Role of SGT1 in resistance protein accumulation in plant immunity

Cristina Azevedo, Shigeyuki Betsuyaku, Jack Peart, Akira Takahashi, Laurent Noël, Ari Sadanandom, Catarina Casais, Jane Parker, Ken Shirasu

Author Affiliations

  1. Cristina Azevedo1,,
  2. Shigeyuki Betsuyaku2,,
  3. Jack Peart1,
  4. Akira Takahashi1,
  5. Laurent Noël2,3,
  6. Ari Sadanandom1,
  7. Catarina Casais1,
  8. Jane Parker*,2 and
  9. Ken Shirasu*,1
  1. 1 The Sainsbury Laboratory, John Innes Centre, Norwich, UK
  2. 2 Max‐Planck‐Institut für Züchtungsforschung, Department of Plant Microbe Interactions, Köln, Germany
  3. 3 Laboratoire de Biologie du Développement des Plantes, UMR6191 CNRS‐CEA, Université Aix‐Marseille II, CEN Cadarache, Saint Paul‐lez‐Durance Cedex, France
  1. *Corresponding authors: Max‐Planck‐Institut für Züchtungsforschung, Department of Plant Microbe Interactions, Carl‐von‐Linné‐Weg 10, 50829 Köln, Germany. Fax:+49 221 5062353; E‐mail: parker{at}mpiz-koeln.mpg.deRIKEN Plant Science Center, 1‐7‐22 Suehiro, Tsurumi, Yokohama 230‐0045, Japan. Tel.: +81 45503 9445; Fax: +81 45503 9573; E‐mail: ken.shirasu{at}
View Full Text


A highly conserved eukaryotic protein SGT1 binds specifically to the molecular chaperone, HSP90. In plants, SGT1 positively regulates disease resistance conferred by many Resistance (R) proteins and developmental responses to the phytohormone, auxin. We show that silencing of SGT1 in Nicotiana benthamiana causes a reduction in steady‐state levels of the R protein, Rx. These data support a role of SGT1 in R protein accumulation, possibly at the level of complex assembly. In Arabidopsis, two SGT1 proteins, AtSGT1a and AtSGT1b, are functionally redundant early in development. AtSGT1a and AtSGT1b are induced in leaves upon infection and either protein can function in resistance once a certain level is attained, depending on the R protein tested. In unchallenged tissues, steady‐state AtSGT1b levels are at least four times greater than AtSGT1a. While the respective tetratricopeptide repeat (TPR) domains of SGT1a and SGT1b control protein accumulation, they are dispensable for intrinsic functions of SGT1 in resistance and auxin responses.


Most eukaryotic organisms are able to recognize potential pathogens and express appropriate defences to prevent disease. In plants, one of the most effective surveillance systems is controlled by Resistance (R) proteins that recognize, directly or indirectly, specific pathogen effectors, thereby triggering a rapid immune response (Belkhadir et al, 2004). The largest class of R proteins shares structural motifs with animal proteins involved in innate immunity, suggesting that related processes may operate in plants and animals (Inohara and Nunez, 2003). R protein recognition of different pathogens amplifies a common set of basal defences (Tao et al, 2003) and often involves massive localized accumulation of reactive oxygen species and programmed cell death known as the hypersensitive response (HR) (Shirasu and Schulze‐Lefert, 2000).

To date, few components that regulate fundamental aspects of R protein triggered responses have been isolated and characterized in detail (Hammond‐Kosack and Parker, 2003). Among those identified, RAR1 (require for Mla12 resistance), HSP90 (heat shock protein 90) and SGT1 (suppressor of the G2 allele of skp1) are required for resistance mediated by multiple R proteins recognizing viral, bacterial, oomycete or fungal pathogens (Shirasu and Schulze‐Lefert, 2003; Schulze‐Lefert, 2004). RAR1 contains two highly conserved zinc‐binding domains called CHORD‐I and CHORD‐II (cysteine and histidine rich domain). The functions of CHORD containing proteins vary in different eukaryotic organisms. In mouse, a CHORD protein known as Melusin acts as a biomechanical sensor that prevents cardiac failure in response to pressure overload (Brancaccio et al, 2003). In Aspergillus, the CHORD‐containing protein (CHP1) is essential for maintenance of diploidy (Sadanandom et al, 2004). Caenorhabditis elegans requires CHP for proper development while yeast does not have proteins with CHORD domains (Shirasu et al, 1999). Involvement of CHP proteins in immune responses in these organisms has not been demonstrated.

We showed previously that CHORD‐I of RAR1 binds to the molecular chaperone HSP90 (Takahashi et al, 2003). Inhibitor treatments, partial silencing, or mutations of cytosolic HSP90 isoforms resulted in attenuated resistance responses, implying that RAR1 may function closely with HSP90 (Hubert et al, 2003; Lu et al, 2003; Takahashi et al, 2003). In support of this idea, both RAR1 and HSP90 are required to stabilize certain NB‐LRR proteins. The Arabidopsis NB‐LRR type R protein RPM1 did not accumulate in Arabidopsis rar1 or hsp90.2 mutants (Tornero et al, 2002; Hubert et al, 2003). Similarly, accumulation of potato Rx and barley MLA1 and MLA6 was reduced in backgrounds depleted in RAR1 (Bieri et al, 2004). A current hypothesis is that direct association of RAR1 with HSP90 stabilizes R proteins in a restrained conformation that is competent to receive pathogen signals in a manner broadly similar to steroid receptor complexes in animals (Hubert et al, 2003; Schulze‐Lefert, 2004).

Both RAR1 and HSP90 bind to SGT1 (Takahashi et al, 2003). SGT1 is also a conserved eukaryotic protein that functions in multiple biological processes through interaction with different protein complexes (Shirasu and Schulze‐Lefert, 2003). SGT1 contains three distinct domains: a tetratricopeptide repeat domain (TPR), the CS motif (present in CHP and SGT1 proteins) and the SGS motif (SGT1 specific sequence). In yeast, the TPR domain of SGT1 is essential but not sufficient for binding to HSP90 (Bansal et al, 2004). By contrast, the CS motif of barley SGT1 is sufficient to bind to the ATPase domain of HSP90 (Takahashi et al, 2003). The CS motif also interacts with the CHORD‐II domain of RAR1. It is not clear whether the CS domain binds to RAR1 and HSP90 cooperatively or competitively. The SGS motif of yeast SGT1 mediates interaction with the LRR domains of CYR1/CDC35 (Dubacq et al, 2002). Similarly, barley SGT1 interacts with the LRR domain of MLA1 via its SGS domain in a yeast two‐hybrid assay (Bieri et al, 2004).

Arabidopsis contains two SGT1 isoforms, AtSGT1a and AtSGT1b, which are highly conserved in their TPR–CS–SGS domain structures and 87% similar at the amino‐acid level. Several genetic screens identified AtSGT1b as a component of certain R gene triggered resistance responses (Austin et al, 2002; Tör et al, 2002). Also, a mutation in AtSGT1b (eta3) was identified as a genetic enhancer of the tir1‐1 mutation that exhibits impaired responses to auxin (Gray et al, 2003). TIR1 is an F‐box protein that interacts with AtCUL1, RBX1, and an SKP1‐like protein to form an SCF complex (Gray et al, 2001). An auxin stimulus causes the SCFTIR1 complex to target negative regulators of the pathway, such as auxin/indoleacetic acid (Aux/IAA) family proteins, for ubiquitin‐mediated degradation (Gray et al, 2001).

Mutations in AtSGT1a were not isolated in various genetic screens to identify components of either disease resistance or the auxin response, suggesting that AtSGT1b may be preferentially recruited to these pathways. In this study, we investigated the contribution of AtSGT1a to disease resistance and auxin hormone signalling. We establish that AtSGT1a is able to contribute positively to resistance triggered by the NB‐LRR type R proteins Rx, N, and RPP5 and can complement for loss of AtSGT1b in auxin signalling. Consistent with these findings, AtSGT1a expression is induced by pathogen infection. However, a certain amount of AtSGT1a must be attained for resistance and this level depends on the R protein tested. AtSGT1a is inherently less stable than AtSGT1b in plant tissues, and we have identified two threonine residues in the TPR domain that are, at least in part, responsible for the difference in accumulation of these two SGT1 isoforms. Intriguingly, the TPR domain of SGT1 is dispensable in both resistance and auxin responses, suggesting that this domain acts primarily at the level of SGT1 stability. We demonstrate that SGT1 is required for accumulation of Rx, suggesting that SGT1 positively controls steady‐state levels of preactivated R proteins. We propose that plant R proteins differ in the amounts of SGT1 needed to trigger effective resistance.


SGT1 controls the abundance of Rx in Nicotiana benthamiana

Two SGT1 binding proteins, RAR1 and HSP90, are needed to stabilize R proteins. To test if SGT1 is also involved in stabilization of R proteins, we depleted levels of N. benthamiana SGT1 (NbSGT1) by virus induced gene silencing (VIGS). In these experiments, we used a transgenic line of N. benthamiana that expresses an HA‐tagged NB‐LRR protein, Rx. Silencing of NbSGT1 in these N. benthamina plants resulted in reduction of steady‐state levels of Rx protein (Figure 1). This was not the case for silencing of NbEDS1, a plant defence component that is not essential for Rx resistance (Peart et al, 2002). Steady‐state levels of HSP90 were unchanged in NbSGT1 silenced plants. We conclude that, similar to RAR1 and HSP90, NbSGT1 is essential for stabilizing the Rx resistance protein in its preactivation state.

Figure 1.

NbSGT1 positively controls steady‐state levels of Rx. Western blot analysis of Rx, NbSGT1 and HSP90 levels using α‐HA, α‐SGSa and α‐HSP90 antibodies, respectively. The Rx‐4HA transgenic N. benthamiana plants were inoculated with TRV:00 (control), TRV:NbSGT1 or TRV:NbEDS1. Plants were sampled at 13–16 days postinoculation. Protein extracts were prepared as previously described (Bieri et al, 2004). Equal loading of total protein was checked by Ponceau S staining of Rubisco. For each treatment results, representative of three independent experiments are shown.

AtSGT1a is dispensable for resistance to Pseudomonas syringae pv. tomato (Pst)

Recent studies showed that AtSGT1b antagonizes RAR1‐ and HSP90‐dependent accumulation of an Arabidopsis NB‐LRR protein, RPS5, suggesting that AtSGT1b assists RPS5 degradation rather than stabilization (Holt et al, 2005). To explore this apparent discrepancy with our data, we investigated possible involvement of the second Arabidopsis SGT1 gene, AtSGT1a in disease resistance by isolating an AtSGT1a T‐DNA insertion line, designated sgt1a‐1 (Figure 2A). For comparison, we also isolated an AtRAR1 T‐DNA insertion line in the same background (Ws‐0), designated rar1‐1 (Figure 2A). Western blot analysis of these lines revealed that sgt1a‐1 and rar1‐1 are likely null mutations (Figure 2B). Inoculation of rar1‐1 leaves with different Pst DC3000 strains expressing avrRpm1, avrRpt2, avrRps4 or avrPphB confirmed the genetic requirement for AtRAR1 in RPM1, RPS2, and RPS5 resistance and showed that RAR1 is genetically dispensable for RPS4 resistance in Ws‐0 (Figure 2C–G). Growth of Pst harboring avrRpm1, avrRpt2, avrRps4 or avrPphB in sgt1a‐1 was similar to that observed in wild‐type Ws‐0, indicating that AtSGT1a is not genetically required for these resistance responses (Figure 2D–G). An sgt1a‐1 rar1‐1 double mutant also exhibited responses equivalent to rar1‐1 (Figure 2C–G).

Figure 2.

Analysis of of Arabidopsis sgt1a‐1 and rar1‐1 mutants. (A) Relative position of T‐DNA insertions within the AtSGT1a and AtRAR1 genes. Exons are indicated by the black boxes. (B) Protein expression of AtSGT1a, AtSGT1b and AtRAR1. Western blot analysis of total protein extracts from mutant and wild‐type plants probed with α‐SGSa or α‐RAR1 antibody, as indicated. (C) Bacterial growth analysis of Pst DC3000 (vector) after infiltration of leaves of 6–7 week‐old plants with bacterial suspensions (1 × 105 c.f.u./ml). Leaves were harvested at 0 (white column) and 3 days (grey column) after inoculation. (D) Same analysis for Pst DC3000 (avrRpm1), (E) Pst DC3000 (avrRpt2), (F) Pst DC3000 (avrRphB), (G) Pst DC3000 (avrRps4). Asterisks indicate that bacterial growth is significantly different (P<0.05) from the wild‐type control. Experiments were repeated at least three times with similar results. (H) DIC image of cleared seed from wild‐type plant. (I) DIC image of cleared seed from a self‐pollinated mutant plant homozygous for sgt1a‐1 mutation and heterozygous for sgt1b‐1.

An sgt1a‐1 sgt1b‐1 double mutant is embryo lethal

Since neither sgt1a‐1 (this study) nor sgt1b mutants (Muskett et al, 2002; Tör et al, 2002) were compromised in resistance to Pst strains, we reasoned that AtSGT1a and AtSGT1b may have redundant functions in these signalling pathways. To test this hypothesis, we attempted to create an sgt1a‐1 sgt1b‐1 double mutant. As AtSGT1a and ATSGT1b are on the same chromosome IV, we first created parental lines homozygous for one allele and heterozygous for the other (SGT1a sgt1a/sgt1b sgt1b or sgt1a sgt1a/SGT1b sgt1b) and characterized their seeds. The double mutant was embryo lethal (Figure 2H and I, Supplementary Table 1). These results show that SGT1 activity is essential for early development in Arabidopsis and that AtSGT1a and AtSGT1b are redundant in this process.

AtSGT1a is induced upon pathogen infection

We examined the modes of AtSGT1a and AtSGT1b expression upon pathogen infection by making stable transformants in which the promoters of AtSGT1a or AtSGT1b were fused to the β‐glucuronidase (GUS) gene. Three independent transgenic lines for each construct were inoculated with Hyaloperonospora parasitica isolate Noco2 (avirulent; recognized by RPP5) or isolate Cala2 (virulent) and examined at different time points after inoculation. In the incompatible (Figure 3A–D) and compatible (Figure 3E and F) interactions, there was induction of GUS activity of both pAtSGT1a∷GUS and pAtSGT1b∷GUS around pathogen infection sites and the leaf vasculature. More intense GUS staining of pAtSGT1a∷GUS than pAtSGT1b∷GUS was consistently observed at infection foci and around the vascular tissues (Figure 3A–F). This may reflect higher basal levels of AtSGT1a expression in the leaves. However, data from gene expression microarrays of Arabidopsis leaves indicate similar levels of AtSGT1a and AtSGT1b mRNAs in healthy Arabidopsis leaves and early responsiveness (2 h) of AtSGT1a but not AtSGT1b transcripts to infiltration with Pst DC3000 expressing avrRpm1 (Zimmermann et al, 2004). We concluded that AtSGT1a is responsive to multiple biotic stresses consistent with a role of AtSGT1a in defence, despite preferential genetic recruitment of AtSGT1b in R gene‐triggered resistance.

Figure 3.

AtSGT1a and AtSGT1b expression is induced by pathogen infection. Induction of β‐glucuronidase (GUS) activity in Ler transgenics expressing pAtSGT1a∷GUS (A, C, E) or pAtSGT1b∷GUS (B, D, F) was examined 3 and 7 days after inoculation (dpi) with avirulent (Noco2) or virulent (Cala2) H. parasitica, as indicated. GUS‐stained leaves were viewed under a light microscope. Pictures are representative of three independent experiments using three independent transgenic lines for each construct. HR, hypersensitive response at pathogen infection foci; M, pathogen mycelium; O; pathogen oospores; V, vasculature. Bars represent 200 μm.

AtSGT1a can function in R gene‐mediated resistance

We explored whether differences in AtSGT1a and AtSGT1b expression levels in Arabidopsis may account for preferential genetic recruitment of AtSGT1b in certain R gene mediated responses. Since RPP5 resistance is compromised in the sgt1b‐3 null mutant (Austin et al, 2002), we tested whether overexpression of AtSGT1a would revert this phenotype. sgt1b‐3 plants were transformed with AtSGT1a expressed under the control of either the constitutive 35S or native AtSGT1b promoter (respectively denoted 35S∷gAtSGT1a and pAtSGT1b∷gAtSGT1a). Homozygous transgenic lines with different levels of transgene expression were selected on Western blots (Figure 4A; Supplementary Figure 1) and infected with H. parasitica isolate Noco2 (Figure 4B; Supplementary Table 2). Lines 7.1 and 8.5 with high AtSGT1a protein levels (⩾10 fold higher than in the sgt1b‐3 mutant; Figure 4A; Supplementary Figure 1) fully complemented the sgt1b‐3 defect, manifested by restoration of spatially restricted HR lesions after pathogen infection of leaves. Lower levels of AtSGT1a expression in lines 3.4, 3.6 and 8.10 (<8‐fold higher than in sgt1b‐3; Figure 4A; Supplementary Figure 1) only partially compensated for loss of AtSGT1b in RPP5 resistance, as seen by the development of HR lesions together with trailing necrosis (TN) and occasional sporulation on leaves infected with Noco2 (Figure 4B; Supplementary Table 2). For comparison, sgt1b‐3 was also transformed with AtSGT1b controlled either by its own (pAtSGT1b∷gAtSGT1b) or by the AtSGT1a promoter (pAtSGT1a∷gAtSGT1b). As observed for AtSGT1a, we found a dose‐dependent complementation of the sgt1b‐3 defect by AtSGT1b. For example, AtSGT1b expression in line 2.3 was not sufficient to fully complement sgt1b‐3 in RPP5 resistance (Figure 4; Supplementary Figure 1 and Supplementary Table 2). In contrast, transgenic lines 5.1, 5.3 and 6.3 that expressed greater levels (⩾3‐fold higher than in line 2.3) of AtSGT1b fully complemented sgt1b‐3 (Figure 4; Supplementary Figure 1 and Supplementary Table 2). These results demonstrate that AtSGT1a, like AtSGT1b, is able to function in R gene‐mediated resistance and that a minimum level of either protein needs to be attained for full expression of disease resistance.

Figure 4.

Expression of AtSGT1a or AtSGT1b in transgenic Arabidopsis complements sgt1b‐3 in RPP5 resistance. (A) Western blot analysis of total leaf extracts from different sgt1b‐3 plants transformed with AtSGT1a or AtSGT1b under different promoters. SGT1 proteins were visualized by the α‐SGSa antibody. (B) Lactophenol trypan blue staining of Noco‐2 infected leaves 7 days after inoculation reveals necrotic plant cells and pathogen structures. HR, hypersensitive response; M, mycelium; TN, trailing plant cell necrosis; S, sporangiophore. Bars represent 500 μm.

To determine if AtSGT1a can participate in defence signalling pathways other than those tested in Arabidopsis, we developed a heterologous transient complementation assay in N. benthamiana. This assay was based on the fact that VIGS of NbSGT1 leads to compromised R gene triggered resistance against several pathogens (Liu et al, 2002; Peart et al, 2002). If AtSGT1a is able to function in resistance, we expected it to complement the loss of NbSGT1. We monitored both hypersensitive plant cell death (HR) associated with transient expression of PVX coat protein (PVX‐Tk) and resistance against green fluorescent protein (GFP)‐tagged recombinant PVX conditioned by Rx (Peart et al, 2002). For transient assays, we used two types of Agrobacterium strain/plasmid sets, C58C1 (pBin19 based vector, 35S promoter) and GV3101 (pPM90RK, pPAM_MCS based vector, 35S promoter) for high and low expression, respectively. Analyses of AtSGT1a expression mediated by the two different strains showed that levels with C58C1 were two to five times higher than that observed for GV3101 (Figure 5A). As previously reported (Peart et al, 2002), no HR was observed in TRV:NbSGT1 plants co‐infiltrated with Agrobacterium cells carrying the vector control and a plasmid expressing PVX‐Tk (Figure 5B). By contrast, transient expression of either AtSGT1a or AtSGT1b using the low expression system restored the HR (Figure 5B). Similarly, there was a significant but incomplete reduction of PVX‐GFP accumulation in TRV:NbSGT1 plants transiently expressing AtSGT1a or AtSGT1b while no strong reduction of PVX:GFP expression was observed with the control vector (Figure 5C). Expression of AtSGT1a or AtSGT1b in wild‐type N. benthamiana TRV:00 plants had no effect on HR or PVX:GFP (Figure 5B and C). These data show that AtSGT1a has the capacity to function in resistance mediated by Rx.

Figure 5.

AtSGT1a and AtSGT1b are functional in Rx resistance. (A) Western blot analysis of differential AtSGT1a protein expression levels by Agrobacterium strains C58C1 with a pBIN61‐based vector and GV3101 with a pPAM‐MCS vector. Protein extracts from N. benthamiana leaves infiltrated with indicated strains were separated by SDS–PAGE and blotted on a membrane. AtSGT1a was visualized by α‐SGSa antibody. (B) Appearance of HR elicited by co‐infiltration of Agrobacterium expressing the PVX‐Tk (PVX coat protein) and the test constructs. One week‐old wild type and transgenic N. benthamiana plants (Rx, N) were inoculated with TRV:NbSGT1 or TRV:00 by Agrobacterium infiltration. Approximately 3 weeks later PVX elicitor was co‐expressed with the test constructs AtSGT1a, AtSGT1b or empty vector by Agrobacterium infiltration. Plants were photographed 5 days postinoculation under white light. (C) Rx resistance against PVX. Co‐infiltration of Agrobacterium carrying PVX:GFP (OD600=0.001) and the test construct (OD600=0.25) 3 weeks after inoculation of TRV:00 or TRV:NbSGT1. Accumulation of PVX:GFP was monitored by GFP fluorescence under UV illumination 5 days post inoculation. Similar results were obtained in three independent experiments.

We then investigated whether AtSGT1a and AtSGT1b can function in N‐mediated resistance against GFP‐tagged recombinant TMV (TMV:GFP; Peart et al, 2002). Using the low expression system, transient expression of AtSGT1b, but not AtSGT1a, suppressed GFP expression in TRV:NbSGT1 plants, indicating that AtSGT1a was unable to complement the loss of NbSGT1 in N resistance (Figure 6A). However, when AtSGT1a was expressed using the high expression system, full complementation of N resistance was observed in TRV:NbSGT1 transgenic plants (Figure 6B). These results show that AtSGT1a can function in N resistance when highly expressed. They also demonstrate that higher levels of AtSGT1a are required for resistance mediated by N compared to Rx.

Figure 6.

Dose‐dependent recruitment of AtSGT1a in N resistance. (A) AtSGT1b but not AtSGT1a is functional in N resistance against TMV when expressed from the Agrobacterium strain GV3101 (pMP90RK) in a pPAM‐MCS‐based vector. Wild type and transgenic N. benthamiana plants were treated as described in Figure 5. Co‐infiltration of Agrobacterium carrying TMV:GFP (OD600=0.025) and the test construct (OD600=0.25) as indicated. Accumulation of TMV:GFP was monitored by GFP fluorescence under UV illumination 6‐7 days post inoculation. Similar results were obtained for three independent experiments. (B) Comparison of two Agrobacterium strains, C58C1 with a pBIN61‐based vector and GV3101 with a pPAM‐MCS vector for AtSGT1a functionality in N resistance.

Differential accumulation of AtSGT1b and AtSGT1a in planta is determined by the TPR domain

To investigate if there is an inherent difference between AtSGT1a and AtSGT1b accumulation in Arabidopsis, we generated multiple independent stable transgenic lines of sgt1b‐3 expressing AtSGT1a or AtSGT1b under their own promoters and C‐terminally fused to a StrepII (SII) affinity purification tag (Witte et al, 2004). AtSGT1b was more highly expressed than AtSGT1a in three independent transformants, as shown in Western blot analysis of one representative line for each construct (Figure 7A). Levels of AtSGT1b‐SII were at least four times higher than AtSGT1a‐SII when measured in a dilution series of leaf soluble extracts on a Western blot probed with anti‐SII antibody (data not shown). Affinity purification of silver stainable amounts of both proteins after transient expression in N. benthamiana leaves established that the SII‐tagged AtSGT1a and AtSGT1b were detected to the same efficiency by anti‐SII antibody (Supplementary Figure 2). Additionally, we found that α‐SGSa has higher affinity for AtSGT1a than for AtSGT1b since the difference in protein expression was not apparent using this antibody (Figure 7A, lanes 3 and 4).

Figure 7.

The TPR domain controls steady‐state levels of Arabidopsis SGT1 proteins. (A) Western blot analysis of AtSGT1a‐SII (indicated by black arrow) and AtSGT1b‐SII (white arrow). (B) Relative abundance of different AtSGT1a and AtSGT1b constructs transiently expressed in TRV:NbSGT1 plants determined by Western blot analysis with the α‐SGSa antibody. A chimeric construct AtSGT1a/b/b containing the TPR domain (M1‐V163) from AtSGT1a and the remaining protein from AtSGT1b (V171‐Y357) while the reciprocal chimera AtSGT1b/a/a contains M1‐A170 from AtSGT1b fused to V164‐I350 of AtSGT1a. (C) AtSGT1a(T91A+T100A) and AtSGT1b(A91T+A100T) contructs were transiently expressed in TRV:00 or TRV:NbSGT1 plants as indicated and analysis of N resistance performed as in Figure 6A. Accumulation of TMV:GFP was monitored by GFP fluorescence under UV illumination 6–7 days post inoculation.

To identify domains responsible for differential accumulation, we created chimeric constructs by swapping the TPR domains of AtSGT1a and AtSGT1b, designated AtSGT1a/b/b and AtSGT1b/a/a (Figure 7B). Western blot analyses revealed that AtSGT1b/a/a accumulated to higher levels than AtSGT1a wild‐type protein after transient expression in N. benthamiana leaves. Conversely, AtSGT1a/b/b accumulated to lower levels than wild‐type AtSGT1b. Further sequence comparison between AtSGT1a and other plant SGT1 proteins revealed the existence of three conserved amino acids in the N‐terminal TPR domain of all plant SGT1 proteins except AtSGT1a (Supplementary Figure 3). In AtSGT1a, amino acids at positions 91, 100 and 118 in the TPR domain are threonines whereas all other plant SGT1 proteins contain alanines at these positions. Phosphorylation predictions indicated that Thr91 and Thr100 are strong candidates for phosphorylation (scoring of Thr91‐0.630 and Thr100‐0.828; based on NetPhos 2.0 server When both Thr91 and Thr100 in AtSGT1a were mutated to alanines (AtSGT1a(T91A+T100A)) the protein accumulated to greater levels than wild‐type AtSGT1a after transient expression in N. benthamiana leaves (Figure 7B). By contrast, exchange of the corresponding Ala91 and Ala100 to Thr in AtSGT1b (AtSGT1b(A91T+A100T)) caused the protein to accumulate to lower levels than wild‐type AtSGT1b (Figure 7B). From these data we concluded that the N‐terminal TPR domain of AtSGT1a reduces the steady‐state level of Arabidopsis SGT1 proteins whereas the same domain from AtSGT1b enhances SGT1 accumulation.

Next, we tested whether the AtSGT1a and AtSGT1b amino‐acid exchange variants are functional in N. benthamiana pathology assays using the low expression system, GV3101. As expected, AtSGT1a(T91A+T100A) and AtSGT1b(A91T+A100T) were able to complement NbSGT1 for Rx resistance (data not shown). Similarly, both AtSGT1a(T91A+T100A) and AtSGT1b(A91T+A100T) complemented NbSGT1 for N resistance in TRV:NbSGT1 plants (Figure 7C). Therefore, an increase in protein stability observed in AtSGT1a(T91A+T100A) is sufficient to attain a level required for N resistance while the amount of AtSGT1b(A91T+A100T) was above the minimal level. Analysis of protein expression of SGT1 derivatives in yeast did not show significant differences in stability (data not shown), suggesting that instability due to the presence of the AtSGT1a TPR domain is a plant‐specific phenomenon.

The TPR domain is dispensable for SGT1 resistance function and auxin signalling

A link between the nature of the TPR domain and stability of Arabidopsis SGT1 proteins led us to speculate that the TPR domain may have a regulatory role. We therefore transiently expressed AtSGT1a or AtSGT1b that lack their corresponding TPR domain (ΔTPRa or ΔTPRb) in TRV:NbSGT1 N. benthamiana transgenic plants and tested for restoration of Rx resistance. The ΔTPRa and ΔTPRb both complemented NbSGT1 in Rx mediated resistance as seen by the lack of GFP fluorescence after inoculation with PVX:GFP (Figure 8). These results demonstrate that the TPR domain of either AtSGT1a or AtSGT1b is not essential for resistance conditioned by Rx, at least when they are transiently expressed.

Figure 8.

The TPR domain is dispensable for Rx resistance in N. benthamiana. ΔTPRa and ΔTPRb were expressed using the Agrobacterium strain GV3101 (pMP90RK) with a pPAM_MCS‐based vector. Experiments were performed as described in Figure 5.

In order to assess the influence of the AtSGT1a and AtSGT1b TPR domains in Arabidopsis, we transformed the sgt1b‐1 null mutant with ΔTPRa and ΔTPRb and selected independent lines with different levels of transgene expression by Western blot analyses (Supplementary Figure 4). First, we tested whether constitutive expression of either ΔTPRa or ΔTPRb is able to complement the sgt1b‐1 defect in RPP5 resistance to H. parasitica isolate Noco2. Complementation was observed in a dose‐dependent manner by both ΔTPRa and ΔTPRb (Supplementary Figure 4). Lines with higher levels of ΔTPRa or ΔTPRb expression exhibited complete resistance whereas lines with lower levels of expression were partially resistant to Noco2 as seen by the appearance of a combination of discrete HR lesions and TN (Supplementary Figure 4). Next, we tested these lines for auxin sensitivity. We found that ΔTPRa or ΔTPRb transgenic lines also complemented sgt1b‐1 as seen by restoration of auxin sensitivity (Supplementary Figure 4). These results indicate that the TPR domain of Arabidopsis SGT1 is dispensable for resistance function and auxin signalling.


AtSGT1a and AtSGT1b are functionally redundant

Arabidopsis contains two highly similar proteins, AtSGT1a and AtSGT1b, that bind to RAR1, a component required for the resistance response mediated by several R proteins. Despite the high similarity between AtSGT1a and AtSGT1b, genetic screens for loss of resistance identified mutations in AtSGT1b but not AtSGT1a. In this study, we show that both AtSGT1b and AtSGT1a mediate resistance in a dose‐dependent manner and that there is functional redundancy between the two proteins during embryo development. These data are consistent with previous findings that all tested R genes were compromised in NbSGT1 silenced N. benthamiana plants (Peart et al, 2002) and that both Arabidopsis SGT1 proteins are able to complement sgt1 mutations in yeast (Azevedo et al, 2002). An explicit prediction from these results is that Arabidopsis R genes previously identified not to require AtSGT1b (Austin et al, 2002; Tör et al, 2002) utilize AtSGT1a in an sgt1b mutant. Supporting this idea, we found that AtSGT1a promoter activity was highly induced by pathogen infection (Figure 2). AtSGT1a may thus reach a level required for the full resistance triggered by certain R proteins in the absence of AtSGT1b. We did not observe loss of resistance phenotype in sgt1a‐1 against Pst strains (Figure 2C–G) probably because there is sufficient AtSGT1b protein for these R proteins to signal effectively.

R proteins differ in the levels of SGT1 needed for resistance

Our studies in N. benthamiana and Arabidopsis reveal that the levels of SGT1 proteins required for resistance depends on the R protein tested. Transient complementation using high and low expression vectors showed that Rx requires lower amounts of AtSGT1a than N does to function (Figures 5 and 6). Similarly, RPP5 resistance is restored in transgenic plants expressing high levels of AtSGT1a (Figure 4) even though endogenous AtSGT1a is insufficient to mediate resistance in the sgt1b mutant (Austin et al, 2002; Tör et al, 2002). Transgenic lines expressing lower amounts of AtSGT1b did not fully complement the sgt1b‐3 defect in RPP5 resistance, suggesting that there is also a critical dose of AtSGT1b required for proper R protein function. In barley, transient single cell silencing of HvSGT1 compromised resistance conferred by MLA6 but not MLA1 (Azevedo et al, 2002). In the light of our new data, we consider it likely that MLA6 requires higher levels of HvSGT1 than MLA1. Transient silencing may not have been complete, leaving sufficient HvSGT1 for MLA1 to operate. In yeast, HvSGT1 can associate with MLA1‐LRR but not MLA6‐LRR (Bieri et al, 2004). Perhaps, interaction between MLA6‐LRR and HvSGT1 is weak or transient and not detectable in this assay and this may explain why MLA6 requires more HvSGT1 than MLA1 for resistance.

The TPR is a regulatory domain in Arabidopsis SGT1 proteins

The TPR domain of SGT1 has high similarity to that of the serine threonine protein phosphatase 5 (Pp5) (Takahashi et al, 2003). Recently, the TPR domain of Pp5 was shown to maintain an auto‐inhibited conformation with a C‐terminal domain suppressing the phosphatase activity (Yang et al, 2005). Here, we demonstrate that the TPR domain is at least in part responsible for the difference in AtSGT1a and AtSGT1b accumulation (Figure 7). In particular, amino acid residues Thr91 and Thr100 in the TPR domain of AtSGT1a confer reduced levels of this protein. It is possible that phosphorylation of Thr91 and Thr100 by a plant specific kinase may cause a conformational change rendering AtSGT1a less stable. Consistent with this idea, we did not observe significant differences in the amounts of Arabidopsis SGT1 proteins expressed in yeast or Escherichia coli (Marta Boter and Ken Shirasu, unpublished).

AtSGT1b is required genetically for the auxin response mediated by SKP1‐containing SCFTIR1 (Gray et al, 2003). Yeast SGT1 mediates CBF3 kinetochore assembly also through its interaction with SKP1 (Kitagawa et al, 1999). The TPR domain of SGT1 is necessary but not sufficient for the interaction (Bansal et al, 2004; Lingelbach and Kaplan, 2004). In contrast, C. elegans and Drosophila SGT1 proteins do not contain a TPR domain (Kitagawa et al, 1999). We show that although the TPR domain controls Arabidopsis SGT1 protein stability, it is not important for intrinsic disease resistance and auxin signalling functions of SGT1. We reason therefore that interaction of SKP1 and SGT1 via the TPR domain is not crucial to resistance and auxin signalling pathways. While association between SGT1 and SKP1 has been demonstrated in barley and tobacco co‐IP experiments (Azevedo et al, 2002; Liu et al, 2002), we were unable to demonstrate interaction between Arabidopsis SGT1 proteins and SKP1 homologues either by yeast two hybrid assays or by co‐IP experiments (data not shown). SGT1 has low affinity for SKP1 (Bansal et al, 2004; Lingelbach and Kaplan, 2004) and our current protocols may not detect such weak or transient interactions.

SGT1 functions in R protein accumulation in disease resistance

An outstanding question is how SGT1 proteins regulate plant R protein function. We show that silencing of NbSGT1 results in reduced steady‐state levels of Rx and conclude that SGT1, like RAR1 and HSP90, has a role in maintaining stability of preactivated R proteins. A previous study showed that silencing HSP90 resulted in greater reduction of Rx protein levels compared to those in NbSGT1 silenced plants, suggesting that HSP90 may also affect other processes, such as translation of Rx (Lu et al, 2003). Holt et al provided evidence that AtSGT1b antagonizes RAR1‐ and HSP90‐dependent accumulation of RPS5, suggesting that AtSGT1b assists RPS5 degradation. However, interpretation of SGT1 activities in Arabidopsis is complicated by the presence of two functional SGT1 proteins with different expression characteristics, as shown here. Our data lead us to consider an alternative model to explain the results of Holt et al (2005). In this model (Figure 9), RPS5 and other NB‐LRR protein stabilization in sgt1b may be conferred by AtSGT1a when expressed above a certain level. In wild‐type Arabidopsis, AtSGT1b provides the principle SGT1 activity due to its higher abundance and possibly also preferential HSP90 association (Hubert et al, 2003). In the absence of AtSGT1b, AtSGT1a activity may be invoked. We have demonstrated that the amounts of both SGT1 proteins required for resistance depends on the R protein tested. Therefore, for some R proteins, such as RPP5 (Austin et al, 2002), the amount of active AtSGT1a in the absence of AtSGT1b may not be sufficient for proper R protein function. If certain R proteins, such as RPS5, require lower AtSGT1a levels for assembly of competent complexes, this would explain recovery of RPS5 protein in rar1 sgt1b mutants observed by Holt et al (2005). By the same argument, instability of RPM1 in rar1 is not recovered in rar1 sgt1b double mutants, possibly because RPM1 requires higher levels of AtSGT1a than does RPS5 to signal resistance. The balance of assembly and degradation of such complexes is likely to be tightly regulated, and we cannot exclude a role for SGT1 or other co‐chaperones in degradatory processes once R proteins are activated (Figure 9).

Figure 9.

Model for SGT1 function in assembly of plant NB‐LRR protein complexes. An NB‐LRR type R protein is depicted as part of a ‘preactivated’ multiprotein complex with other plant protein(s) (shown as x), whose assembly requires the cooperative activities of the HSP90 chaperone and RAR1. Our data argue for a role of SGT1 in assembly (solid straight black arrow) of pre‐existing NB‐LRR proteins. In Arabidopsis, two active SGT1 isoforms, AtSGT1a and AtSGT1b, are expressed. The predominant SGT1 activity in leaves is exerted by AtSGT1b due to its higher accumulation (indicated by dark grey shading) than AtSGT1a (light grey shading) before pathogen challenge, and possibly also preferential HSP90 binding. In the absence of AtSGT1b and RAR1, AtSGT1a activity is sufficient for accumulation of some NB‐LRR proteins, such as RPS5, but not others, such as RPP5 or RPM1. This function is supported by an intrinsic capability of AtSGT1a to complement the sgt1b mutant when expressed at a certain level and by differences in the amounts of SGT1 needed for resistance conferred by various R proteins tested. It is likely that a fine balance between NB‐LRR protein assembly and degradation is mediated by chaperone and co‐chaperone associations. While the model depicts a principle activity of SGT1 in R protein assembly, it does not preclude an SGT1 contribution to degradation of R proteins (broken straight black arrow) or downstream signalling components.

Holt et al (2005) reported that instability of RPM1 and RPS5 caused by the treatment with a specific HSP90 inhibitor, geldanamycin (GDA) in wild‐type plants was not observed in sgt1b mutants. One explanation for this phenomenon is that AtSGT1b, but not AtSGT1a, requires HSP90 activity to stabilise R proteins. In this scenario, GDA is no longer able to inhibit the SGT1 activity in the absence of AtSGT1b, because AtSGT1a functions independently of HSP90 activity. A yeast two‐hybrid analysis showing that AtSGT1b but not AtSGT1a interacts with full‐length HSP90 (Cristina Azevedo and Ken Shirasu unpublished) is consistent with this hypothesis. In this sense, AtSGT1a may be an exceptional SGT1 protein since all other tested SGT1 proteins associate with HSP90 (Takahashi et al, 2003; Bansal et al, 2004; Lee et al, 2004).

In yeast, SGT1 and HSP90 cooperate in maintaining transient protein interactions essential for proper assembly of kinetochore complexes (Bansal et al, 2004; Lingelbach and Kaplan, 2004). Here, we showed that SGT1 positively controls the steady‐state levels of an R protein, similar to RAR1 and HSP90 (Figure 9). It is tempting to speculate that SGT1 functions closely with HSP90 and RAR1 to mediate of R protein complex assembly in plants. SGT1 was shown to play a role in intramolecular interactions of the Bs2 R protein in transient N. benthamiana expression assays (Leister et al, 2005). Presumably, proper folding of Bs2 is important for complex formation and SGT1 may function through interaction with Bs2 via the SGS domain and interaction with HSP90 via CS domain. Since barley RAR1 and SGT1 also act additively on MLA6 resistance (Azevedo et al, 2002), the function of RAR1 may be to enhance the performance of SGT1–HSP90 complexes.

Materials and methods

Plant materials, bacteria and virus strains

The sgt1a‐1 and rar1‐1 mutants were identified by screening the Ws‐0 Wisconsin T‐DNA KO lines for insertions in the AtSGT1a or AtRAR1 genes as described (Krysan et al, 1999). Transgenic N. benthamiana plants carrying N and Rx resistance genes, their growth conditions, VIGS of the NbSGT1 using tobacco rattle virus vector were described previously (Peart et al, 2002). Pst DC3000 strains and H. parasitica isolates used in this study are described (Austin et al, 2002; Takahashi et al, 2003). The virus strains PVX:GFP and TMV:GFP were described previously (Peart et al, 2002). Agrobacterium carrying PVX‐Tk was kindly provided by G Farnham. Agro‐infiltration to transiently express genes in N. benthamiana was performed according to Peart et al (2002). Details of Agroinfiltration protocol are also provided in the Supplementary Material.

Plasmid constructs

The AtSGT1a and AtSGT1b derivatives used in agro‐infiltration were amplified from existing clones (Azevedo et al, 2002) with primers containing specific restriction sites that allowed subcloning into different destination vectors. Details of construction schemes are described in Supplementary Material.

Whole‐mount preparation of seed

Seeds were fixed in ethanol:acetic acid (3:1) for 1 h followed by washing in 70% ethanol twice. Clearing was obtained after 30 min in a derivative of Hoyer's medium (chloral hydrate:distilled water:glycerol [8:3:1 g]). Observations were performed using differential interference contrast (DIC) optics.

Western blot analysis

Rat anti‐SGSa and rabbit anti‐SGT1b antibodies used here are described by Azevedo et al (2002) and Austin et al (2002), respectively. Anti‐RAR1 and anti‐HSP90 antibodies were described previously (Azevedo et al, 2002; Takahashi et al, 2003). Rabbit anti‐SGSa antibodies were raised against the SGSa domain and affinity purified. The protocol for StrepII tag detection using StrepTactin AP‐conjugate is described by Witte et al (2004). Total protein extracts from agro‐infiltrated N. benthamiana leaves were prepared by grinding 2 cm2 of leaf tissue in 100 μl of ice‐cold 1 × GTEN extraction buffer (10% glycerol, 25 mM Tris–HCl pH 7.5, 1 mM EDTA, 150 mM NaCl, 5 mM dithiothreitol (DTT) and 1% of plant protease inhibitor inhibitor cocktail (Sigma)) and 4 × of SDS–PAGE sample buffer. Samples were boiled for 5 min, quick spun and 5 μl run on a 12% SDS–PAGE gel, blotted onto PVDF membranes and visualized using antibodies described with ECL‐Plus (Amersham). ImageJ software (, US National Institute of Health, Bethesda, MD) was used for quantitative comparison of SGT1 proteins on the blots (see Supplementary Figure 1 for details).

Supplementary data

Supplementary data are available at The EMBO Journal Online.

Supplementary Information

Supplemental Material [emboj7601084-sup-0001.pdf]


We thank David Baulcombe for providing the N. benthamiana transgenic plants and VIGS constructs. This work was supported by grants from the Gatsby Charitable Organization (KS) and the UK Biotechnology and Biological Sciences Research Council (KS) and fellowships from Fundação para a Ciência e a Tecnologia to CA. Support to JP, SB and LN by a Deutsche Forschungsgemeinschaft ‘SFB 635’ Grant, The Alexander von Humboldt Foundation and The Max‐Planck Society is gratefully acknowledged. SB was supported in part by Research Fellowships of the Japan Society for the Promotion of Science for Young Scientists.


  • Equal first authors


View Abstract