Protein Polyubiquitination Plays a Role in Basal Host Resistance of Barley^,

Abstract

To study protein ubiquitination pathways in the interaction of barley (Hordeum vulgare) with the powdery mildew fungus (Blumeria graminis), we measured protein turnover and performed transient-induced gene silencing (TIGS) of ubiquitin and 26S proteasome subunit encoding genes in epidermal cells. Attack by B. graminis hyperdestabilized a novel unstable green fluorescent protein fusion that contains a destabilization domain of a putative barley 1-aminocyclopropane-1-carboxylate synthase, suggesting enhanced protein turnover. Partial depletion of cellular ubiquitin levels by TIGS induced extreme susceptibility of transformed cells toward the appropriate host pathogen B. graminis f. sp hordei, whereas papilla-based resistance to the nonhost pathogen B. graminis f. sp tritici and host resistance mediated by the mlo gene (for mildew resistance locus O) remained unaffected. Cells were rescued from TIGS-induced ubiquitin depletion by synthetic genes encoding wild-type or mutant barley monoubiquitin proteins. The strongest rescue was from a gene encoding a K63R mutant form of ubiquitin blocked in several ubiquitination pathways while still allowing Lys-48–dependent polyubiquitination required for proteasomal protein degradation. Systematic RNA interference of 40 genes encoding all 17 subunits of the proteasome 19S regulatory particle failed to induce hypersusceptibility against B. graminis f. sp hordei. This suggests a role for Lys-48–linked protein polyubiquitination, which is independent from the proteasome pathway, in basal host defense of barley.

INTRODUCTION

Protein ubiquitination regulates many processes in eukaryotic cells, ranging from cell division through communication/signaling to cell death. There are three main types of ubiquitin signals: (1) monoubiquitination that provides signals for endosomal sorting, lysosomal degradation, DNA repair, histone regulation, and nuclear export; (2) Lys-63–linked polyubiquitination that is involved in DNA repair, lysosomal degradation, and protein activation; and (3) Lys-48–linked polyubiquitination that is essential for targeting proteins to the 26S proteasome for degradation (Smalle and Vierstra, 2004; Haglund and Dikic, 2005). In plants, the ubiquitin/proteasome pathway of protein degradation has been implicated in plant responses to internal and external stimuli, including phytohormones, abiotic stress, and pathogen attack. Experimental evidence is derived mainly from gene regulation studies of pathway components, from pharmacological experiments using pathway inhibitors, and from mutant plants (Vierstra, 2003). By contrast, very little is known about the monoubiquitination and Lys-63–linked polyubiquitination pathways in plants.

The molecular structure of the 26S proteasome has been well characterized and can be separated into a 20S catalytic core plus a 19S regulatory particle (RP) (Smalle and Vierstra, 2004). The RP can be subdivided further into a base consisting of six AAA-ATPase (RPT1-6) plus two non-ATPase (RPN1-2) subunits and a lid consisting of nine non-ATPase subunits (RPN3-12). While the core particle is responsible for enzymatic protein degradation, the RP is required for recognition and docking of ubiquitinated protein complexes, protein unfolding, and presumably regulation of substrate entry into the core particle.

Several lines of evidence suggest that the ubiquitin/proteasome pathway plays a role in pathogen-attacked plants. First, several genes encoding pathway components like E2 or RING E3 ligases have been found to be transcriptionally upregulated by pathogen-derived elicitors (Dahan et al., 2001; Takai et al., 2002). Second, the RPM1 (for resistance to Pseudomonas syringae pv maculicola 1) resistance protein in Arabidopsis thaliana has been found to become destabilized upon pathogen attack, suggesting enhanced proteasome-mediated turnover (Boyes et al., 1998). There is also evidence for co-option of the ubiquitin/proteasome pathway by pathogens like Agrobacterium tumefaciens that inject an F-box protein (VirF) or P. syringae that either injects or recruits intrinsic E3 ligases in(to) susceptible host cells (Schrammeijer et al., 2001; Abramovitch et al., 2006; Janjusevic et al., 2006). Finally, the F-box protein SON1 and the U-box protein Spl11 represent suppressors of resistance and cell death, indicating the existence of negative regulation of pathogen defense by the ubiquitin/proteasome pathway (Kim and Delaney, 2002; Zeng et al., 2004).

The pathosystem of barley (Hordeum vulgare) attacked by the barley powdery mildew fungus Blumeria graminis f. sp hordei (Bgh) represents one of the best studied model systems of plant–fungus interactions as well as being of high economic importance. The restriction of the Bgh interaction to the epidermal mono-cell layer of attacked shoot tissue makes it ideally suited for detailed cytological, biochemical, and molecular analysis. Two basic types of host and nonhost defense responses have been described: (1) the papilla-based localized response and (2) the hypersensitive response (Hückelhoven et al., 1999; Collins et al., 2002; Panstruga and Schulze-Lefert, 2002). It is generally accepted that the localized response is a hallmark for a race-nonspecific, durable, and sometimes quantitative type of resistance, whereas the hypersensitive response is typical for race-specific, nondurable resistance mediated by major resistance (R) genes like the Mla (for mildew-resistance locus A) allelic series of barley. It is interesting to note that several barley R proteins, including Mla, require Rar1 (for required for Mla-mediated resistance 1) that by itself interacts with SGT1 (for suppressor of G-two allele of Skp1), an SCF-E3 activator (Shirasu et al., 1999; Azevedo et al., 2002). Rar1 was recently described as stabilizing specific allele products of the barley Mla gene conferring resistance against the barley powdery mildew Bgh (Bieri et al., 2004). These data provide evidence that regulation of protein turnover, possibly through the ubiquitin/proteasome pathway, is also important for effective defense against fungal attack in the barley–Bgh interaction.

Here, we studied the role of protein (poly)ubiquitination pathways in basal host defense against Bgh and in nonhost resistance against the wheat powdery mildew B. graminis f. sp tritici (Bgt). The starting point for this study was a screen based on the recently established transient-induced gene silencing system (TIGS) as a high-throughput phenomics tool to address gene function in epidermal cells of barley interacting with powdery mildew fungi (Douchkov et al., 2005).

RESULTS

TIGS of Candidate Genes for Basal Host Resistance

A functional screen for a possible role of 389 pathogen-upregulated barley genes in basal host defense revealed a number of putative unigenes that either increased or reduced the number of fungal haustoria in epidermal cells upon TIGS. Five out of 16 RNA interference (RNAi) constructs that enhanced cell susceptibility by at least twofold were targeted against different polyubiquitin genes of barley (pIPKTA30_HO13K06, pIPKTA30_HO10N01, pIPKTA30_HY04B09, pIPKTA30_HO10A17, and pIPKTA30_HO12M17; for EST sequences and accession numbers, see http://pgrc.ipk-gatersleben.de/est/). This discovery prompted us to study the role of protein ubiquitination in basal host defense in greater detail.

Protein Turnover in Pathogen-Attacked Barley Epidermis

Ubiquitination is essential for protein turnover and degradation. Therefore, we first addressed the question whether pathogen attack influences protein turnover in barley epidermal cells. To develop a sensitive reporter for protein turnover, which might be regulated by powdery mildew attack, we fused the region encoding the N-terminal part of a putative barley 1-aminocyclopropane-1-carboxylate synthase (ACS) to the start of the open reading frame of green fluorescent protein (GFP) (Figure 1). The plant ACS protein has been shown to be rapidly turned over, which probably reflects the need of precise adjustment of metabolic flux leading to transient accumulation of the plant hormone ethylene as shown in powdery mildew–attacked wheat leaves (Kim and Yang, 1992; Ulrichs and Gabler, 2004). Recently, an N-terminal domain of 93 amino acids has been identified to confer protein instability when fused to a dihydrofolate reductase reporter gene (Schlogelhofer and Bachmair, 2002). Two different ACS:GFP fusion constructs were transiently coexpressed together with pBC17 that gave rise to anthocyanin accumulation and served as internal control in barley epidermal cells by biolistic bombardment, as described earlier (Schweizer et al., 2000). In contrast with GFP that gave rise to green fluorescing cells over a period of at least 6 d after bombardment, the ACS:GFP fusion proteins first fluoresced brightly but became nondetectable at ∼3 d after bombardment (Figure 2). In all cases, cobombarded pBC17 gave rise to visible anthocyanin accumulation from 3 d after bombardment onwards (Figures 2D, 2H, and 2L), indicating that the ACS:GFP fusion protein had no cytotoxic effect. We therefore conclude that the transient ACS:GFP fluorescence reflects the transient nature of gene expression from introduced plasmid DNA, together with enhanced ACS:GFP protein turnover. The instability domain of ACS was also recognized in wheat (Triticum aestivum), maize (Zea mays), and tobacco (Nicotiana tabacum) (Table 1, Figure 3).

Figure 1.

Constructs for Transient Overexpression or TIGS in Barley and Wheat.

All constructs contain the 35S promoter (P) and terminator (T). Overexpression constructs based on pIPKTA9 (see Supplemental Figure 1 online) were obtained by conventional ligation, whereas TIGS constructs based on pIPKTA30N were obtained by a combined ligation/recombination method (Douchkov et al., 2005). The polyubiquitin sequence targeted by TIGS constructs pIPKTA30_Ubi_long and pIPKTA30_Ubi_short is indicated by brackets at the bottom of the figure (only one of the inverted repeats is shown). Ubi, ubiquitin; B1 and B2, LR clonase attachment sites; IR1 and IR2, inverted-repeat sequences of target genes; Ubi-wobble_x, synthetic ubiquitin genes (used for RNAi rescue experiments) saturated with silent mutations encoding the identical monoubiquitin unit as barley cDNA clone HO13K06; x, either wild-type or K48R or K63R mutant forms; CaMV, Cauliflower mosaic virus; aa, amino acids.

Figure 2.

An N-terminal Domain of Barley ACS Destabilizes GFP in Barley Epidermis.

Barley leaf segments were cobombarded with pIPKTA25 (113 amino acids; ACS:GFP overexpression [OE]), pIPKTA26 (164 amino acids; ACS:GFP OE), or pGFP encoding an enhanced GFP, together with pBC17, giving rise to anthocyanin accumulation 3 to 6 d after bombardment. Images from identical sites on bombarded leaves were repeatedly taken at the times indicated. Note the high degree of spatial congruence of initial GFP fluorescence with pBC17-induced anthocyanin accumulation.

Table 1.

New Unstable GFP Reporter Genes for Plants Based on a Destabilizing Domain of the Barley ACS Gene

		Number of Fluorescent Cells
Plant	Time (h)a	pGFP	pIPKTA25	pIPKTA26
Barley	24	750	670	487
Barley	48	810	452	283
Barley	72	868	272	180
Maize	24	488	149	224
Maize	48	398	72	91
Maize	72	326	15	7
Tobacco	24	2245	2115	842
Tobacco	48	1909	806	168
Tobacco	72	1322	56	11

^aLeaf segments were bombarded with different GFP constructs, and the number of GFP fluorescing cells was repeatedly counted on the same leaf segments 24, 48, and 72 h after bombardment.

Figure 3.

The Unstable ACS:GFP Protein Encoded by pIPKTA25 Is Hyperdestabilized by Pathogen Attack in Leaf Epidermis.

Barley or wheat leaf segments were cobombarded with a mixture of pIPKTA25 (113 amino acids; ACS:GFP OE) and pUbiGUS, followed by challenge inoculation with Bgt or Bgh. At the times indicated after bombardment, the number of GFP-fluorescing cells was counted, followed by GUS staining and counting of GUS-stained cells, which was used for internal normalization. High ratios of GFP:GUS in control leaves reflect higher efficiency of GFP as reporter gene in transient expression experiments. The data represent the mean of two independent experiments. Bars = range.

To test whether pathogen attack influenced turnover of the ACS:GFP fusion protein, bombarded barley and wheat leaf segments were challenged with Bgh or Bgt 4 h after cobombardment with pIPKTA25 and pUbiGUS (Schweizer et al., 1999). This resulted in a reciprocal pair of two (non)host interactions. As shown in Figure 3, fungal attack dramatically hyperdestabilized the ACS:GFP protein, which was reflected by a decreased GFP-to-β-glucuronidase (GUS) ratio of visibly expressing cells. This effect was seen in all four tested interactions, irrespectively of whether they were susceptible or characterized by papilla-based resistance (barley/Bgt; see also Table 5) or by a hypersensitive response (wheat/Bgh; data not shown). The kinetics of GFP fluorescence was compared between GFP and ACS:GFP in noninoculated control and in pathogen-attacked cells (Figure 4). The number of GFP-expressing cells followed a parallel pattern when noninoculated controls were compared with Bgh-attacked leaf segments. The smaller number of GFP-fluorescing cells upon Bgh attack probably reflects some degree of cell death induced by the combination of bombardment and pathogen, as reflected by an increased number of autofluorescing cells (data not shown). By contrast with GFP, the number of ACS:GFP-fluorescing cells in Bgh-attacked leaf segments followed another kinetics and clearly declined more rapidly than without pathogen attack.

Table 5.

TIGS of Polyubiquitin Genes Does Not Affect Papilla Formation in Nonhost- and mlo-Resistant Cells

Construct	Mildew	Plant	Papilla (%)a	Int.b
Nonec	Bgh (host)	Golden Promise	86.3d	627
None	Bgt (nonhost)	Golden Promise	95.0	581
pIPKTA30N	Bgt (nonhost)	Golden Promise	57.0e	114
pIPKTA30_Ubi_short	Bgt (nonhost)	Golden Promise	53.0e	134
pIPKTA30N	Bgh (host)	Ingrid BC mlo5	70.0e	247
pIPKTA30_Ubi_short	Bgh (host)	Ingrid BC mlo5	65.0e	203

^aPercentage of papilla formation underneath appressorial lobes of Bgh or Bgt.

^bNumber of observed interaction sites.

^cNonbombarded leaves.

^dA majority of these papillae are ineffective, not preventing penetration.

^eMean values of two independent, reinvestigated experiments shown in Tables 3 and 4. For internal controls of RNAi efficiency, please refer to Tables 3 and 4.

Figure 4.

Time Course of the Number of GFP-Expressing Cells in Barley Epidermis.

Barley leaf segments were bombarded with pIPKTA25 (113 amino acids; ACS:GFP OE), followed by challenge inoculation with Bgh 4 h after bombardment. The data represent the mean of two parallel bombardments. Bars = range.

Lys-48–Linked Protein Polyubiquitination Is Important for Basal Resistance in Pathogen-Attacked Barley

To study the role of protein ubiquitination in basal defense mechanisms of pathogen-attacked barley in greater detail, we depleted epidermal cells from ubiquitin to different extent using RNAi constructs targeting a polyubiquitin gene(s) represented by EST clone HO13K06 (Figure 1). Although barley cv Golden Promise is susceptible to Bgh, it possesses basal resistance preventing unlimited fungal ingress. This basal resistance can be enhanced by inducers of resistance, and we observed that it was also enhanced by microprojectile bombardment, resulting in ∼80 to 90% resistance. Figure 5A shows that TIGS of ubiquitin caused severe cell death reflected by strongly reduced numbers of GUS-expressing cells when the RNAi construct RNAi_Ubi_long containing a longer inverted-repeat sequence was used (Figure 1). This result was not unexpected because the targeted gene, which is represented by EST clone HO13K06, shows high homology (90 to 97% sequence identity) to 20 tentative unigene consensi of the TIGR barley gene index. Therefore, cross-silencing of many polyubiquitin genes by RNAi_Ubi_long is likely. The result also demonstrates the efficiency of TIGS in barley.

Figure 5.

Lys-48–Linked Protein Polyubiquitination Is Important for Basal Resistance in Barley Epidermal Cells.

(A) and (B) Mean values ± se of six independent experiments. Different letters indicate significantly different values based on one-way analysis of variance. Control, pUbiGUS plus pIPKTA9 plus pIPKTA30N (empty vectors); RNAi_Ubi_long, RNAi construct pIPKTA30_Ubi_long; RNAi_Ubi_short, RNAi construct pIPKTA30_Ubi_short; Ubi_wobble_WT; RNAi rescue construct pIPKTA9-Ubi_wobble_WT; Ubi_wobble_K48R; RNAi rescue construct pIPKTA9-Ubi_wobble_K48R; Ubi_wobble_K63R, RNAi rescue construct pIPKTA9-Ubi_wobble_K63R (for details, see Figure 1).

(A) Barley leaf segments were cobombarded with a mixture of pUbiGUS, ubiquitin RNAi construct, and RNAi rescue construct containing a synthetic Ubi_wobble gene (either wild type, K48R mutant, or K63R mutant). Three days after bombardment, the number of GUS-expressing cells was counted.

(B) Barley leaf segments were cobombarded with a mixture of pUbiGUS, ubiquitin RNAi construct, and RNAi rescue construct containing a synthetic Ubi_wobble gene (wild type, K48R mutant, or K63R mutant). Three days after bombardment, leaves were challenge inoculated with Bgh, followed by analysis of the haustorium index 40 h after inoculation (number of haustoria in GUS-expressing cells relative to the number of GUS-expressing cells). Each data column is based on an average of 1786.8 counted GUS-expressing cells and on 500.2 observed haustoria.

To obtain more specific information about different ubiquitination pathways, we developed an RNAi rescue approach based on complementation of ubiquitin-depleted cells by cobombarded synthetic genes encoding the identical ubiquitin monomer, which has been targeted by the RNAi constructs but is immune against the RNAi effect due to saturated silent point mutations (Figure 1; see Supplemental Figure 2 online). This approach allowed for testing the effect of wild-type ubiquitin plus dominant-negative K48R and K63R mutant forms that have been shown to specifically block Lys-48- and Lys-63–dependent pathways, respectively (Spence et al., 2000; Kolodziejski et al., 2002; Fujimuro et al., 2005). Addition of the RNAi rescue construct Ubi_wobble_WT partially restored GUS cell numbers, suggesting that normal ubiquitin levels were only partly restored by the transient coexpression of a monoubiquitin gene. RNAi rescue by K48R and K63R mutant forms of ubiquitin also partially restored GUS cell numbers. Therefore, the observed cell number effect by depletion of ubiquitin was probably the result of multiple impaired ubiquitination pathways. A second RNAi construct, RNAi_Ubi_short, containing shorter inverted repeats targeting the same ubiquitin gene(s) had a mild cell number effect, suggesting partial depletion. All tested RNAi rescue constructs appeared to fully restore GUS cell numbers, although differences were statistically not significant due to considerable variation between independent experiments (Figure 5A, right group of columns).

Partial depletion of ubiquitin by the construct RNAi_Ubi_short caused a dramatic increase in successful penetration by Bgh (haustorium index; for details, see Douchkov et al., 2005), indicating that protein ubiquitination is essential for basal resistance of barley (Figure 5B). RNAi rescue by the synthetic genes encoding wild-type or mutated ubiquitin units partially restored basal resistance. Interestingly, the mutant K63R protein produced the strongest effect, suggesting that its inaccessibility for Lys-63–linked polyubiquitination allowed more efficient complementation of the remaining ubiquitination pathways by a limiting number of monoubiquitin molecules in transiently expressing cells. The effect of the K63R mutant protein was significantly stronger compared with wild-type ubiquitin by one-way analysis of variance. A pairwise comparison by Student's t test also revealed a significant difference between K48R and K63R mutant proteins (P = 0.02). This gain of efficiency was not observed using the K48R mutant of ubiquitin for RNAi rescue. Theoretically, the observed difference in complementation strength of the two mutant forms of monoubiquitin could have been due to different protein stability or other trivial reasons. However, the very similar rescue effect of these mutants on GUS cell numbers argues against this possibility (Figure 5A). In summary, it appears likely that the Lys-48–linked polyubiquitination of proteins was more important for basal defense in barley than polyubiquitination by Lys-63–linked units.

We also tested the effect of transient ubiquitin overexpression on basal resistance by bombarding leaves with the construct pIPKTA9_Ubi (Figure 1, Table 2). Clearly, ubiquitin overexpression had no effect on haustorium index, demonstrating that under normal (nonsilenced) conditions, cellular ubiquitin levels were saturated. This result also strongly suggests that the effect observed with the RNAi rescue constructs was indeed due to complementation of ubiquitin-depleted cells and not due to overexpression.

Table 2.

Transient Overexpression of a Polyubiquitin Gene Has No Effect on Basal Host Resistance against Bgh

RNAi Construct	Time Intervala	HI (%)b
pIPKTA9	4 h	10.5 (8.3–12.7)
pIPKTA9_Ubi	4 h	9.9 (7.3–12.5)
pIPKTA9	24 h	5.2 (3.6–6.8)
pIPKTA9_Ubi	24 h	4.6 (4.3–4.9)

^aTime between bombardment and challenge inoculation.

^bHaustorium index (number of haustoria in GUS-expressing cells relative to the number of GUS-expressing cells). Mean (range) of two parallel bombardments. A second independent experiment produced very similar data.

TIGS of Polyubiquitin Genes Has No Effect on mlo-Mediated and Nonhost Resistance

In sharp contrast with its effect on basal host resistance, no effect of ubiquitin depletion was observed in the nonhost interaction with Bgt, although the positive control construct directed against Hv SNAP34 (for synaptosome-accociated protein of 34 kD) that encodes a tSNARE protein did break this type of resistance in the same set of experiments (Table 3; Douchkov et al., 2005). To test if the absence of Bgt haustoria in ubiquitin-depleted, GUS-expressing cells resulted from the inability of these cells to support fungal growth due to stress overkill by particle bombardment, ubiquitin depletion, and pathogen challenge, we looked at their capacity to raise the normal papilla response. As demonstrated in Figure 6 and Table 5, the cells expressing pIPKTA30_Ubi_short were fully normal in their papilla-based response to Bgt. We also found that TIGS of polyubiquitin genes in barley had no effect on mlo-mediated (for mildew resistance locus O) resistance (Table 4). As was the case in the nonhost-resistant interaction, papilla formation of ubiquitin-depleted, mlo-resistant cells appeared to be normal (Table 5). Together, these data suggest that TIGS of ubiquitin almost completely disarmed host cells with respect to basal resistance, while their competence for nonhost- or mlo-mediated resistance remained unaffected.

Table 3.

TIGS of Polyubiquitin Genes Does Not Affect Nonhost Resistance in Barley Epidermal Cells against Bgt

RNAi Constructa	Series	SI (%)b	P (t Test)	nc
pIPKTA30N	1	0.46 ± 0.32		3
pIPKTA30_Ubi_long	1	2.17 ± 1.10	0.100	3
pIPKTA30_SNAP34	1	7.62 ± 1.24	0.003	3
pIPKTA30N	2	0.098 ± 0.098		4
pIPKTA30_Ubi_short	2	0.59 ± 0.33	0.100	4
pIPKTA30_SNAP34	2	5.60 ± 2.24	0.025	4

^aFor constructs, see Figure 1.

^bSusceptibility index (percentage of susceptible GUS-expressing cells bearing at least one haustorium). Mean values ± se.

^cNumber of independent experiments.

Figure 6.

TIGS of Polyubiquitin Genes Does Not Affect Papilla-Based Nonhost Resistance in Barley Epidermal Cells.

Both a transformed cell coexpressing pUbiGUS and pIPKTA30_Ubi_short and a neighboring, nontransformed cell mounted a normal papilla response at the site of attempted penetration by Bgt (arrows). Bars = 25 μm.

Table 4.

TIGS of Polyubiquitin Genes Does Not Affect mlo-Mediated Resistance in Barley Epidermal Cells against Bgh

Experiment	Plant	Constructa	HI (%)b
1	Ingrid	pIPKTA30	13.5
1	Ingrid	pIPKTA30_Ubi_short	NAc
1	Ingrid BC mlo5	pIPKTA30	0.0
1	Ingrid BC mlo5	pIPKTA30_Ubi_short	0.0
2	Ingrid	pIPKTA30	5.4
2	Ingrid	pIPKTA30_Ubi_short	17.6
2	Ingrid BC mlo5	pIPKTA30	0.0
2	Ingrid BC mlo5	pIPKTA30_Ubi_short	0.0
3	Ingrid	pIPKTA30	5.2
3	Ingrid	pIPKTA30_Ubi_short	9.7
3	Ingrid BC mlo5	pIPKTA30	0.0
3	Ingrid BC mlo5	pIPKTA30_Ubi_short	0.0

^aFor constructs, see Figure 1.

^bHaustorium index (number of haustoria in GUS-expressing cells relative to the number of GUS-expressing cells).

^cNA, not analyzed.

A Unigene Set of the 19S RP of the 26S Proteasome in Barley

Lys-48–linked polyubiquitination is known to target protein substrates to the 26S protesome. Therefore, the results obtained so far suggested an essential role of the ubiquitin/proteasome pathway in basal host resistance. To test the relevance of proteasome-mediated protein turnover or degradation in basal host resistance, a nonredundant unigene set of barley encoding the well-characterized set of RP components of the 26S proteasome was calculated based on ∼140,000 ESTs of the IPK (http://pgrc.ipk-gatersleben.de/cr-est), and RNAi constructs directed against this set were produced (see Supplemental Table 1 online). In total, 42 barley unigenes encoding the 17 RP components were identified (see Supplemental Table 2 online). The number of identified, homologous unigenes per RP component ranged from one to five, and each unigene was represented by one to up to 46 EST clones. In general, the genetic redundancy (number of gene homologs) encoding the AAA-ATPase base components appeared to be lower compared with non-ATPase subunits. For a comparison, all RP components of Arabidopsis appear to be encoded by single-copy or pairs of nearly identical, duplicated genes (see http://www.arabidopsis.org/).

TIGS of Proteasome Components in Pathogen-Attacked Barley Epidermis

First, we tested a number of RNAi constructs directed against RP base components, which show a low genetic redundancy, for stabilization of ACS:GFP in the presence of Bgt in a cobombardment experiment with pBC17 that gives rise to anthocyanin accumulation (Table 6). As observed previously, the number of cells with visible ACS:GFP expression dropped upon inoculation, whereas the number of anthocyanin-accumulating cells was not significantly affected. In the presence of any of the four selected RNAi constructs directed against Hv RPN1a, Hv RPT1a, Hv RPT3a, or Hv RPT6a, a partial restoration of ACS:GFP fluorescence in Bgt-attacked cells was observed. Moreover, these RNAi constructs decreased the number of anthocyanin-accumulating cells, indicating some cytotoxic effect. As a result, the ratio of ACS:GFP to anthocyanin-expressing cells was clearly increased by the RNAi constructs against 26S proteasome components surpassing the number obtained in nonattacked cells bombarded with the pIPKTA30 empty-vector control.

Table 6.

TIGS of Components of the 26S Proteasome RP Stabilizes an Unstable ACS:GFP Fusion Protein in Pathogen-Attacked Barley Cells

Construct	EST Clone	Bgta	ACS:GFPb	Anthoc.c	ACS:GFP/Anthd
pIPKTA30N (empty vector)	None	−	109.6 ± 24.5	109.4 ± 16.8	1.03 ± 0.25
pIPKTA30N (empty vector)	None	+	19.4 ± 4.3	67.4 ± 25.7	0.45 ± 0.14
pIPKTA30N_HvRPN1a	HX02I23	+	37.0 ± 24.0	15.5 ± 5.5	2.10 ± 0.80
pIPKTA30N_HvRPT1a	HO12F06	+	44.0 ± 33.0	26.0 ± 13.0	1.41 ± 0.56
pIPKTA30N_HvRPT3a	HW03E18	+	54.0 ± 2.0	30.0 ± 11.0	2.11 ± 0.84
pIPKTA30N_HvRPT6a	HD01P16	+	42.5 ± 7.5	11.0 ± 3.0	3.98 ± 0.41
Mean (all four RP genes)			44.4 ± 3.5	20.6 ± 4.4	2.40 ± 0.55
P (t test; all four RP genes)e			0.008	0.095	0.008

^aChallenge inoculation with Bgt 4 h after bombardment.

^bNumber of cells expressing the ACS:GFP fusion construct (mean ± se).

^cNumber of cells expressing the cobombarded construct pBC17 giving rise to anthocyanin accumulation (mean ± se).

^dNumber of cells expressing ACS:GFP normalized to the number of anthocyanin-accumulating cells (mean ± se).

^eStatistical significance was tested against Bgt-challenged empty-vector controls.

TIGS of polyubiquitin genes had an effect on basal host resistance only. Therefore, in a systematic RNAi approach, we tested all 40 RP subunit–encoding genes for which RNAi constructs could be obtained for their function in basal host resistance. Sixteen RNAi constructs caused at least a twofold effect compared with the empty-vector control (data not shown). The corresponding candidate genes were silenced in four more independent experiments together with the four constructs that had a stabilizing effect on ACS:GFP (Table 6). As shown in Table 7, none of the RNAi constructs targeting 26S proteasome components caused an effect that was comparable to the one obtained by TIGS of polyubiquitin. A positive control plasmid for RNAi efficiency (pIPKTA36) targeting the barley Mlo gene and resulting in a phenocopy of mlo-mediated resistance was active in the corresponding set of experiments (Douchkov et al., 2005). There was even a trend toward enhanced resistance of cells with silenced genes encoding proteasome components. For some of the targeted subunits that are encoded by up to five homologous genes according to the EST-clustering data, the absence of RNAi-induced interaction phenotypes could have been due to genetic redundancy. However, four RNAi constructs (pIPKTA30_HvRPN1a, pIPKTA30_HvRPT1a, pIPKTA30_HvRPT3a, and pIPKTA30_HvRPT6a) directed against RP base subunits, which apparently are encoded by only one or two genes (the targeted one being the most highly expressed), could not reproduce the ubiquitin TIGS effect either, although they protected the ACS:GFP fusion protein from pathogen-induced degradation. The low genetic redundancy of these four proteins was also confirmed by a comparison to the rice (Oryza sativa) genome sequence where two gene paralogs were usually identified by BLASTN analysis (http://www.tigr.org/tdb/e2k1/osa1/). Taken together, these results indicate that a ubiquitin-dependent mechanism other than proteasome-mediated protein degradation is essential for basal host resistance of barley against Bgh.

Table 7.

TIGS of Genes Encoding Subunits of the 26S Proteasome RP Does Not Compromise Basal Resistance against Bgh

RNAi Construct	EST Clone	Relative HI (%)a	nb	P (t Test)c
pIPKTA30N	None	100	5
pIPKTA30N_HvRPT2a	HF14O23	141.3 ± 35.7	5	0.1550
pIPKTA30N_HvRPN5a	HA06K14	139.6 ± 30.9	5	0.1340
pIPKTA30N_HvRPT3ad	HW03E18	114.8 ± 35.2	5	0.3470
pIPKTA30N_HvRPT6ad	HD01P16	104.5 ± 34.7	5	0.4510
pIPKTA30N_HvRPT2b	HD08L01	98.8 ± 6.1	4	0.4250
pIPKTA30N_HvRPN6b	HD07C11	96.7 ± 10.0	4	0.3820
pIPKTA30N_HvRPN9a	HH02K23	93.1 ± 18.0	5	0.3610
pIPKTA30N_HvRPN1a	HI10F03	90.3 ± 8.0	4	0.1540
pIPKTA30N_HvRPN3c	HX05P15	89.4 ± 15.0	5	0.2600
pIPKTA30N_HvRPT6a	HD05P11	89.4 ± 12.4	4	0.2270
pIPKTA30N_HvRPN2d	HZ61F13	89.1 ± 12.7	5	0.2200
pIPKTA30N_HvRPN1ad	HX02I23	86.4 ± 24.7	5	0.3050
pIPKTA30N_HvRPN2e	HT01B14	84.5 ± 21.1	5	0.2510
pIPKTA30N_HvRPT1ad	HO12F06	82.7 ± 25.0	4	0.2690
pIPKTA30N_HvRPN11c	HZ40O15	75.9 ± 2.0	4	0.0006
pIPKTA30N_HvRPN9d	HS03L14	75.6 ± 11.6	4	0.0630
pIPKTA30N_HvRPN12a	HT06M12	75.4 ± 16.5	5	0.1040
pIPKTA30N_HvRPN11b	HM10M03	74.7 ± 11.6	5	0.0470
pIPKTA30N_HvRPT4a	HI05K17	72.2 ± 11.9	4	0.0500
pIPKTA36 (Mlo RNAi)	HO33B01	48.6 ± 11.2	4	0.0090

^aThe haustorium index (number of haustoria in GUS-expressing cells relative to the number of GUS-expressing cells) was calculated relative to the internal empty-vector control (piPKTA30N, set to 100%). Mean ± se of n independent bombardments.

^bNumber of independent bombardments (= independent inoculation experiments).

^cOne-sample t test against hypothetical value 100 (one-sided).

^dConstructs highlighted in bold did stabilize ACS:GFP in pathogen-attacked cells.

DISCUSSION

The role of protein ubiquitination pathways in basal host and nonhost resistance was studied by employing a high-throughput gene-silencing approach in the pathosystem of barley interacting with powdery mildew. To measure protein turnover and monitor the effect of different RNAi constructs, we designed unstable GFP proteins as the potential targets for ubiquitin-mediated protein degradation by fusing a previously identified destabilizing domain of the ACS protein to the N terminus of GFP. Indeed, the ACS:GFP fusion protein became hyperdestabilized in pathogen-attacked wheat and barley epidermis, suggesting enhanced protein turnover. The assumption that the 26S proteasome pathway was involved in enhanced protein turnover was confirmed by RNAi-mediated gene silencing of several components of the 26S proteasome RP, which caused stabilization of the ACS:GFP protein. Protein hyperdestabilization was found to the same extent in both susceptible and resistant interactions, suggesting that it is a rather nonspecific stress symptom, which reflects the need of proteome reorganization for basal and nonhost defense and/or for supporting the invading fungus. However, it cannot be excluded that the observed ACS:GFP protein hyperdestabilization was specific for the ACS fusion partner in the context of pathogen-induced, transient ethylene biosynthesis (Ulrichs and Gabler, 2004) requiring tight regulation of the key biosynthetic enzyme(s). Therefore, no general conclusion from these data as to the role of the ubiquitin/proteasome pathway in pathogen-attacked barley is possible at present.

Because Lys-48–linked and Lys-63–linked polyubiquitination of proteasome and endosome/lysosome targets, respectively, is a prerequitise for their degradation, polyubiquitin depletion might have a major impact on barley–powdery mildew interactions if these pathways are involved in defense of incompatible or support of compatible formae specialis of the pathogen. However, disruption of polyubiquitin gene function will inevitably have very serious consequences for cellular functions, probably leading to embryo lethality in stably transformed plants that express corresponding RNAi constructs. On the other hand, forward genetic approaches like ethyl methanesulfonate mutagenesis would probably never reveal any phenotypes because there are many constitutively expressed, highly homologous polyubiquitin-encoding genes in the barley genome that would probably complement mutations (http://www.tigr.org/tigr-scripts/tgi/T_index.cgi?species=barley). In contrast with those limitations, the transient gene-silencing system used here offers as major advantages the possibility to reveal gene function during transition from the nonsilenced to a silenced state, and its versatility with respect to optimizing the strength of individual silencing constructs (by varying inverted-repeat length) and to targeting larger numbers of genes. Moreover, the RNAi rescue approach introduced here allows for specific testing of hypotheses based on site-directed mutagenesis. By exploiting these advantages, we observed extreme host susceptibility of epidermal cells bombarded with RNAi constructs directed against the major group of highly homologous, highly expressed polyubiquitin genes.

The observation that a K63R mutant of monoubiquitin was most strongly complementing the resistant phenotype in polyubiquitin-depleted cells was in line with the above-mentioned speculation about a major role of the ubiquitin/proteasome pathway in basal host resistance. To our surprise, this observation was not supported by the data from a TIGS approach to all known components of the 19S RP of the 26S proteasome that revealed enhanced basal resistance rather than hypersusceptibility. In this context, it is important to note that successful bacterial plant pathogens have been reported to inject or recruit E3 ligase proteins for host invasion (Abramovitch et al., 2006; Janjusevic et al., 2006). Therefore, in the absence of race-specific or nonhost resistance, the ubiquitin/proteasome pathway may be involved in susceptibility rather than basal defense, in line with our observation from TIGS of proteasome components. Is it possible that Lys-48–linked polyubiquitination required for basal defense of barley against Bgh is involved in other pathways than the proteasomal one? Recent reports in animal systems suggest a more complex role of Lys-48–linked polyubiquitination: In human cells, the receptor for hepatocyte growth factor (Met), which is a Tyr kinase, was found to be degraded in a proteasome-independent manner but still required Lys-48–linked polyubiquitination (Carter et al., 2004). Also, Lys-48 linkages of ubiquitin chains have been reported to be required as an export signal of proteins from the endoplasmatic reticulum into the cytoplasm (Flierman et al., 2003) and to be recognized by the deubiquitinating enzyme UBPY (Row et al., 2006). In barley, we tentatively propose a still unknown pathway of protein modification important for basal host defense that is initiated by attachment of Lys-48–linked polyubiquitin chains (Figure 7). Possibly, Lys-48–linked ubiquitin chains provide a signal for partial deubiquitination, resulting in monoubiquitinated proteins that will not be targeted to the proteasome (Miller et al., 2004).

Figure 7.

Model of Protein Ubiquitination Pathways in Pathogen-Attacked Barley.

Please note that the suppressor of basal host defense, MLO, also negatively influences nonhost resistance against Bgt but, at normal expression levels, is unable to break nonhost resistance (Elliott et al., 2002).

The dependence of protein (poly)ubiquitination was restricted to basal host resistance because TIGS of polyubiquitin genes did not affect nonhost- or mlo-mediated resistance, two other types of race-nonspecific and durable resistance based on effective papillae. This result is clearly different from reported effects of overexpression of MLO or BAX inhibitor proteins or TIGS of a gene encoding the tSNARE protein Hv SNAP34 (Elliott et al., 2002; Eichmann et al., 2004; Douchkov et al., 2005). Misexpression of those genes always affected host basal or mlo-mediated and nonhost resistance. It is tempting to speculate that, in the case of nonhost resistance or host resistance mediated by the loss of MLO protein, which is a negative regulator of basal defense, a step in signaling or defense is bypassed that is highly dependent on protein polyubiquitination (Kim et al., 2002; Piffanelli et al., 2002). This bypass might be based on additional fungal-derived elicitors in the case of nonhost resistance, thereby inducing a signaling pathway that may have its convergence point with the host pathogen–induced pathway downstream of a polyubiquitination step. In the case of mlo-mediated host resistance, the absence of the MLO protein as a potential target of polyubiquitination to limit its defense-suppressive effect might explain the insensitivity of plants carrying the mlo resistance gene to polyubiquitin depletion (Figure 7). In this case, however, one would expect a role of Lys-63–linked rather than Lys-48–linked polyubiquitination because the former has been shown to be responsible for lysosomal degradation of integral membrane proteins to which class MLO belongs (Büschges et al., 1997). Indeed, we also observed partial restoration of basal resistance by the RNAi rescue construct encoding the K48R mutant of monoubiquitin that can still target proteins to the endosome/lysosome (Figure 5B). Moreover, the accumulation of multivesicular bodies, which are the hallmark of an induced endocytotic pathway, has recently been demonstrated in Bgh-attacked barley epidermal cells (An et al., 2006). Nevertheless, functional complementation of basal resistance by the K63R mutant of monoubiquitin was more efficient. Possibly, MLO interacting proteins or downstream signaling components of MLO are targets of proteasome-independent, Lys-48–linked polyubiquitination.

METHODS

Plants and Fungi

Barley (Hordeum vulgare) seedlings cv Golden Promise and cv Ingrid and wheat (Triticum aestivum) seedlings cv Kanzler were grown in pots of compost soil (from the IPK nursery) in a growth chamber (16 h of light from metal halogen lamps; 8 h of dark, 70% relative humidity, 20°C constant temperature). For the study of TIGS effects in mlo-resistant plants, a near-isogenic BC₇ line of cv Ingrid carrying the loss-of-function mlo5 allele was used and compared with the recurrent parent cv Ingrid (Büschges et al., 1997). Blumeria graminis DC Speer f. sp hordei (Bgh) (isolate CH-4.8) was maintained at 22°C under a 16-h-light/8-h-dark cycle by weekly transfer to fresh barley cv Golden Promise. B. graminis DC Speer f. sp tritici Em Marchal (Bgt) (Swiss field isolate FAL 92315) was maintained at 20°C under a 16-h-light/8-h-dark cycle by weekly transfer to fresh wheat cv Kanzler.

Plasmid Constructs

Constructs carrying unstable GFP variants were obtained by PCR amplification of two fragments of barley ACS from cDNA clone HY07A14 (accession no. AL507850) using adapter primers 5′-TTTTTCCCGGGCTTGTCCGTCTCTGTCTCTGCTTCTGCTC-3′ and 5′-TTTTTCCACGACCCTGGACCTGGCCTTCCACCCCCTCAC-3′ or adapter primers 5′-TTTTTCCCGGGCTTGTCCGTCTCTGTCTCTGCTTCTGCTC-3′ and 5′-TTTTTCCACGACCCTGGGCTTCACCCCCGACCTCCAGC-3′ and ligation in frame into the BstXI and SmaI sites of pGermin:GFP after excision of a germin gene fragment (Schweizer et al., 1999). This added 113 and 164 amino acids upstream of GFP in constructs pIPKTA25 and pIPKTA26, respectively.

RNAi constructs were made by a combined ligation/recombination method using plasmid pIPKTA30N as final GATEWAY destination vector as described (Douchkov et al., 2005). For RNAi constructs directed against polyubiquitin genes or subunits of the proteasome 19S RP, PCR fragments were amplified from barley EST clones of the IPK (http://pgrc.ipk-gatersleben.de/est/) using two gene-specific primers per clone (see Supplemental Table 1 online). This resulted in a collection of RNAi constructs carrying inverted repeats of ∼500 bp in length except for pIPKTA30_Ubi_short, where inverted repeat length was reduced to 246 bp (Figure 1). All TIGS constructs used for repeated experiments were verified by sequencing of both inverted repeats after EcoRV digestion and gel elution of the inverted-repeat double bands (Douchkov et al., 2005).

Construct pIPKTA9_Ubi for overexpression of a polyubiquitin gene was obtained by subcloning a BamHI-ApaI fragment of full coding sequence EST clone HA12I21 (accession no. BU978086) into the BamHI-ApaI sites of pIPKTA9 (see Supplemental Figure 1 online). This resulted in overexpression of a polyubiquitin gene encoding five monoubiquitin units.

The RNAi rescue construct pIPKTA9_wobble_x was obtained by designing three pairs of overgo-oligonucleotides of ∼120 kb in length that overlapped at a length of 20 bp. The recessing single strands were annealed by heating to 95°C for 2 min, followed by cooling to 68°C at a ramp of 0.05°C s⁻¹. The fill-in reaction was performed using Taq polymerase (Qiagen) at 68°C for 1 min (one elongation step). The ends of the resulting double-stranded synthetic gene fragments of ∼250 bp were restriction digested by XbaI and BamHI and ligated into pIPKTA9 (see Supplemental Figure 2 online). These gene fragments are saturated in silent mutations compared with the original barley gene represented by EST clone HO13K06, while avoiding rare codons in barley (http://www.kazusa.or.jp/codon/H.html), and encode either wild-type monoubiquitin or K48R or K63R mutant versions of the protein.

Microprojectile Bombardment and Challenge Inoculation for TIGS

Leaf segments of 7-d-old barley or wheat seedlings were used for microprojectile bombardment as described (Christensen et al., 2004). For experiments measuring the effect of pathogen attack and/or TIGS on stability of ACS:GFP fusion proteins, leaf segments were cobombarded with a GFP construct and pBC17 or pUbiGUS plus an RNAi construct where indicated, followed by challenge inoculation with either Bgh or Bgt 4 h after bombardment. Plasmid pBC17 giving rise to anthocyanin staining (Schweizer et al., 2000) or pUbiGUS (Schweizer et al., 1999) served as internal controls for transformation efficiency. For experiments measuring the effect of TIGS on (non)host–pathogen interactions, barley leaf segments were cobombarded with an RNAi construct plus pUbiGUS, followed by challenge inoculation with Bgh and Bgt 3 and 4 d after bombardment, respectively. Inoculation density was usually 180 to 220 conidia mm⁻². Interaction phenotypes of GUS-stained, transformed epidermal cells were determined 40 h after inoculation via light microscopy by counting the number of GUS-stained cells and the number of fungal haustoria.

19S RP Unigene Selection

The crop EST database of the IPK (http://pgrc.ipk-gatersleben.de/cr-est) was queried for 19S RP components using two keyword-based searches (proteasome and non-ATPase, and proteasome and ATPase). This yielded 510 hits from 25 EST libraries (library ID HA to HZ) corresponding to ∼140,000 EST sequences. The selected EST sequences were clustered using the StackPack software (Electric Genetics). This clustering resulted in 53 consensus and 21 singleton sequences. The BLASTX results of primary consensi and singletons were analyzed, and sequences associated with E-values weaker than 10⁻¹⁰ or with best BLASTX hits against fungal RP homologs were discarded. The contaminating fungal sequences were derived from library HO, which was prepared from Bgh- and Bgt-attacked barley epidermis. These steps resulted in a final set of 42 unigenes encoding all 17 subunits of the 19S RP of barley.

Accession Numbers

Sequence data of synthetic monoubiquitin genes can be found in the GenBank/EMBL data libraries under accession numbers DQ986351, DQ986352, and DQ986353.

Supplemental Data

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

• Supplemental Figure 1. Schematic Drawing of Plasmid pIPKTA9 Used for Transient Overexpression.

• Supplemental Figure 2. Design of the Synthetic Ubi_wobble Genes.

• Supplemental Table 1. cDNA Clones and Oligonucleotide Primers Used for RNAi Constructs Targeting 19S RP Components.

• Supplemental Table 2. A Unigene Set of Barley Encoding Subunits of the Proteasome 19S RP.