Brachypodium genome databases. The St gene encoded proteins from the other grass species, grouped with the sorghum St proteins with high edge support (100) (Additional file 2). The rice genome contains orthologs of sorghum St1A, St1B, St2A, St2B and an St3 gene (Figure 1A, B). This indicates that the ancestor of rice and sorghum likely had a copy of these genes. Sorghum St3A and St3B are likely a result of a more recent genome duplication event after the split between the rice and sorghum species (Figure 1G). The rice genome also contains multiple copies of St1A, St2A and St2B orthologs, likely produced from gene duplications after the species split from sorghum. Likewise, the Setaria italica (foxtail millet) genome contains orthologs of St1A, St1B, St2A and St2B, with seven of the nine genes found in a cluster within the same scaffold, as complete chromosome annotation have yet to be determined (Figure 1C). An St3 homolog was not found in millet. In addition to the maize gene pair identified in our cDNA-AFLP analysis, BLASTP and BLASTN searches revealed a third single gene homolog, GRMZM2G050959, St2A on maize chromosome 2 (Figure 1D). The model grass Brachypodium genome, on the other hand, has a gene pair orthologous to St1B on chromosome 4, and one to St2B on chromosome 5, but lacks all other gene homologs (Figure 1E). The St gene cluster is maintained between sorghum, rice and possibly millet genomes but is smaller in maize and Brachypodium with St genes located across or on different chromosomes.

Sequence homology was also found between sorghum St proteins and Arabidopsis CC-NB-LRR encoding genes (Figure 3; Additional file 4). All six St proteins formed a cluster together with the CC rather than TIR domain containing R proteins from Arabidopsis indicating a closer evolutionary relationship as expected. The nearest related Arabidopsis gene is RPM1, a gene mediating resistance to Pseudomonas syringae isolates expressing the avrRpml or avrB genes 27 .

Genetic transformation of sorghum and maize is possible but laborious and requires other genotypes than those used in this study to be successful 28 29 . Hence, our candidate genes were further studied using virus induced gene silencing (VIGS) using the Brome mosaic virus (BMV) system, previously used to silence genes in monocots 30 . VIGS was followed by fungal inoculation to assess the potential defense function of the St genes. In our hands, the VIGS procedure was not successful when applied to the A619Ht1 maize genotype. Because the St genes were up-regulated upon fungal inoculation with S. turcica in our sorghum GA06/18 genotype (Figure 2), we continued the studies on our sorghum materials.

Two VIGS constructs (1 and 2) with high identity to the 6 St genes in sorghum were designed (Figure 4) including examination for their off-target gene silencing capacity. The highest non-St sorghum gene similarity belongs to a related R-gene pair, Sb10g028720 and Sb10g028730, located in a different subgroup upon phylogenetic analysis (Additional file 2), and used as a control for off-target gene silencing. The selected sequences were amplified and ligated into the third plasmid (pF13m) in the BMV system, and used to infect the sorghum plants.

The VIGS procedure was first optimized. Sorghum seeds were surface sterilized before sowing to minimize additional stress by other microorganisms. mRNA was produced by in vitro transcription, added to inoculation buffer and rubbed directly onto the second leaf of three week old sorghum plants. No intermediate step involving barley as virus host was used. The virus spreads systemically throughout the plant with silencing greatest in the second and third leaves above the inoculation site and complete silencing rarely achieved 30 . Seven days post infection (dpi), light green colored streaks were visible on the third leaf, indicating viral symptoms and successful infection by the virus. In order to confirm onset of silencing quantitative real time-PCR was carried out on leaf samples from the VIGS treated plants (Figure 5). There was a significant decrease in the relative transcript levels in relation to control plants inoculated with empty plasmid suggesting a clear down-regulation of five of the six targeted genes, particularly by construct 1, in both sorghum genotypes. Relative transcript levels of Sb10g028720 and Sb10g028730 were not influenced in VIGS treatments indicating no off-target silencing.

Fungal colonization and growth on plants inoculated with the different VIGS constructs compared with control material was carefully monitored. The different phenotypic observations are summarized in Figure 6; and Additional file 5. Fungal growth was further assessed by detaching infected leaves and placing them in a petri dish containing moist filter paper followed by incubation in the dark at 25°C for two days, as described by Levy 31 . The development of conidiophores protruding through leaf lesions followed by rapid asexual spore development indicated fungal colonization of the leaf material, and a susceptible phenotype.

A hypersensitive response (small dark/red spots) occurred at 2 dpi on the resistant GA06/18 genotype upon fungal challenge while the plants treated with empty vector produced a somewhat delayed HR phenotype 3 dpi. When VIGS construct 1 was applied to GA06/18 plants prior to fungal inoculation, larger and more numerous lesions with chlorotic halos developed compared to the control plants. Disease lesions spread laterally along the leaf and fungal conidiophores and spores were produced under sporulating conditions. Similarly, when the effect of construct 2 was assayed, the disease lesions were seen 2 dpi and spread laterally to form large lesions that produced large numbers of fungal spores. The disease lesions were larger than those induced by construct 1, at 7 dpi. On the susceptible Sila plants clear disease symptoms, necrotic spots, and chlorotic halos around fungal appressoria were seen 2 dpi. Large numbers of asexual fungal spores were produced on conidiophores protruding from necrotic lesions. When Sila plants were inoculated with the empty VIGS vector, prior to fungal inoculation, similar disease symptoms occurred 2 dpi. In contrast, on Sila plants inoculated with our VIGS construct 1, slightly larger and more frequent lesions appeared compared to control plants. The disease symptoms were further amplified when construct 2 was used, resulting in larger necrotic lesions, and profuse fungal sporulation. In order to correlate these observed disease phenotypes with fungal growth, fungal DNA was quantified in the VIGS materials (Figure 7). S. turcica DNA increased to 1.5 ± 0.4 pg/ng sorghum DNA in GA06/18 leaves inoculated with VIGS construct 1, and to 3.6 ± 0.9 pg/ng sorghum DNA when using construct 2, from a near zero level in control plants (non-VIGS and empty vector). A significant (P < 0.005) increase in fungal DNA was also found in samples from Sila inoculated with construct 1 (1.2 ± 0.4 pg/ng sorghum DNA), and construct 2 (0.8 ± 0.9 pg/ng sorghum DNA), compared to control samples with approximately 0.5 pg/ng sorghum DNA.

Taken together, as expected the resistant GA06/18 genotype showed a compromised defense response when inoculated with VIGS construct 1 or 2 prior fungal inoculation. Interestingly, we observed enhanced disease phenotypes on the susceptible Sila genotype upon corresponding VIGS treatments.

Resistant (R) and susceptible (S) Sorghum bicolor genotypes from Uganda, GA06/18 (R) and Sila (S), and maize lines A619Ht1 (R) and A619 (S) provided by USDA ARS, were used in the study. The plants were grown in a growth chamber (Percival) using a 12/12 h photoperiod at 22°C. A single spore isolate from S. turcica infected sorghum (Ig1), or infected maize (Mb1), collected from Iganga and Mbale, Uganda, was used for all sorghum and maize analysis, respectively. The fungal DNA was extracted using a modified CTAB method 42 . DNA was analyzed by using S. turcica specific ITS1 and ITS2 primers (F -GCAACAGTGCTCTGCTGAAA and R-ATAAGACGGCCAACACCAAG). PCR was carried out using the following conditions: 10 ng of template DNA was added to a 24 μl mix consisting of H₂O, 2.5 mM MgCl₂, 2.5 μl Taq buffer (Fermentas, Helsingborg, Sweden) 0.2 mM of each dNTP, 0.25 μM of forward and reverse primers and 1 U of Taq polymerase (Fermentas) with: 3 min at 94°C, 35 cycles of (1 min at 94°C, 1 min at 60°C, and 1.5 min at 72°C), and final extension at 72°C for 10 min. The PCR products were separated on 1% agarose gels to confirm fragment size, (344 bp) followed by sequencing (Macrogen Inc., Seoul, Korea).

Three-week old seedlings were inoculated on the third leaf whorl with 25 μl conidia suspension (5 × 10⁵ conidia/ml) as described by Carson 43 . Inoculated leaves from three to four plants were pooled and harvested at 24, 48, and 78 hours post inoculation (hpi) for cDNA-AFLP analysis. Water treated control samples were harvested at the same time-points.

Total RNA was isolated from the leaf samples using the BioRad RNA isolation kit (BioRad, California, USA) followed by mRNA preparation with the mRNA capture kit (Roche, California, USA). cDNA was synthesized with Oligo-dT primer and RevertAid™ H Minus M-MuLV Reverse Transcriptase (Fermentas). Second strand was synthesized using E. coli DNA Polymerase I (Fermentas). The double stranded cDNA was digested with BstY1 and Mse1 (Fermentas) and ligated to respective adaptors, pre-amplified and later selectively amplified using the BstYI +N (³³P labeled) and MseI +N primers. Pre-amplification was carried out with the adapter-ligated cDNA, Taq DNA Polymerase (Fermentas) and the non-selective primers specific to the BstYI and MseI adapters using 25 cycles of 94°C for 30 s; 56°C for 1 min and 72°C for 1 min. The pre-amplified reaction mixture was diluted 600-fold and 5 μl was used for final selective amplification with 24 primer combinations, carried out with BstYI +N (³³P labeled) primers (Additional file 1) and touchdown amplification 44 . The selective amplification products were resolved on 6% polyacrimide gel run at 100 W until 4300 Vh was reached. Gels were dried and exposed to Kodak Biomax film (Amersham Pharmacia, California, USA) for 5-7 days.

Approximately 150 transcripts (unique, up and down-regulated) from the resistant genotypes in relation to the susceptible genotypes, were excised from the dried PAGE gels, eluted in H₂O and PCR amplified using the non-selective primers under the same conditions as earlier described in the pre-amplification step. The products were cloned into the pJET 1.2 blunt vector (CloneJET™ PC, Fermentas) and sequenced. The sequences were analyzed using the BLASTN and BLASTX programs 45 and compared with sequences deposited in NCBI, GRAMENE and PHYTOZOME databases. Identified fungal sequences were excluded.

The VIGS system used is based on the monocot-infecting Brome mosaic virus (BMV) as previously described 30 but pre-inoculation on barley was excluded. The BMV VIGS vector consists of three plasmids harboring BMV RNA1 (p1-1), RNA2 (p2-2) and RNA3 (pF13m, formally pF3-5/13_A/G), respectively. To generate VIGS constructs, PCR fragments ranging from 246 to 253 bp in size were amplified from the sorghum candidate gene using genomic DNA of the resistant GA06/18 genotype and gene-specific primers harboring NcoI and AvrII restriction sites using the Primer 3 version 0.4.0 http://frodo.wi.mit.edu/primer3/ software (Additional file 6). Prior to PCR amplification, off-target gene searches were undertaken to design optimal VIGS constructs (Figure 4). After restriction, each fragment was cloned into the corresponding site of the pF13m plasmid. The identity of the inserts was verified by sequencing. P1-1, p2-2 and the pF13m containing different constructs were digested with SpeI, PshAI and PshAI, respectively. Infectious RNA transcripts were synthesized from linearized plasmids through in vitro transcription using T3 Polymerase (Fermentas), according to manufacturer instructions. 1 μl of the reaction product was run on a 1.5% agarose gel to confirm presence of a transcript.

Plant inoculation procedures were performed as described 30 with slight modifications. A 10 μl aliquot of the transcription mix from each of the plasmids p1-1, p2-2 and pF13m-insert was combined with 30 μl FES inoculation buffer and used directly to rub inoculate the second and third leaves of 3-week-old sorghum and maize plants. As a control, plants were inoculated in the same way with water or combined transcripts from p1-1, p2-2 and empty pF13m. Maize and sorghum plants were challenged with S. turcica as earlier described one week after viral inoculation (when faint chlorosis and vein clearing started to appear) to assess the effect of the different constructs. Plants were randomized and coded to reduce potential bias in the scoring of fungal colonization and growth.

Prior to fungal inoculation of the VIGS treated sorghum plants, approximated 3 cm of the second leaf above the VIGS inoculated leaf was collected from 3 independent plants in triplicates for each condition and used for RNA extraction as previously described. First strand-cDNA was synthesized from 1 μg of total RNA, with Oligo-dT primer and RevertAid™ H Minus M-MuLV Reverse Transcriptase (Fermentas) according to the manufacturer's instructions. Real-time PCR was carried out using the first strand cDNA in an iQ5 cycler (Bio-Rad). Maxima Sybr Green/Fluorescein qPCR Master Mix (Fermentas) was used for PCR amplification in a 20 μl total reaction volume consisting of 10 μl of SYBR Green qPCR Master Mix, 0.3 μM forward and reverse primers and 5 ng of cDNA template. All PCRs were performed in triplicate under the following amplification conditions; 10 min at 95°C followed by 40 cycles of 95°C, for 15 s, 60°C for 30 s, and 72°C for 30 s, followed 1 min at 95°C, and melt curve analysis. Primers sequences for St genes were designed using the Primer 3 version 0.4.0 http://frodo.wi.mit.edu/primer3/ software (Additional file 7). The sorghum elongation factor 1-alpha (Sb02g036420) and Actin (Sb01g010030) were used as reference genes and relative transcript values were calculated. All calculations and statistical analyses were performed as described in the ABI PRISM 7700 Sequence Detection System User Bulletin #2 (Applied Biosystems, USA) slightly modified by Vetukuri et al. 46 . Quantification of S. turcica DNA on VIGS material 7 days post fungal inoculation was carried out as earlier described 41 . Approximately 3 cm of leaf material from three plants was pooled and DNA extracted. Three biological samples per treatment were analyzed. Statistical significance was calculated using Student's t-test.

The amino acid sequences of St1A (Sb05g008280), St1B (Sb05g008140), St2A (Sb05g008350), St2B (Sb05g008030), St3A (Sb05g008250) and St3B (Sb05g008270) were aligned to sorghum, maize, millet, rice, Brachypodium and Arabidopsis genome databases using BLASTN and BLASTP (PHYTOZOME). Predicted domains were identified using coiled-coil prediction 47 ), LRRfinder 48 and CD-Search 49 . St-like gene loci were identified using Genomic Viewer (PHYTOZOME). Phylogenetic analysis was conducted using Treefinder and maximum likelihood and 10, 000 replicates 50 . The JTT+G model 51 was found to best fit the data using ProtTest v2.4 52 . Confidences were calculated using local rearrangement of expected likelihood weights (LR-ELW) 53 . Phylograms were drawn using Treeview 1.6.6 54 .