Transcript-Based Cloning of RRP46 , a Regulator of rRNA Processing and R Gene–Independent Cell Death in Barley–Powdery Mildew Interactions

Programmed cell death (PCD) plays a pivotal role in plant development and defense. To investigate the interaction between PCD and R gene–mediated defense, we used the 22K Barley1 GeneChip to compare and contrast time-course expression proﬁles of Blumeria graminis f. sp hordei ( Bgh ) challenged barley ( Hordeum vulgare ) cultivar C.I. 16151 (harboring the Mla6 powdery mildew resistance allele) and its fast neutron–derived Bgh-induced tip cell death1 mutant, bcd1 . Mixed linear model analysis identiﬁed genes associated with the cell death phenotype as opposed to R gene–mediated resistance. One-hundred ﬁfty genes were found at the threshold P value < 0.0001 and a false discovery rate <0.6%. Of these, 124 were constitutively overexpressed in the bcd1 mutant. Gene Ontology and rice ( Oryza sativa ) alignment-based annotation indicated that 68 of the 124 overexpressed genes encode ribosomal proteins. A deletion harboring six genes on chromosome 5H cosegregates with bcd1 -speciﬁed cell death and is associated with misprocessing of rRNAs but segregates independent of R gene–mediated resistance. Barley stripe mosaic virus -induced gene silencing of one of the six deleted genes, RRP46 ( rRNA-processing protein 46 ), phenocopied bcd1- mediated tip cell death. These ﬁndings suggest that RRP46, a critical component of the exosome core, mediates RNA processing and degradation involved in cell death initiation as a result of attempted penetration by Bgh during the barley–powdery mildew interaction but is independent of gene-for-gene resistance. demonstrate that RRP46 within the bcd1 deletion is required for R gene–independent PCD and the normal processing of rRNAs.


INTRODUCTION
Programmed cell death (PCD) is regulated by several diverse pathways in plants. Examples include the initiation and execution of senescence during plant development, the hypersensitive response (HR) during pathogen attack, and cellular damage sustained from various abiotic stresses (Lam, 2004). PCD can be initiated by internal or external factors and is under the control of active genetic programs. During leaf senescence, cell death usually starts from the tips or margins and progresses toward the base; however, earlier senescence can occur on targeted parts of the plant under environmental stress (Lim et al., 2007). In contrast with this cellular aging program, the HR is triggered when the plant recognizes a pathogen and generates a local cellular suicide, with the intent of halting pathogen progression (Sasabe et al., 2000). Each PCD program has discrete outcomes, and yet, conserved genes do exist among defenseand senescence-induced pathways (Quirino et al., 1999;Quirino et al., 2000). One such example is the tobacco (Nicotiana tabacum) harpin-induced1 gene, an HR cell death marker, which is also expressed at late stages of leaf senescence (Takahashi et al., 2004). Conversely, gene expression has also been found to be PCD specific, such as the induction of tobacco hsr203J for incompatible interactions with the bacterial pathogen Ralstonia solanacearum compared with the induction of Arabidopsis thaliana SAG12, encoding a putative Cys protease, during senescence (Pontier et al., 1999). Functional dissection of gene expression at key moments in cell death has facilitated our understanding of the mechanisms of cell-to-cell crosstalk and PCD as a whole.
In the interaction between host plants and their pathogens, cell death occurs in both incompatible and compatible responses. It is normally believed that, in the incompatible response, HR is activated by the R-AVR interaction, which restricts pathogen ingress, whereas compatibility triggered cell death may facilitate pathogen infection (Kim and Palukaitis, 1997;Greenberg et al., 2000). By contrast, several studies indicate that cell death and the prevention of pathogen invasion likely have separable functions in disease resistance (Bendahmane et al., 1999). For instance, the Rx gene conditions resistance to Potato virus X without cell death, but the Potato virus X coat protein elicits cell death when transformed into the Rx plants, leading to HR. Genetic investigations in oat (Avena sativa) revealed that Rds and Rih mediate HR, although they act independent of genefor-gene resistance to the oat crown rust pathogen Puccinia coronata f. sp avenae (Yu et al., 2001). The Arabidopsis dnd1 and dnd2 mutants retain gene-for-gene mediated defense but with the reduced ability to produce HR (Yu et al., 1998(Yu et al., , 2000. The dnd mutants also exhibit increased broad-spectrum resistance as a result of the activation of multiple defense pathways, none of which are required for the reduced HR (Genger et al., 2008). Hence, diverse pathways mediate cell death and each may have roles or control different aspects of plant-pathogen interactions.
Cellular features of PCD include fragmentation of nuclear DNA, signal transduction involving Ca 2+ fluxes, changes in protein phosphorylation, increase in nuclear heterochromatin, and induction of reactive oxygen species (Greenberg, 1996;Pennell and Lamb, 1997). In plants, cell death is often correlated with development and metabolism and their associated differentiation, reproduction, and vegetative growth (Beers, 1997;Tamagnone et al., 1998). RNA processing represents an essential metabolic pathway throughout these stages. Central to this pathway is the exosome complex, which mediates 39 to 59 RNA processing and degradation in early pre-rRNA processing steps (Mitchell et al., 1997;Allmang et al., 2000). Recently, Chekanova et al. (2007) used genome-wide transcript mapping of exosome substrates to demonstrate the regulatory role of the exosome in mediating RNA quality control and stable structural RNA metabolism in plants. Other studies have shown that the exosome exhibits cell death-related nuclease function, leading to apoptotic DNA degradation in Caenorhabditis elegans and mammals (Parrish and Xue, 2006).
Barley (Hordeum vulgare) powdery mildew, caused by Blumeria graminis f. sp hordei (Bgh), is an ideal system to study the interactions between obligate fungal biotrophs with their hosts . The well-characterized Bgh infection stages (Clark et al., 1993;Hall et al., 1999) provide the opportunity to monitor the host response and associated changes in gene expression upon pathogen attack. In this report, we isolated the fast neutron-derived, Bgh-induced tip cell death1 mutant, bcd1, and used Barley1 GeneChip expression profiles to discern the reprogramming of the transcriptome associated with the bcd1 mutation compared with its progenitor upon challenge by Bgh. Further bioinformatic and functional analytic methods were used to elucidate that RRP46, encoding a predicted rRNA processing protein, regulates cell death in an R geneindependent manner, implying a shared signaling pathway for rRNA processing and cell death, independent of the resistance response.

Mutant Characterization and Experimental Design
Mutants isolated for their pathogen-induced cell death provide an initial look at the cellular processes underlying crosstalk of cell death-mediated pathways (Lorrain et al., 2003;Love et al., 2008). In concert, the diversity of host responses to pathogen attack presents an opportunity to answer specific questions by selecting appropriate contrasts of plant and pathogen genotypes . We used the bcd1 mutant, selected from a group of fast neutron-derived C.I. 16151 mutants (see Methods), where bcd1 was identified by the developing tip cell death upon Bgh inoculation while retaining resistant or susceptible responses to avirulent or virulent isolates, respectively ( Figure 1). The bcd1 mutant exhibits rapid cell death in the top 2 to 3 cm of first leaves within 3 d after inoculation, even though Bgh conidia are attempting penetration at thousands of sites throughout its entire 10-cm length. To further characterize bcd1-mediated cell death, Bgh inoculated leaves were examined at 24, 48, 72, and 96 h after inoculation (HAI) using fluorescence microscopy. As shown in Figures 2A and 2B, Mla6-specified HR was observed in both wild-type C.I. 16151 and bcd1 leaves at 24 HAI. At 48 HAI, C.I. 16151 and bcd1 epidermal cells continued to exhibit HR ( Figures 2C and 2D). In addition, bcd1 mesophyll cells displayed Transcript profiling was based on a split-split-plot design with replications as blocks, Bgh isolate as the whole-plot factor, plant genotype as the split-plot factor, and time as the split-split-plot factor. Seven-day-old plants (1st leaf) of the wild type (C.I. 16151) and fast neutron-derived mutant, bcd1, were inoculated with respective Bgh isolates 5874 (AVR a6 ) and K1 (avr a6 ). Fifteen first leaves of inoculated barley seedlings were harvested at 0, 8, 16, 20, 24, and 32 HAI. One Barley1 GeneChip was used for each of the 72 split-split-plot experiment units corresponding to three replications 3 two isolates 3 two genotypes 3 six time points. The infection types shown above were photographed 7 d after inoculation. Infection type is classified on a scale of 0 to 4 and is shown in the top right of each panel. An infection type of 0 to 2 is resistant ("À" designates incompatible/no sporulation), whereas an infection type of 3 to 4 is susceptible ("+" designates compatible/abundant sporulation). 0, Immune; 1n, few small necrotic flecks (0.5 mm); 1 to 2n, significant small necrotic flecks (1 mm); 2N, abundant large necrotic flecks (1.5 mm); 4sp, completely susceptible; c, limited chlorosis; C, abundant chlorosis. accumulation of phenolic compounds beneath epidermal cells that exhibit HR ( Figure 2D), as well as in mesophyll below non-HR-producing epidermal cells ( Figure 2E). At 72 HAI, bcd1 mesophyll exhibited more severe symptoms (Figures 2G and 2H), resulting in collapse at 96 HAI ( Figure 2J). Pleiotropic effects are observed in bcd1, which manifest as macroscopic cell death on first and then second and third leaves of tillers prior to anthesis and a 23% increase in the number of sterile florets (P < 0.0001) compared with the wild type. Under normal greenhouse conditions (18 to 228C, 16 h light), overall plant stature and tiller number are statistically equivalent; however, there is a trend toward a reduction for both in bcd1 mutants as plants near maturity. Additionally, cell death can occur in heat-shocked plants, suggesting a threshold model in the induction of this phenotype. As shown in Supplemental Table 1 online, bcd1 mutant plants treated at 378C for six or more hours and returned to 248C begin to exhibit cell death symptoms within 24 h, becoming more severe thereafter, whereas wild-type (C.I. 16151) plants were unaffected. By contrast, bcd1 mutant plants were identical to wild-type plants when treated at 4, 24 (control), or 308C.
To investigate the degree of interaction between PCD versus R gene-mediated resistance, our experimental design used two barley genotypes, C.I. 16151 (harboring the Mla6 powdery mildew resistance allele) and the bcd1 mutant described above, challenged with the Bgh isolates 5874 (AVR a6 ) and K1 (avr a6 ), respectively ( Figure 1). A time-course expression profile experiment was conducted based on a split-split-plot design described by Caldo et al. (2004) with replications as blocks, Bgh isolate as the whole-plot factor, plant genotype as the split-plot factor, and time as the split-split-plot factor. Fifteen first leaves of inoculated barley seedlings were harvested at 0, 8, 16, 20, 24, and 32 HAI. One Barley1 GeneChip (Close et al., 2004) was used for each of the 72 split-split-plot experiment units corresponding to three replications 3 two isolates 3 two genotypes 3 six time points. The resulting data set was then used for two purposes: (1) transcript-based isolation of the gene(s) responsible for the bcd1 phenotype, and (2) an assessment of cell death-mediated transcriptome reprogramming in both compatible and incompatible interactions in response to challenge with the biotrophic pathogen Bgh. A Six-Gene Deletion Cosegregates with R Gene-Independent, bcd1-Mediated Tip Cell Death It is known that fast-neutron mutagenesis generates deletions (Li et al., 2001); thus, we hypothesized that differential expression in the bcd1 mutant compared with the wild type could be caused by deletion of certain regulator(s), which control a coexpressed gene cluster associated with leaf tip cell death. Hence, to identify candidate genes that may be in such a deletion, we filtered the expression profiling database on two general criteria: (1) probe sets for which natural log expression in the bcd1 mutant was less than three through all six time points, indicative of a lack of measurable expression, and (2) probe sets for which expression was at least 4.5 times greater in wild-type C.I. 16151 than the bcd1 mutant at any one time point, indicative of a functional gene in wild-type plants, but not in the mutant. Twenty-two candidate genes were identified from this analysis (see Supplemental Table  2 online).
Deletion candidates were screened by both genomic PCR and RT-PCR. Six out of the 22 genes, represented by Barley1 probe sets Contig15422_at, Contig12722_s_at, Contig24342_at, Contig8225_at, Contig4201_s_at, and Contig9277_s_at, displayed no amplification in the bcd1 mutant in both assays and thus were concluded to be deleted (Table 1, Figure 3C). Two sets of genetic crosses were performed to test whether (1) the deleted genes cosegregated with the cell death phenotype and 2) bcd1mediated cell death was genetically independent of R genemediated resistance. Specifically, the bcd1 mutant (Mla6/Mla6, bcd1/bcd1) was crossed to (1) the progenitor C.I. 16151 (Mla6/ Mla6, Bcd1/Bcd1) and (2) cv Morex (mla/mla, Bcd1/Bcd1), respectively. Seven-day-old F2 seedlings were inoculated with Bgh 5874 and infection types were recorded another 7 d after inoculation. In cross 1, the ratio of the wild type versus plants displaying the tip cell death phenotype was 42:18, respectively, which fit the expected 3:1 ratio (Table 2). In the second test, 59 plants displayed resistance without cell death, nine plants showed resistance and cell death, 16 plants exhibited susceptibility without cell death, and three plants presented suscepti-bility and cell death, fitting a 9:3:3:1 ratio (Table 2). DNA was extracted from each plant in the two populations and used as template to amplify sequences representative of the six probe sets (Contig15422_at, Contig12722_s_at, Contig24342_at, Contig8225_at, Contig4201_s_at, and Contig9277_s_at). In both populations, except for Contig9277_s_at, no PCR-amplified products were observed from DNA of those plants displaying the cell death phenotype compared with those without cell death ( Figure 4). As the genome sequence of barley is currently unavailable, we used Model Genome Interrogator at PLEXdb (http://www.plexdb.org/modules/MGI/) to derive the most probable genomic positions in rice (Oryza sativa). All five cosegregating candidate genes aligned to the region between Os03g63710 and Os03g63860 on rice chromosome 3, while the gene represented by Contig9277_s_at aligned to rice chromosome 5. Three additional genes were identified from the syntenic region on rice chromosome 3 (www.gramene.org) that were not on the Barley1 GeneChip. These sequences were used to identify their closest barley homologs and were subsequently tested for presence or absence via PCR amplification as described above. One of these genes (represented by GenBank accession number BF267800) was deleted and cosegregated with the cell death phenotype ( Figure 3). Thus, a total of six genes were deleted in the bcd1 mutant and cosegregated the tip cell death phenotype. At the same time, results from cross 2 demonstrated that these six genes segregated independently from the Mla6 R gene, as plants homozygous for the deletion cosegregated with cell death regardless of the expected incompatible or compatible responses with Bgh isolate 5874 (AVR a6 ) ( Figure 4B). Together with the autofluorescence experiments described above in the previous section, these results indicate that bcd1-mediated cell death is genetically separable from Mla6-specified HR.

Genetic and Rice Synteny Mapping Positions the Deleted Region on Barley Chromosome 5H
We used 23,738 expression quantitative trait loci and 1596 transcript-derived markers from the Steptoe 3 Morex (St 3 Mx) doubled haploid mapping population (Potokina et al., 2008b(Potokina et al., , 2008a to genetically map the deleted probe sets to barley chromosome 5H. The region encompassing the bcd1 deletion is highly syntenous with rice as described above, and the Barley1 probe sets identified (using tBLASTn) are ordered correctly within this region based on the transcript-derived marker and expression quantitative trait loci data ( Figure 3). The proximal (left) border of the deletion is gene rich and permits a clear delineation of the breakpoint between Contig8931_at (LOC_Os03g63710) and Contig15422_at /CA031190 (LOC_Os03g63720). The distal (right) border is extremely gene poor, where most annotations of rice genes are retrotransposons of the classes Ty1-copia, Ty3-gypsy, a novel gene potentially encoding a protein with a WRKY domain, and several unclassified genes. The closest proximal gene that could be identified to the deletion breakpoint was Contig10531_at (LOC_Os03g63860) and was found to be present and expressed ( Figure 3). Thus, we could predict the size of the deletion in rice to be between 30 and 70 kb, although barley is expected to have a much larger size due to repetitive elements.

Barley stripe mosaic virus-Induced Silencing of CA031190, Representing RRP46, Causes Tip Cell Death in Barley Leaves
Barley stripe mosaic virus (BSMV)-mediated virus-induced gene silencing (VIGS) was used to assess the functions of the six genes contained within the cosegregating bcd1 deletion (Table  1). Two primer sets per gene (see Methods) were used to amplify fragments for each of the six deleted genes from cDNA of wildtype C.I. 16151 (Bcd1/Bcd1) and inserted into BSMV-VIGS constructs according to Meng et al. (2009  occurred at the third leaf stage, which was used for phenotypic observation (Meng et al., 2009). As shown in Table 1, unigenes Hv21_15423 (represented by GenBank accession number CA031190) and Hv21_15422 (represented by Contig15422_at) both align to the genomic clone FJ652571. CA031190 represents residues 8 to 222 of the annotated open reading frame (ORF), predicted as RRP46 (see next section), while Hv21_15422 represents the 39 end of the ORF and 39 untranslated region. Hence, Contig15422_at was used for determination of transcript accumulation, while both CA031190 and Hv21_15422-derived sequences were used for BSMV-VIGS experiments ( Figure 5A). Plants infected with two constructs, BSMV:CA031190 283 and BSMV:15422 284 , repeatedly developed tip cell death equivalent to the bcd1 mutant ( Figure 5B). Tip cell death equivalent to the bcd1 mutant was not observed in the empty vector (BSMV:00) and mock or in experiments performed with the other five candidate gene constructs. Silencing experiments were repeated at least four times with each replicate exhibiting similar results. An additional three replications of 14 to 24 plants each were performed for BSMV:CA031190 283 and BSMV:15422 284 (example shown in Figure 5B). RNA gel blot analysis showed that the 1.2-kb RRP46 transcript was degraded in silenced plants with the cell death phenotype compared with the mock or BSMV:00 controls ( Figure 5C). These results indicate that the RRP46-derived sequences within the cosegregating deletion in bcd1 mutant plants regulate Bgh-induced tip cell death.  Observed  42  18  59  9  16  3  Expected  45  15  49  16  16  6  Model  3  1  9  3 3 1 x 2 = 0.8, P value = 0.37 x 2 = 6.45, P value = 0.09

RRP46 Encodes a rRNA Processing Protein Conserved among Eukaryotes
To further investigate the gene structure from which CA031190 was transcribed, inverse and RT-PCR were used to recover 59, intron, and 39 sequences. The assembled sequences were further confirmed by amplification using primers flanking the CA031190 ORF using genomic DNA and cDNA template. The newly characterized sequence, Hv RRP46 (GenBank FJ652571), is comprised of seven exons ( Figure 5A), the same number as the Os RRP46 (GenBank accession number AK069107) and Zm RRP46 (GenBank accession number EU964443) orthologs in rice and maize (Zea mays), respectively. The exon-intron borders are conserved as well. In barley, however, the introns are larger and mainly account for the difference in length between Hv RRP46 (>3.5 kb) and Zm RRP46 (2924 bp (Allmang et al., 1999a). As illustrated in Figure 6, a cross-taxa alignment was conducted using BioEdit (http://www.mbio.ncsu.edu/ BioEdit/bioedit.html), demonstrating the conservation of this protein and specific residues among eukaryotes.
Constitutively Overexpressed Genes in the bcd1 Mutant are Associated with Ribosome Biosynthesis Next, we used our time-course expression profiling data to investigate the effect of the RRP46 deletion on cell deathmediated transcriptome reprogramming in both compatible and incompatible interactions in response to challenge by Bgh. The statistical analysis was based on the overall pattern of expression over time, reflecting the kinetics of plant responses to pathogen infection . Within this context, our primary strategy was focused on the identification of genes with time-course expression profiles that differed between the wild type and mutant. As summarized in Figure 7 (detailed data in Supplemental Data Sets 1 to 6 online), three contrasts were made to group genes associated with the bcd1-mediated cell death phenotype as opposed to R gene-mediated responses. The first contrast compared the time-course expression profile of the C.I. 16151 line (containing Mla6) to its fast neutron-derived mutant bcd1, both challenged with Bgh isolate 5874 (AVR a6 ). This comparison yielded those genes that were significantly differentially expressed between C.I. 16151 and bcd1 in response to the avirulent Bgh isolate 5874. Similarly, the second contrast was made between the time-course expression profiles of C.I. 16151 and bcd1 in response to the virulent Bgh isolate K1 (avr a6 ). The third contrast took the average time-course expression profile in response to the two isolates for each  genotype before contrasting C.I. 16151 with bcd1. The intersection of all three contrasts represent probe sets in which the timecourse expression profile differs between the wild type and bcd1 mutant genotypes in a consistent manner for both isolates (Figure 7). A set of P values was obtained for each contrast as part of our mixed linear model analysis of the expression data as described in Methods. The intersection of the three contrasts contained 150 genes with a cutoff P value <0.0001 and an estimated false discovery rate (FDR) <0.6% . With the exception of 12 genes with variable patterns, 124 of these 150 genes were found to be constitutively overexpressed in the bcd1 mutant compared with C.I. 16151; while 14 genes displayed constitutively lower expression (Figure 8). To view these differentially expressed genes in a biological context, Gene Ontology-based functional enrichment analysis was performed using EasyGO (Zhou and Su, 2007). Enrichment for genes associated with the ribosome was found in all three categories of the Gene Ontology (biological process, cellular component, and molecular function) (binomial test, P value = 7.4e-45). Cytoplasmic ribosomal proteins include both the large and small subunits, with no specific bias for either. In addition, predictive functional annotation for the 150 genes was performed by comparison to rice genome alignments, TIGR transcript assemblies, and the Dana-Farber Cancer Institute H. vulgare gene index via the PLEXdb probe set annotation tool (http://www.plexdb.org/modules/PD_general/tools.php) (see Supplemental Data Set 1 online). As shown in Table 3, 68 of the 150 genes had annotations connected to the ribosome. This Gene Ontology and rice alignment-based overrepresentation suggests that the bcd1 mutation influences a regulatory pathway that controls the expression level of these genes (Barakat et al., 2001). Further investigation revealed that if we relaxed the FDR to 5%, all annotated ribosomal genes in barley were overexpressed in the bcd1 mutant. The remaining 26 genes, which had patterns other than constitutive overexpression, did not have significant overrepresentation in any category.

Misprocessing of rRNAs Is Associated with Deletion of RRP46
Depletion of exosome components impacts RNA maturation and RNA quality (Houseley et al., 2006;Chekanova et al., 2007). In yeast, the depletion of Rrp46p protein increases the 39 extended forms of the 5.8S rRNA that extends to the length of the 7S pre-rRNA (Allmang et al., 1999a). To investigate the function of RRP46 in 5.8S rRNA maturation, RNA gel blot hybridization was performed to analyze the 39 extended forms of the 5.8S rRNA. The extended forms can be detected with a probe specific to internal transcribed spacers 2 (Mitchell et al., 1997). Seven-dayold wild-type C.I. 16151 and bcd1 mutant seedlings were challenged with Bgh 5874 and harvested at 0, 24, 48, 72, and 96 HAI for RNA extraction. As shown in Figure 9A, the bcd1 mutant rapidly accumulates 7S pre-rRNA and the 5.8S rRNA extended forms compared with the C.I. 16151 progenitor at all time points in both inoculated and noninoculated plants.
Recently, it has been reported that the depletion of exosome components cause aberrant accumulation of polyadenylated rRNA (Kuai et al., 2004;Chekanova et al., 2007;Lange et al., 2008). To further assess the role of RRP46 in rRNA processing, we analyzed the expression of polyadenylated 18S and 26S rRNA, represented by Barley1 probe sets Contig372_at and Contig711_at, respectively ( Figure 9B). As shown in PLEXdb accession BB46 (this study), we found both polyadenylated rRNAs are significantly accumulated in bcd1 mutants compared with the wild type, in both compatible and incompatible interactions. Accumulation of these rRNAs was equivalent among inoculated and noninoculated wild-type plants (BB10) and within the range observed in 13 stages from seed to seed in barley development (BB3). Taken together, these data indicate that RRP46 in the bcd1 deletion is required for 5.8S rRNA maturation and also plays a role in the control of the level of polyadenylated rRNA. Amino acid sequence alignment of 12 RRP46 proteins from plants (H. vulgare, O. sativa, Z. mays, and Arabidopsis), vertebrates (Homo sapiens and Mus musculus), insects (Anopheles gambiae and Drosophila melanogaster), nematode (C. elegans), yeast (Saccharomyces cerevisiae and Schizosaccharomyces pombe), and fungi (Neurospora crassa) was conducted using BioEdit (http://www.mbio.ncsu.edu/bioedit/bioedit.html). Sequences were derived from the available family information in the PANTHER database (www.pantherdb.org) (Thomas et al., 2003;Mi et al., 2007). The RNase-PH domain is positioned from amino acid residues 14 to 134. Amino acids highlighted in black are completely conserved, whereas those highlighted in gray contain weakly similar residues. Symbols: A colon indicates >50% observed for the most prevalent residue; a period indicates >75% observed for the most prevalent residue; and an asterisk indicates 100% observed for the most prevalent residue. Totals of differentially expressed genes derived from mixed linear model analyses are shown in each cell. Contrasts were based on comparisons between C.I. 16151 (Bcd1) versus bcd1 mutant, each inoculated with Bgh isolates 5874 (AVR a6 ) or K1 (avr a6 ). After specifying a P value cutoff of 0.0001, contrasts 1, 2, and 3 correspond to an FDR of 0.6, 0.2, and 0.09%, respectively. Individual cell summaries for P value, FDR, and natural log expression value per time point are provided in Supplemental Data Sets 1 to 6 online.

Deletion of RRP46 Preconditions Barley Tip Cell Death
Contrasting cell death pathways have been found to have distinct and overlapping levels of interaction. Regardless of the path taken, the final destination is cellular suicide. Unlike lesionmimic mutants that develop cell death constitutively, inoculation with Bgh induces tip cell death in bcd1 mutant plants; silencing of RRP46 contained within the bcd1 deletion generates an equivalent phenotype. Hence, data presented in this report demonstrate that RRP46 within the bcd1 deletion is required for R gene-independent PCD and the normal processing of rRNAs.
Physiological processes appear to be compromised by deficiencies caused by the bcd1 deletion, creating a cellular environment in which the plant is more vulnerable. Challenge by Bgh results in thousands of synchronous penetration attempts, which facilitates extensive collapse of both epidermal and mesophyll cells. This cellular collapse leads to accelerated leaf senescence, as senescence similarly begins from the tip or margins (Lim et al., 2007). A model of accelerated senescence does not account for the accompanying necrosis, which may be exacerbated by the underlying plant-pathogen interactions and resulting signal cascades. Thus, the absence of regulation by RRP46 may also lower the threshold for cell death in pathogen-attacked or presenescing cells (Shirasu and Schulze-Lefert, 2000).
During plant-pathogen interactions, cell death is believed to be associated with R gene-mediated defense and the ensuing HR (Greenberg and Yao, 2004). However, nonspecific induction of gene expression during the early stages of infection is not fundamentally different in incompatible and compatible barley-Bgh interactions (Boyd et al., 1995;Caldo et al., 2004;Caldo et al., 2006). Since bcd1-mediated cell death occurs in both incompatible and compatible responses (Figure 1), it is possible that the initiation of cell death may occur during these early pathogen infection stages. Interestingly, a GCC-box (GCCGCC) is found 10 times in the barley RRP46 promoter compared with 2.05% of all rice genes with 10 or more. The GCC-box is also found in other pathogen-responsive genes and is known to function as an ethylene-and jasmonate-responsive element (Lorenzo et al., 2003;Bouchez et al., 2007). Moreover, the GCCbox is the DNA target of ETHYLENE RESPONSE FACTOR1 (ERF1), which encodes a transcription factor that regulates the expression of pathogen response genes that prevent disease progression (Lorenzo et al., 2003). Hence, a possible scenario might involve the GCC-box as an RRP46 promoter element that is responsive to an ERF1-type transcription factor. In wild-type plants, RRP46 may potentiate a form of R gene-independent cell death, and without RRP46, plants succumb to the stress response generated by Bgh.
Destabilization of the rRNA/Protein Signaling Pathway rRNA, the catalytic component of the ribosomes, accounts for ;80% of the total RNA in the eukaryotic cell (Kampers et al., 1996). The primary transcripts produced from most rRNA genes are extensively processed to yield the mature, functional forms (Lodish et al., 2000). Functional coordination of rRNAs and ribosomal proteins (r-proteins) assemble the ribosome, subsequently translating RNA into protein. Thus, preservation of steady state levels and cellular functions of ribosomes ensures normal metabolism in plants. However, in bcd1 mutant plants, genes encoding known constituents of ribosomes are constitutively expressed compared with its wild-type progenitor (see Supplemental Data Set 1 online; Figure 8). Since sufficient amounts of each r-protein must be produced to allow correct assembly and processing of rRNAs (Laferté et al., 2006;Michels and Hernandez, 2006), the misprocessing caused by the absence of a functional RRP46 would likely result in their overaccumulation (Figure 9). This, in turn, may activate a regulatory pathway that generates an overabundance of r-proteins as opposed to the normal steady state requirement. This destabilizes the rRNA/protein signaling pathway, compromising the plants' normal metabolism and physiological processes (Quirino et al., 1999;Chekanova et al., 2000;Stirpe and Battelli, 2006). Therefore, the overabundance of rRNA and protein alters the plants metabolism, impairing its defense system and facilitating cell death development upon pathogen attack.

Dual Function of RRP46 and Tip Cell Death
RRP46 belongs to the exosome, a conserved multiexonuclease complex that mediates RNA processing and degradation (Mitchell et al., 1997). The exosome core is characterized by a hexameric ring composed of six RNase PH domain-type proteins (RRP41, RRP42, RRP43, RRP45, RRP46, and MTR3) (Mian, 1997). In yeast and human, studies have shown that all the core subunits are vital to the normal activity of exosome to ensure its function (Allmang et al., 1999a(Allmang et al., , 1999bLiu et al., 2006). Recently, Xue (2003, 2006) described the degradeosome, an apoptotic DNA degradation pathway in C. elegans. One of its components, CRN-5, appears to be a functional homolog of RRP46. Silencing of crn-5 generates accumulation of TUNEL-reactive DNA intermediates in apoptotic cells, indicating its importance in apoptosis (Parrish and Xue, 2003). In our study, silencing of RRP46 resulted in a phenotype equivalent to the bcd1 mutant after challenge with Bgh. However, at this time, we have not observed evidence of bcd1dependent DNA laddering (L. Xi and R.P. Wise, unpublished data), suggesting a lack of apoptosis in the generation of the cell death phenotype. More likely, the abundance of constitutively overexpressed ribosomal genes observed in the bcd1 mutant is consistent with the RNA processing function of RRP46. Thus, it is likely that without RRP46, the superfluous rRNA and ribosomal protein compromise the plants' normal metabolism and impairs the defense system, making them more susceptible to stress, resulting in cell death upon pathogen penetration. As cell death and overexpression of ribosomal-related genes were both observed in the bcd1 mutant, we hypothesize that bcd1mediated cell death can be caused either by the direct cell death regulatory function of RRP46 or by the impaired pathways for metabolism and defense due to loss-of-regulation of rRNA/r-protein processing or by the interaction of these perturbed systems.

Isolation of Fast Neutron-Derived Mutant bcd1
The C.I. 16151 line was obtained by introgression of the Mla6 gene into the universal susceptible cv Manchuria (Moseman, 1972). Seeds of C.I. 16151 were treated with fast neutrons at 4 Gy Nf at the International Atomic Energy Agency, Vienna, Austria. M1 seeds were space planted at Figure 9. Effects of RRP46 on rRNA Processing.
(A) RNA gel blot analysis of total RNAs extracted from 7-d-old seedlings of C.I. 16151 and bcd1 mutant. Paired flats were inoculated with Bgh isolate 5874 (AVR a6 ) or noninoculated and harvested at 0, 24, 48, 72, and 96 HAI. Five micrograms of total RNA per sample was fractionated on 8% polyacrylamide gels, and hybridization was performed with a probe specific to 39 extended forms for 5.8S rRNA (GenBank accession number Z11759.1). 5S and 5.8S rRNAs were visualized after ethidium bromide staining (bottom panel).
(B) Time-course expression profiling of polyadenylated 18S rRNA and 26S rRNA in C.I. 16151 and bcd1 mutant. The natural logarithms of signal intensities from PLEXdb accessions BB46 (this study), BB10 (Meng et al., 2009), and BB3 (Druka et al., 2006) were plotted in graphs. Standard errors were calculated based on the mixed linear model used for the analysis.
the USDA-Agricultural Research Service Small Grains Laboratory in Aberdeen, Idaho. Single spikes from each individual M1 plant were harvested to represent the M2 family, which was screened for mutant segregants by sowing intact spikes consisting of 25 to 40 seeds in potting soil following the method of Wise and Ellingboe (1985). When the first leaves were completely unfolded (;10 cm high), plants were inoculated with Bgh isolate 5874 (AVR a6 ) and families were scored for infection type 7 d after inoculation. Families segregating three wild-type to one mutant for either cell death symptoms or sporulating Bgh colonies were identified as putative mutants. Mutant segregants were selected for rescue, advanced to the M3 generation, and then retested for homozygosity with Bgh 5874 as described by Meng et al. (2009). Fast neutron-induced mutant m11542, harboring the bcd1 deletion, was confirmed as a single locus by genetic complementation, DNA gel blot (Halterman et al., 2001), and Barley1 GeneChip  analyses.

Analysis of HR
First leaves were harvested at time points and fixed and cleared in fixation solution (ethanol:acetic acid = 3:1). Hypersensitive responses were examined using a fluorescence microscope (Leitz Fluovert) with a fluorescein filter set.

Microarray Experiment Design
Planting, stage of seedlings, harvesting, and experimental design were part of a larger experiment described by Caldo et al. (2004). Briefly, C.I. 16151 (Mla6/Mla6, Bcd1/Bcd1) and the bcd1 mutant (Mla6/Mla6, bcd1/ bcd1) were planted in separate 20 3 30-cm flats using sterilized potting soil. Each experimental flat consisted of six rows of 15 seedlings, with rows randomly assigned to one of six harvest time points (0,8,16,20,24,and 32 HAI). Seedlings grown to the 1st leaf stage with 2nd leaf unfolded were inoculated with a high density of fresh conidiospores (80 6 20 spores/mm 2 ). Groups of flats were placed at 188C (8 h darkness, 16 h light) in separate controlled growth chambers corresponding to the Bgh isolates. Rows of plants were harvested into liquid nitrogen at the assigned time points. The entire experiment was repeated three times in a standard split-split-plot design with 72 experimental units (Kuehl, 2000).

Transcriptome Analysis
Total RNA was isolated using a hot (608C) phenol/guanidine thiocyanate method as described by Caldo et al. (2004). Probe synthesis, labeling and GeneChip hybridization, washing, staining, and scanning were performed at the Iowa State University GeneChip Core facility. Data processing and normalization were performed according to Caldo et al. (2004). A geneby-gene mixed linear model approach (Wolfinger et al., 2001) was employed to analyze the 22,792 probe sets on the Barley1 GeneChip using the SAS MIXED procedure. Each mixed linear model included genotype, isolate, time, and all their interactions as fixed effects. Random effects were included for replications, replication-by-isolate interactions (corresponding to whole-plot experimental units), replication-by-isolateby-genotype interactions (corresponding to split-plot experimental units), and error terms (corresponding to split-split-plot experimental units).
Three contrasts (described in Results) were tested for each gene as part of each mixed linear model analysis. For each contrast, the set of genes with P values < 0.0001 was estimated to have an FDR <0.7% using the histogram-based method described by Nettleton et al. (2006). The set of 150 genes with P values < 0.0001 for all three contrasts was identified for further analysis (see Supplemental Data Set 1 online). Gene Ontology enrichment was done using the EasyGO tool located at http:// bioinformatics.cau.edu.cn/easygo/. Cluster analysis was performed using the cluster library in R. Average scaled signal intensities for the 150 genes were calculated from three replications and standardized; the dendrogram and heat map were generated using divisive clustering (diana) and heatmap functions in R, respectively.

Data Access
Detailed GeneChip data are publicly available at PLEXdb (http://www. plexdb.org/) under accession number BB46 and at the National Center for Biotechnology Information-Gene Expression Omnibus under accession number GSE14930. Complete cDNA and genomic clones of Hv RRP46 are annotated within GenBank under accession number FJ652571.
Results of the analysis summarized in Figure 7 are provided as Supplemental Data Sets 1 to 6 online.

Genomic PCR and RT-PCR
RNA extraction of C.I. 16151 and the bcd1 mutant was performed as described above. For RT-PCR, single-strand cDNA was synthesized using SuperScript III reverse transcriptase (Invitrogen) using oligo (dT) 20 as a primer. PCR was performed subsequently using Taq DNA polymerase, recombinant (Invitrogen) with specific primers according to each unigene sequence (Table 1; see Supplemental Table 2 online). PCR was also performed with 1 mg genomic DNA of the wild type and the mutant, respectively, using the same sets of primers. Sequences derived from probe sets corresponding to deletion borders (Contig8931_at and Contig10531_at) were used as positive controls (Figure 3; see Supplemental Table 2 online).

Cosegregation Analysis
bcd1 was crossed to the wild-type C.I. 16151 and Morex (C.I. 15773). Seven-day-old seedlings from each F2 population were inoculated with Bgh 5874. Infection phenotypes were scored 7 d after inoculation. A single leaf from each seedling was harvested for DNA extraction. PCR was performed using the population DNA as the template and primers representing the deletion gene candidates (see Supplemental Table 2 online). DNAs of C.I. 16151, the bcd1 mutant, Morex, and Manchuria were used as controls.
Isolation of genomic and cDNA Copies of Hv RRP46 All primers are listed in Supplemental Table 3 online. The 59 end of RRP46 was obtained by inverse PCR performed according to Meng et al. (2007) with minor modifications. One microgram of C.I. 16151 genomic DNA was subjected to overnight digestion with 5 units of PstI followed by self ligation. Subsequently, the self-ligation product was used as template for amplification with ThermalAce DNA polymerase (Invitrogen) using primers RRP46IPp1f and RRP46IPp1r. Reverse primer RRP46IPp1r was used for sequencing the PCR product. RT-PCR was used to retrieve the 39 end of RRP46. Total RNA was extracted from C.I. 16151 plants according to the method of Caldo et al. (2004). First-strand cDNA was synthesized using 2 mg of total RNA, oligo d(T) 20 primer and Superscript reverse transcriptase III (Invitrogen) and subsequently used as the template to amplify the 39 end with AccuPrime High Fidelity Taq DNA Polymerase (Invitrogen) with the RRP46 genespecific primer RRP46IPp1f and a poly-T primer with a linker sequence DTRAND1. One microliter of the resulting PCR product was used as template for a second round of PCR using the linker primer RAND1 and RRP46IPp1f. RRP46IPp1f was used for sequencing the PCR product. New 59 and 39 RRP46 primers (RRP46cdp1f and RRP46cdpr1) were designed from the resulting sequences and used to amplify complete cDNA and genomic clones, which are annotated within GenBank under accession number FJ652571.

BSMV Silencing Constructs
A DNA-based BSMV-VIGS system was used to silence candidate genes in wild-type C.I. 16151 (Meng et al., 2009). Total RNA was extracted from C.I. 16151 plants 20 HAI with Bgh 5874, according to the method of Caldo et al. (2004). First-strand cDNA was synthesized using 2 mg of total RNA, oligo d(T) 20 primer and Superscript reverse transcriptase III (Invitrogen). Primers were designed according to the RRP46 EST sequences, GenBank accession numbers CA031190 and CB881151 (Barley1 Contig15422_at), and are listed in Supplemental Table 3 online. Firststrand cDNA was used as the template to amplify two independent EST sequence fragments of 283 and 284 bp, with introduced PacI and NotI recognition sites at the 59 and 39 ends, respectively. Amplified fragments were ligated in the antisense orientation into the PacI and NotI sites of the BSMV:g vector, with the resulting constructs designated as BSMV: CA031190 283 and BSMV:15422 284 .

Mechanical Infection of BSMV and Powdery Mildew Inoculation
Seven to ten days after bombardment, plants displaying a BSMV infection phenotype (brown streak on the first leaf and chlorotic mosaics on the second leaf) were selected according to Meng et al. (2009). Leaves from infected plants were ground with 2 to 5 volumes of 0.05 M phosphate buffer, pH 7.2, in an ice-cold mortar. Carborundum (0.05 g; Sigma-Aldrich) was added to the buffer for optimal grinding. Seven-day-old healthy barley seedlings were then infected with the appropriate recombinant virions by rubbing the first leaf with crude virus extract four to six times between thumb and index finger, with new gloves used for each construct to prevent contamination.

RNA Gel Blot Analysis
RNA was extracted from silenced plants using a modified guanidium/ acid-phenol protocol. Eight micrograms of total RNA were denatured with DMSO for 1 h at 508C, followed by size fractionation through a 1.2% Seakem GTG agarose (FMC) gel with 0.01 M iodoacetic acid to inhibit RNases. Electrophoresis was performed with 10 mM Na 2 HP0 4 , pH 7, followed by transfer to Hybond XL (Amersham), baked at 808C for 1 h, and cross-linked with 220 mJ of light emitted by 302 nm bulbs in a Stratalinker 2400 (Stratagene). Hybridization was performed as described (Wise et al., 1996). Actin was used as a loading control.
Analysis of RRP46 Function on 5.8S rRNA Maturation C.I. 16151 and the bcd1 mutant were grown in separate 20 3 30-cm flats as described above, and rows of plants were harvested into liquid nitrogen at the assigned time points. RNA was prepared as described by Caldo et al. (2004). Five micrograms of total RNA per sample was size fractionated on 8% polyacrylamide gels containing 8 M urea in 0.53 TBE. RNAs were electroblotted to Hybond-XL membranes (GE Healthcare). The blots were dried and UV cross-linked with a Stratagene 2400. Hybridization was performed in Church buffer (0.5 M NaPO 4 , pH 7.2, 7% SDS, 1 mM EDTA, and 1% BSA) with 20 pmol end-labeled oligonucleotide 59-AAGAGGGTGGTTGGGAGCGT-39. The blots were washed twice with 13 SSPE-0.1% SDS at 428C. Subsequently, the blots were exposed to a storage phosphor screens (GE Healthcare), and screens were scanned by a Typhoon scanner 9410 (GE Healthcare).

Accession Numbers
Complete cDNA and genomic clones of Hv RRP46 are annotated within GenBank accession number FJ652571. Additional cDNA sequences representing Hv RRP46 can be found under GenBank accessions numbers CA031190 and CB881151.

Supplemental Data
The following materials are available in the online version of this article.