Heat shock protein 90 and its co-chaperone proteinphosphatase 5 interact with distinct regions of the tomatoI-2 disease resistance protein

Sergio de la Fuente van Bentem1,†, Jack H. Vossen1,‡, Klaas Jan de Vries1, Saskia van Wees2, Wladimir I.L. Tameling1,§, Henk L.

Dekker3, Chris G. de Koster3, Michel A. Haring2, Frank L.W. Takken1 and Ben J.C. Cornelissen1,*

1Plant Pathology, Swammerdam Institute for Life Sciences, University of Amsterdam, Kruislaan 318, 1098 SM Amsterdam, The

Netherlands,2Plant Physiology, Swammerdam Institute for Life Sciences, University of Amsterdam, Kruislaan 318, 1098 SMAmsterdam, The

Netherlands, and3Mass Spectrometry, Swammerdam Institute for Life Sciences, University of Amsterdam, Nieuwe Achtergracht 166, 1018 WV,

Amsterdam, The Netherlands

Received 16 March 2005; revised 29 April 2005; accepted 6 May 2005.*For correspondence (fax þ31 0 20 525 7934; e-mail cornelissen@science.uva.nl).†Present address: Institute of Genetics, Vienna Biocenter, University of Vienna, Dr Bohrgasse 9, 1030 Vienna, Austria.‡Present address: Phytopathology, Wageningen University and Research Centre, Binnenhaven 9, 6709 PD Wageningen, The Netherlands.§Present address: The Sainsbury Laboratory, John Innes Centre, Colney Lane, Norwich NR4 7UH, UK.

Summary

Recent data suggest that plant disease resistance (R) proteins are present in multi-protein complexes. Tomato

R protein I-2 confers resistance against the fungal pathogen Fusarium oxysporum. To identify components of

the I-2 complex, we performed yeast two-hybrid screens using the I-2 leucine-rich repeat (LRR) domain as bait,

and identified protein phosphatase 5 (PP5) as an I-2 interactor. Subsequent screens revealed two members of

the cytosolic heat shock protein 90 (HSP90) family as interactors of PP5. By performing in vitro protein–protein

interaction analysis using recombinant proteins, we were able to show a direct interaction between I-2 and

PP5, and between I-2 and HSP90. The N-terminal part of the LRR domain was found to interact with HSP90,

whereas the C-terminal part bound to PP5. The specific binding of HSP90 to the N-terminal region of the I-2 LRR

domain was confirmed by co-purifying HSP90 from tomato lysate using recombinant proteins. Similarly, the

interaction between PP5 and HSP90 was established. To investigate the role of PP5 and HSP90 for I-2 function,

virus-induced gene silencing was performed in Nicotiana benthamiana. Silencing of HSP90 but not of PP5

completely blocked cell death triggered by I-2, showing that HSP90 is required for I-2 function. Together these

data suggest that R proteins require, like steroid hormone receptors in animal systems, an HSP90/PP5 complex

for their folding and functioning.

Keywords: PP5, co-chaperone, HSP90, LRR, disease resistance signalling, R protein complex.

Introduction

Plant disease resistance (R) genes mediate resistance to a

large variety of pathogens, including fungi, bacteria,

nematodes, viruses, insects and oomycetes. Flor (1942)

postulated that for each R gene there is a matching avir-

ulence (Avr) gene in the pathogen, and that the presence

of both is required to induce resistance. This so-called

gene-for-gene hypothesis was later translated into a

receptor-ligand model in which the R protein functions as

receptor and the corresponding Avr protein as ligand;

upon binding of the ligand to the receptor defence re-

sponses are triggered. However, the lack of evidence for a

direct interaction between most R proteins and their elici-

tors, together with the identification of plant proteins that

directly interact with both Avr and R proteins, have re-

cently led to a new interpretation of the gene-for-gene

hypothesis: the guard model (van der Biezen and Jones,

1998; Dangl and Jones, 2001; van der Hoorn et al., 2002).

In this model R proteins guard proteins (termed

284 ª 2005 Blackwell Publishing Ltd

The Plant Journal (2005) 43, 284–298 doi: 10.1111/j.1365-313X.2005.02450.x

‘guardees’) that are targets of Avr proteins. The interaction

between the Avr protein and the guardee is sensed by the

R protein that subsequently triggers defence responses.

The model implies that R proteins function in recognition

complexes containing their guardees.

R proteins can be assigned to different classes based on

their domain composition. By far the largest class of R

proteins known to date consists of proteins with a central

nucleotide-binding (NB-ARC) domain and a C-terminal

leucine-rich repeat (LRR) domain. The LRR domain is

involved in regulating the signalling activity of the R

protein and single amino acid changes in this domain can

lead to constitutively active proteins (Bendahmane et al.,

2002; Hwang and Williamson, 2003; Hwang et al., 2000). At

their N-terminus, R proteins of the NB-ARC-LRR class

contain either a TIR (for homology with animal Toll and

Interleukin-1 receptors) or a coiled-coil (CC) domain (Dangl

and Jones, 2001; Martin et al., 2003). The structure of NB-

ARC-LRR proteins suggests a conservation of the mech-

anism by which they activate signalling cascades leading

to disease resistance. Several components of signal trans-

duction routes shared by different R proteins have been

identified. Some of these components are required specif-

ically for members of either the CC or the TIR subclass

(Martin et al., 2003). Others, like RAR1 and SGT1b, are

required for a subset of R proteins irrespective of their N-

terminus. They interact with each other and both associate

with different R proteins (Austin et al., 2002; Azevedo

et al., 2002). Both RAR1 and SGT1 interact with heat shock

protein 90 (HSP90) in plants (Hubert et al., 2003; Liu et al.,

2004; Takahashi et al., 2003). In animal cells, HSP90 is a

molecular chaperone with substantial specificity for sub-

strates, designated as clients, involved in signal transduc-

tion. Among the many clients are steroid hormone

receptors such as the glucocorticoid and estrogen receptors

(ER), and many kinases (Pratt and Toft, 2003). In plants,

HSP90 is essential for resistance mediated by several R

proteins of the NB-ARC-LRR class (Hubert et al., 2003; Liu

et al., 2004; Lu et al., 2003; Takahashi et al., 2003). Further-

more, HSP90 associates with the R proteins RPM1 and N in

vivo (Hubert et al., 2003; Lu et al., 2003). Although many

HSP90 clients have been identified in animal cells (Pratt

and Toft, 2003), R proteins are the first examples of HSP90

clients in plants. The (selective) physical associations of R

proteins with RAR1, SGT1b and HSP90 support the idea

that R proteins are present in protein complexes (Shirasu

and Schulze-Lefert, 2003). Although a few additional

components of R protein complexes have been identified

(Axtell and Staskawicz, 2003; Holt et al., 2002; Mackey

et al., 2002, 2003), the composition of most R protein

complexes is still unknown.

Tomato I-2 is a CC-NB-ARC-LRR protein that confers

resistance to race 2 of Fusarium oxysporum forma specialis

lycopersici (Ori et al., 1997; Simons et al., 1998). Here, we

focus on the identification of components of the I-2 protein

complex. Because of the regulatory function of LRR domains

(Hwang et al., 2000), and as they are in general protein-

binding domains (Kobe and Deisenhofer, 1994), we hypo-

thesize that this region of I-2 might be involved in the

interaction with proteins in the I-2 complex. In order to

identify such proteins we pursued a yeast two-hybrid

approach and identified protein phosphatase 5 (PP5) as an

interactor. PP5 interacts with other R proteins as well as with

two HSP90 isoforms. Moreover, HSP90 binds directly to the

LRR domain of I-2, although to a different region as PP5.

Silencing of HSP90 abrogated I-2-mediated cell death in

Nicotiana benthamiana leaves, showing that HSP90 is

essential for I-2 function. Because PP5 is a co-chaperone of

HSP90 in animal cells, we speculate that PP5 acts as a

co-chaperone during the maturation of R proteins by HSP90

in plants.

Results

The I-2 disease resistance protein interacts with PP5

Previously we annotated 17 LRRs in the C-terminal part of I-2

(Simons et al., 1998). After a re-evaluation we now propose

that the LRR domain contains 29 repeats (Figure S1). In the

new alignment the third LRR aligns very well with the con-

served third LRR of other R proteins of the CC-NB-ARC-LRR

class (Axtell et al., 2001; Warren et al., 1998; Figure S1).

Moreover, the spacing between the NB-ARC domain and the

LRR domain is reduced now to a size more commonly

observed in R proteins: the region between the MHDL motif

(a conserved motif within the NB-ARC domain) and the first

LRR is 52 amino acids, while for Arabidopsis thaliana CC-NB-

ARC-LRR proteins this is on average 40 amino acids (Meyers

et al., 2003).

To identify proteins that interact with the LRR domain of

I-2, we performed yeast two-hybrid screens of a Fusarium–

tomato interaction cDNA library (see Experimental proce-

dures). The C-terminal part of I-2 containing the 29 LRRs was

used as bait. Two classes of cDNAs were identified that

activated the two-hybrid markers in an I-2-dependent man-

ner. The sequences of the first class corresponded to the

complete coding sequence of a 17-kDa small heat shock

protein of the alpha crystalline family (HSP17). The charac-

terization of this I-2 interactor will be reported elsewhere.

The second class of cDNAs represented a sequence enco-

ding the 55 kDa isoform of protein serine/threonine phos-

phatase 5 (PP5) (de la Fuente van Bentem et al., 2003). As

shown in Figure 1(a), this protein contains a N-terminal

tetratricopeptide repeat (TPR) domain, consisting of three

TPR motifs followed by a sequence that shares low homol-

ogy (H) with the TPR motif, a short spacer region, a type 5

protein serine/threonine phosphatase domain and a

C-terminal, putative nuclear localization signal. TPRs are

The I-2 complex contains HSP90 and PP5 285

ª Blackwell Publishing Ltd, The Plant Journal, (2005), 43, 284–298

generally involved in protein–protein interactions (D’Andrea

and Regan, 2003).

The interaction of PP5 with the LRR domain of I-2

prompted us to determine interactions of full-length PP5

with the LRR domains of other R proteins of the CC-NB-ARC-

LRR class in a yeast two-hybrid assay. To this end, LRR baits

were tested of two tomato I-2 homologues, I-2C-1 and I-2C-2,

tomato Mi-1.2, potato Rx and Arabidopsis RPM1 and RPS5.

The presence of bait and prey plasmids was selected on

plates lacking tryptophan and leucine ()WL) (Figure 1b).

Interactions were determined by monitoring colony growth

on selective plates additionally lacking histidine ()HWL) or

adenine ()AWL) over a period of 9 days. Protein baits alone

or the PP5 prey alone resulted in growth on )HWL or )AWL

plates (data not shown). When the LRR domain of I-2 was

used as bait, growth of colonies was clearly visible on both

)HWL and the more selective )AWL plates from day 2 on,

indicative for a strong interaction (Figure 1b). The LRR

domain of I-2C-1, I-2C-2, Mi-1.2 and RPM1 showed a good

interaction as well, although the slower growth on the

selective plates suggested that the interaction is weaker than

that with I-2 (Figure 1b). The LRR baits of Rx and RPS5

showed only a weak interaction (Figure 1b). As negative

control, the CC-NB-ARC part of I-2 was used and found not to

interact with PP5 (Figure 2). Apparently, PP5 interacts with

LRR domains of various R proteins in yeast, suggesting that

it is a general component of R protein complexes.

Determination of I-2 and PP5-binding domains

To resolve the region in I-2 that serves as docking domain for

PP5, full-length I-2 and different N- and C-terminal trunca-

tions were tested as baits for their interaction with full-length

PP5 in the yeast two-hybrid system (Figure 2). Proper

expression of the bait proteins was verified by Western

blotting (data not shown). Some truncations resulted in a

complete loss of interaction (LRR1–15, LRR12–19, LRR17–22

and LRR21–29), while others (LRR12–22, LRR12–27 and

LRR12–29) gave rise to an increased interaction when com-

pared with the complete LRR domain. Full-length I-2 did not

interact with PP5 (Figure 2). From these results it can be

concluded that LRRs 12–22 are sufficient for the interaction

with PP5 in the yeast two-hybrid system.

Four different truncations of PP5 were tested to determine

the region of the protein involved in the interaction with I-2

LRR12–29 (Figure 3). The TPR domain and spacer region

(a)

(b)

Figure 1. PP5 interacts with the LRR domains of different R proteins of the

CC-NB-ARC-LRR family.

(a) Domain structure of the PP5 protein. The tetratricopeptide repeat (TPR)

domain consists of three TPRs (1–3) and a sequence with low homology (H)

with a TPR motif. Phosphatase domain: protein serine/threonine phosphatase

domain. NLS: putative nuclear localization signal.

(b) Growth of yeast co-transformed with full-length PP5 as prey and LRR baits

derived from the R proteins indicated at the left. Growth on )WL control plates

(selects for presence of both bait and prey plasmids) was scored after 2 days.

Two-hybrid interactions were determined on selective )HWL and )AWL

plates, and growth on these plates was scored at the day indicated.

Figure 2. Determination of the I-2 region that binds to PP5. Full-length and

various truncated I-2 baits were tested for interaction with full-length PP5 in

the yeast two-hybrid system. The interaction strength was scored as positive

(þ to þþþþ) depending on the growth rate on selective )HWL and )AWL

plates. Lack of growth is defined as no interaction ()). Interactions were

scored after 9 days. The grey area indicates the I-2 region minimally required

for an interaction with PP5.

Figure 3. Identification of regions of PP5 necessary for the interactions with I-

2 and HSP90. Different parts of PP5 were tested for interaction with both I-2

(LRR12–29) and HSP90 (amino acids 643–699 of HSP90-2) in the yeast two-

hybrid system. The names of the truncated clones indicate amino acid

residues. The interaction strength was scored as positive (þ) based on the

growth rate on selective )HWL and )AWL plates, no growth was scored as no

interaction ()). Interactions were scored after 9 days. ND, not determined.

286 Sergio de la Fuente van Bentem et al.

ª Blackwell Publishing Ltd, The Plant Journal, (2005), 43, 284–298

(construct 1–214) showed a strong interaction with LRR12–

29, whereas the phosphatase domain does not interact

(construct 131–485). Truncation of the spacer region (con-

struct 1–151) weakened the interaction, suggesting that

amino acids in this region are involved in the interaction.

Further truncation of the C-terminus to amino acid 130

(construct 1–130) abolished the interaction completely,

indicating the presence of amino acids in the C-terminal

part of the TPR domain that are critical for the interaction.

This region alone, however, was not sufficient for the

interaction with I-2 (construct 131–485). In conclusion, the

TPR domain together with the spacer region is required and

sufficient for binding to the I-2 LRR domain.

PP5 interacts with HSP90 in the yeast two-hybrid system

Next, we searched for additional components of the I-2

protein complex by using PP5 as bait in yeast two-hybrid

screens. In total, 18 cDNAs were isolated from the cDNA

library and they were divided into three distinct classes

containing one, fourteen and three cDNAs respectively. The

first class represented a clone corresponding to a tomato

EST present in the TIGR database (TC51447) and its encoded

protein shows similarity with Arabidopsis proteins contain-

ing Agenet domains. Proteins containing Agenet domains

are potentially involved in chromatin remodelling (Maurer-

Stroh et al., 2003). The second and largest class of cDNAs

corresponded to HSP90-2 (previously termed Hsc80), which

encodes an 80 085 Da protein belonging to the family of

cytosolic HSP90s (Koning et al., 1992; Liu et al., 2004). The

three cDNAs of the third class encoded an 80 108 Da

homologue of LeHSP90-2, designated LeHSP90-1. The se-

quences of these homologues show an identity of 90 and

96% on nucleotide and amino acid level, respectively, and

both proteins show high levels of identity with Arabidopsis

and human HSP90 homologues (about 90 and 68%

respectively). All HSP90 cDNAs isolated in the screens were

derived from the 3¢-end of the mRNA. The shortest cDNA

picked up in the screen encoded the C-terminal 57 amino

acids of LeHSP90-2. This indicates that the C-terminus of

HSP90 is sufficient for the interaction with tomato PP5.

To determine the region of PP5 involved in binding to

HSP90, different truncations were tested for interaction with

the C-terminus of HSP90-2 (amino acids 643–699) in a yeast

two-hybrid assay. Both full-length PP5 and the complete TPR

domain (construct 1–151) interacted with HSP90 (Figure 3).

The C-terminally truncated TPR domain (construct 1–130)

interacted with HSP90 as well, and even showed an

increased interaction strength (Figure 3). This PP5 construct

did not interact with I-2 LRR12–29, showing that distinct

regions of PP5 are involved in the interaction with HSP90

and I-2.

PP5 and HSP90 interact with different regions of the LRR

domain of I-2 in vitro

To confirm the interaction between PP5 and the LRR domain

of I-2, we pursued an in vitro glutathione S-transferase (GST)

pull-down approach. To this end we produced in Escherichia

coli a GST fusion protein with the PP5 TPR domain (amino

acids 1–151), and a maltose-binding protein (MBP) fusion

protein with I-2 LRR12–29. Intact GST (as a control) and GST-

TPR were readily purified from E. coli (Figure 4a). In contrast,

substantial amounts of partially degraded fusion protein

were formed and co-purified with full-length MBP-I-2

LRR12–29 (Figure 4b, input lane). Purified GST or GST-TPR

was incubated with purified MBP-I-2 LRR12–29, and GST and

GST-TPR were subsequently pulled down using glutathione

Sepharose beads. To determine whether MBP-I-2 LRR12–29

co-purified with the beads, indicative for an interaction,

SDS-PAGE and Western blot analysis using anti-MBP anti-

bodies was performed. As shown in Figure 4(b), MBP-I-2

(a) (b) (c)

Figure 4. Distinct regions of the LRR domain of I-2 interact with PP5 and HSP90 in vitro.

(a) Coomassie-stained SDS-PAGE gel loaded with the amounts of GST, GST-TPR and GST-LeHSP90-2 that were used in the pull-down experiments shown in (b) and

(c).

(b) In vitro GST pull-down of the MBP-conjugated C-terminal half of the I-2 LRR domain (MBP-I-2 LRR12–29) with GST (negative control), GST-TPR or GST-LeHSP90-

2. The presence of MBP-I-2 LRR12–29 was detected on a Western blot (WB) using an anti-MBP antibody. In the input lane, 2% of the amount of MBP-I-2 LRR12–29 as

used for the pull-down assay was loaded.

(c) In vitro GST pull-down of the MBP-conjugated N-terminal half of the I-2 LRR domain (MBP-I-2 LRR1–11) with GST, GST-TPR and GST-LeHSP90-2. The presence of

MBP-I-2 LRR1–11 was detected on a WB using an anti-MBP antibody. In the input lane, 2% of the amount of MBP-I-2 LRR12–29 as used for the pull-down assay was

loaded.

The I-2 complex contains HSP90 and PP5 287

ª Blackwell Publishing Ltd, The Plant Journal, (2005), 43, 284–298

LRR12–29 was specifically pulled down by GST-TPR but not

by GST. Moreover, as the partially degraded forms of the

MBP-I-2 LRR12–29 fusion protein were not pulled down

efficiently (Figure 4b), it can be concluded that GST-TPR

binds the intact MBP-I-2 LRR12–29 fusion protein better than

its truncated forms. In a control experiment a fusion protein

containing MBP and the first 11 LRRs of I-2 (MBP-I-2 LRR1–

11) was used. This fusion protein could not be pulled down

by GST-TPR (Figure 4c). Together, these results confirm the

interaction found in the yeast two-hybrid system and show

that PP5 can interact directly and specifically with the

C-terminal part of the LRR domain (LRR12–29) of I-2.

In animal and yeast cells, PP5 is a co-chaperone of HSP90

and is present in HSP90-client heterocomplexes. This raised

the possibility that I-2 is a client of HSP90 and thus that both

proteins physically interact. In line with this idea, HSP90 was

recently found to interact with the LRR domain of N, a

tobacco disease resistance protein of the NB-ARC-LRR class

(Liu et al., 2004). To determine an interaction between

HSP90 and the I-2 LRR domain, an in vitro pull-down

approach as described above was pursued. Full-length

LeHSP90-2 was purified from E. coli as a GST fusion protein

(Figure 4a) and tested for interaction with MBP-I-2 LRR1–11

and with MBP-I-2 LRR12–29 fusion proteins. MBP-I-2 LRR1–

11 was pulled down with GST-LeHSP90-2 (Figure 4c),

whereas no interaction between I-2 LRR12–29 and HSP90

could be detected (Figure 4b). This indicates that HSP90 is

able to interact directly with the N-terminal region of the LRR

domain (LRR1–11) of I-2. Together, these data show that PP5

and HSP90 interact directly with distinct regions of the I-2

LRR domain.

HSP90 interacts with I-2 and PP5 in tomato lysates

To confirm the interaction between HSP90 and PP5, we set

out to co-purify HSP90 from the soluble fraction of tomato

stem lysate using the PP5 TPR domain (amino acids 1–151)

as bait. Purified GST or GST-TPR fusion protein (Figure 5a)

was immobilized on glutathione Sepharose beads and

subsequently incubated with the soluble fraction of a tomato

stem lysate. Proteins present in the GST-TPR pull-down

sample but absent in the control GST pull-down sample

(Figure 5b) were cut out from a Coomassie-stained SDS-

PAGE gel. These proteins were digested by trypsin and the

masses of the obtained peptides were determined by matrix-

assisted laser desorption/ionization time-of-flight mass

spectrometry. Most proteins, except one of approximately

80 kDa, yielded tryptic peptide masses that matched the

GST-TPR fusion protein. The 17 peptide masses of the

80 kDa protein matched LeHSP90-2 (30% coverage), but 16

of the 17 peptides matched LeHSP90-1 as well (Table 1). The

amino acid sequence of five tryptic peptides was confirmed

by MS/MS (Table 1), showing that the peptides were indeed

derived from HSP90. The identity of this protein was con-

firmed by Western blot analysis using antibodies raised

against recombinant HSP90. A protein with a molecular

mass comparable to that of HSP90 was detected in the GST-

TPR sample but not in the GST sample (Figure 5c). Together,

these data indicate that PP5 interacts with HSP90 in tomato

lysates.

Subsequently, we tested whether we could confirm the

interaction between HSP90 and I-2 by co-purifying HSP90

from the tomato stem lysate using recombinant I-2

(a) (b)

(d)

(e)

(c)

Figure 5. Purification of HSP90 from a tomato stem lysate using recombinant

fragments of PP5 and I-2 as baits.

(a) Coomassie-stained SDS-PAGE gel showing GST and GST-TPR proteins

purified from Escherichia coli.

(b) Pull-down of HSP90 from a tomato stem lysate. Purified GST and GST-TPR

proteins immobilized on beads were incubated with tomato lysate. Subse-

quently, GST and GST-TPR beads containing interacting proteins were

washed and the proteins were subjected to SDS-PAGE and Coomassie

staining. The panel at the right is a magnification of the protein band that was

identified as HSP90 by mass spectrometric analysis.

(c) Western blot (WB) analysis of GST and GST-TPR samples after incubation

with tomato lysate (as shown in b) using anti-HSP90 antibodies. A band of

approximately 80 kDa, corresponding to the expected mass of LeHSP90, was

specifically detected in the GST-TPR lane.

(d) Coomassie-stained SDS-PAGE gel showing GST-conjugated I-2 regions

purified from E. coli that were used for the pull-down experiments. CN, I-2 CC-

NB-ARC; LRR1–11, I-2 LRR1–11; LRR12–29, I-2 LRR12–29.

(e) Pull-down experiment from tomato lysate using the I-2 fusion proteins

displayed in (d). HSP90 was detected by WB using anti-HSP90 antibodies. A

band of approximately 80 kDa was specifically present in the GST-LRR1–11

lane.

288 Sergio de la Fuente van Bentem et al.

ª Blackwell Publishing Ltd, The Plant Journal, (2005), 43, 284–298

fragments as baits. For this purpose, GST-I-2 CC-NB-ARC

(GST-CN), GST-I-2 LRR1–11 (GST-LRR1–11) and GST-I-2

LRR12–29 (GST-LRR12–29) fusion proteins were used (Fig-

ure 5d). The fusion proteins were incubated with tomato

lysate. As shown by Western blot analysis using anti-HSP90

antibodies, HSP90 was specifically pulled down by GST-I-2

LRR1–11, but not by GST-CN nor by GST-I-2 LRR12–29

(Figure 5e), confirming that HSP90 interacts specifically with

the N-terminal region of the I-2 LRR domain.

Analysis of PP5 function in disease resistance

The physical interaction between PP5 and I-2 suggests that

they are present in one complex and that PP5 may be in-

volved in I-2-mediated signalling. Therefore, the role of PP5

in I-2-mediated resistance to F. oxysporum f. sp. lycopersici

(Fol) race 2 was assessed. A tomato cultivar containing the

I-2 gene (Mogeor) was stably silenced through Agrobacte-

rium tumefaciens-mediated transformation with a binary

vector containing the silencing construct [the cauliflower

mosaic virus 35S promoter, antisense PP5 cDNA, nopaline

synthase (NOS) transcriptional terminator]. T1 seedlings

were tested for altered I-2-mediated resistance. PP5 expres-

sion levels in lines 03 and 39 were 7 and 9%, respectively, of

that of the wild type (Figure S2). PP5-silenced tomato did not

show an obvious phenotype in comparison with wild-type

plants (data not shown). Wild type and PP5-silenced lines

were infected with Fol race 2. Both lines were as resistant to

this race as wild-type plants (Figure S2), indicating that PP5

is not essential for I-2-mediated resistance in tomato.

Because there was still residual PP5 expression in the

silenced tomato lines, we set out to identify a full pp5 knock-

out. As there are no extensive T-DNA insertion libraries

available in tomato we screened the Arabidopsis (ecotype

Ws-2) collection in Madison, WI. A pp5 null mutant contain-

ing a single T-DNA was isolated (line 39-6; Figure S2).

Similar to PP5-silenced tomato, the Arabidopsis pp5 null

plants did not show any obvious morphological abnormality

(data not shown). As PP5 interacts with RPM1, we assessed

RPM1-mediated resistance in line 39-6 and in Ws-2. Infection

assays with Pseudomonas syringae pv. tomato (Pto) DC3000

carrying avrRpm1 did not show an effect of PP5 knock-out on

RPM1 resistance (Figure S2). Therefore, we conclude that

PP5 has a minor or no effect on RPM1 resistance. Also basal

resistance to virulent Pto DC3000 (without avrRpm1) was not

significantly altered in pp5 null plants (Figure S2).

I-2 triggers cell death in N. benthamiana leaves

Cell death is often part of the defence response mediated by

R proteins. As PP5 is not essential for I-2- or RPM1-mediated

resistance in tomato and Arabidopsis, respectively, we

wanted to test whether PP5 and HSP90 are essential for

I-2-mediated cell death instead of disease resistance. To test

this idea, we set out to construct a mutant I-2 protein that

constitutively induces defence responses in the absence of

the elicitor. A single amino acid substitution (D460V) in the

MHD motif of Rx resulted in an elicitor-independent cell

death response when expressed in N. benthamiana leaves

(Bendahmane et al., 2002). I-2 containing the analogous

substitution (D495V) activated a similar cell death response

when expressed in N. benthamiana leaves (Figure 6b).

Agrobacterium tumefaciens-mediated expression of wild-

type I-2 gave no visible response (Figure 6a). One of the

characteristics of R protein-mediated responses is its

dependence on SGT1 and RAR1 (Liu et al., 2002; Peart et al.,

Table 1 Identification of LeHSP90-2 peptides by MALDI-TOF mass spectrometry analysis

Mr(expt)a Mr(calc)a Position SequenceMissedcleavage

Present inLeHSP90-1

Confirmedby MS/MS

958.52 958.53 264–270 QKPIWMR Yes1133.53 1133.58 472–481 AVENSPFLEK Yes Yes1168.58 1168.57 418–427 LGIHEDSQNR Yes1209.63 1209.63 317–326 RAPFDLFDTK 1 Yes1256.67 1256.70 88–99 ADLVNNLGTIAR Yes1291.58 1291.64 34–45 ELISNSSDALDK Yes Yes1304.59 1304.65 253–263 EVSNEWSLVNK Yes Yes1320.65 1320.66 299–309 HFSVEGQLEFK Yes1323.62 1323.61 180–189 EDQLEYLEER Yes Yes1527.81 1527.80 358–371 GIVDSEDLPLNISR Yes1541.77 1541.76 286–298 SLTNDWEEHLAVK Yes1616.68 1616.74 457–470 EGQNDIYYITGESK Yes Yes1734.95 1734.94 57–71 LDGQPELFIHIIPDK Yes1845.95 1845.93 271–285 KPEEITKEEYAAFYK 1 Yes2071.03 2071.04 174–189 MVLYLKEDQLEYLEER 1 No2426.12 2426.06 141–161 HNDDEQYVWESQAGGSFTVTR Yes2444.10 2444.07 563–583 VVDSPCCLVTGEYGWTANMER Yes

am/z value of singly-protonated peptides, with Cys modified by iodoacetamide.

The I-2 complex contains HSP90 and PP5 289

ª Blackwell Publishing Ltd, The Plant Journal, (2005), 43, 284–298

2002). Therefore, we tested by virus-induced gene silencing

(VIGS) of NbSGT1 and NbRAR1 whether the I-2 D495V cell

death response was dependent on SGT1 and/or RAR1. Three

weeks after infection via agroinfiltration of the empty pTV00

virus vector (TRV:00), the SGT1 silencing vector (TRV:SGT1)

or the RAR1 silencing vector (TRV:RAR1), the I-2 D495V

construct was agroinfiltrated into the upper leaves of the

silenced plants. As expected, leaves of the control TRV:00

plants became necrotic upon I-2 D495V expression

(Figure 6c), whereas silencing of NbSGT1 resulted in com-

plete suppression of I-2 D495V-mediated cell death

(Figure 6d). Silencing of NbRAR1 blocked I-2 D495V-medi-

ated cell death in most leaves (Figure 6e), whereas in some

leaves a faint cell death response was visible. These results

suggest that the I-2 D495V mutant triggers a cell death re-

sponse similar to that of other R proteins and that this sys-

tem may be used to study the involvement of PP5 and HSP90

in I-2-mediated signalling.

VIGS of PP5 and HSP90

To address the question whether PP5 and HSP90 are

required for I-2 D495V-mediated cell death we silenced their

homologues in N. benthamiana. For silencing of NbPP5,

(a) (b) (c)

(d)

(g) (h)

(i)

(e) (f)

1

Actin

PP5

2 3 4 5

1 2 3 4 5

TRV:00

TRV:PP5

TRV:00

TRV:00

TRV:PP5

TRV:HSP90

HSP90

HSP90 protein level (%) 100 91 98 37 47 55

Rubisco

Figure 6. Silencing of HSP90 but not of PP5 abrogates cell death triggered by I-2 in Nicotiana benthamiana. Expression of wild-type I-2 (a) and the I-2 D495V mutant

(b) in leaves of N. benthamiana. Expression of I-2 D495V in TRV:00 control plants (c) and in TRV:SGT1 (d), in TRV:RAR1 (e), TRV:HSP90 (f) and TRV:PP5 plants (g).

Three weeks after induction of silencing using the TRV constructs, Agrobacterium tumefaciens containing the I-2 D495V construct was infiltrated into the upper

leaves. Photographs were taken 7 days after infiltration.

(h) Silencing of NbPP5. RT-PCR on cDNA of TRV:00 and TRV:PP5 plants with NbPP5 primers (lower panel). As a control, the Actin gene was used (upper panel).

Fragments were amplified by 20, 25, 30, 35 or 40 cycles of PCR (lanes 1–5 respectively). Fragments were visualized on a 1% agarose gel containing ethidium bromide.

(i) Effect of silencing on total HSP90 protein levels. Western blot analysis with anti-HSP90 antibodies was performed on leaf extracts of three independent TRV:00

and TRV:HSP90 plants. The relative percentage (%) of HSP90 protein corrected for loading (see Rubisco lane of Coomassie-stained gel below) compared with lane 1

is indicated below each lane.

290 Sergio de la Fuente van Bentem et al.

ª Blackwell Publishing Ltd, The Plant Journal, (2005), 43, 284–298

a 344-bp NbPP5 fragment was amplified from N. bentha-

miana cDNA and cloned into the pTV00 vector (Ratcliff et al.,

2001). First, transcript levels of NbPP5 were determined in

TRV:00 and TRV:PP5-inoculated leaves by reverse tran-

scriptase-PCR analysis using NbPP5-specific primers. The

level of NbPP5 transcript in the TRV:PP5-infected plants was

about 60% lower than that of the TRV:00-infected plants. As a

control for cDNA quality an RT-PCR using Actin primers was

carried out which shows that equal amounts of cDNA were

used (Figure 6h, upper panel). The lower level of PP5 tran-

script in TRV:PP5 plants indicated that silencing occurred,

but was partial (Figure 6h, lower panel, lanes 1–5). Silencing

of NbPP5 did not result in any obvious phenotype (data not

shown). Three weeks after VIGS of PP5, the I-2 D495V con-

struct was agroinfiltrated into the upper leafs. The I-2 D495V

response in PP5-silenced plants was similar to that observed

in TRV:00 control plants (Figure 6c,g).

For silencing of NbHSP90, a 502-bp fragment of LeHSP90-

2 was cloned into pTV00. Based on the sequence identity

with the LeHSP90-2 fragment it is likely that both the

N. benthamiana NbHSP90-1 and NbHSP90-2 homologues

would be silenced. Silencing was determined by Western

blot analysis using anti-HSP90 antibodies. The level of total

HSP90 protein was reduced by about 50% in TRV:HSP90

plants compared with TRV:00 plants (Figure 6i). Tomato

(TIGR, release date 17 April 2003) contains at least two other,

more distantly related, cytosolic HSP90 homologues. The

low sequence identity of these two tomato homologues with

the silencing fragment makes it unlikely that the two

corresponding NbHSP90 homologues (Lu et al., 2003) were

silenced in our experiment. This is reinforced by the notion

that still 50% of the total level of HSP90 protein was present

in TRV:HSP90 leaves (Figure 6i). Two weeks after initiation

of VIGS of HSP90, phenotypic symptoms became visible and

after 3 weeks plants were stunted and leaf morphology was

altered compared with TRV:00 plants (Figure S3). Three

weeks after VIGS of HSP90, the I-2 D495V construct was

agroinfiltrated into the upper leafs. Partial silencing of

HSP90 completely abolished cell death activated by I-2

D495V (Figure 6f), whereas in TRV:00 control plants a cell

death reaction was observed (Figure 6c). This shows that

HSP90 is required for I-2 function.

Discussion

Identification of components of the I-2 protein complex

The aim of our study was to identify components of the

tomato I-2 protein complex. A combination of yeast two-

hybrid screens, in vitro GST pull-downs and GST pull-downs

from tomato lysates resulted in the identification of PP5,

HSP17 and HSP90 as candidate components. In tomato PP5

is a single-copy gene encoding two protein isoforms of

which one is localized in the endoplasmic reticulum and the

other in the cytosol and nucleus. The isoform identified as I-2

interactor is the one that is predominantly found in the

cytosol (de la Fuente van Bentem et al., 2003). PP5 directly

interacts with the C-terminal region of the I-2 LRR domain.

Targeted two-hybrid studies showed that PP5 binds to the

LRR domains of various R proteins of the CC-NB-ARC-LRR

class from several plant species. Interestingly, in a yeast

two-hybrid screen PP5 was also identified as an interactor of

the LRR domain of the TIR-NB-ARC-LRR protein N (Liu et al.,

2004). This suggests that PP5 could be a common interactor

of R proteins of the NB-ARC-LRR class. In both the yeast two-

hybrid system and in vitro, PP5 specifically bound to the

C-terminal region, but not to the N-terminal region of the

I-2 LRR domain.

By testing recombinant proteins, we show that HSP90

directly interacts with the N-terminal region of the I-2 LRR

domain, whereas no binding to either the CC-NB-ARC part or

to a C-terminal region of the LRR domain was detected. In

addition, using the N-terminal I-2 LRR region we were able to

specifically co-purify HSP90 from a tomato lysate. Unfortu-

nately, we were unable to pull down sufficient HSP90 for

mass spectrometric analysis; therefore we could not identify

which HSP90 isoform interacts with I-2. HSP90 likely inter-

acts with the LRR domains of R proteins in general, because

also for N and MLA proteins HSP90 was found to interact

with this domain (Bieri et al., 2004; Liu et al., 2004). Alto-

gether, our interaction and silencing data indicate that I-2 is a

client of HSP90. As other R proteins are HSP90 clients as well

(Hubert et al., 2003; Liu et al., 2004; Schulze-Lefert, 2004),

this indicates that HSP90 is broadly involved in the regula-

tion of R protein function.

PP5 interacts with HSP90 in plants

In yeast and animal cells PP5 interacts with HSP90 (Chen

et al., 1996; Dobson et al., 2001; Gavin et al., 2002). So far,

evidence for an interaction between these two proteins in

plants was lacking. Here we have provided evidence

showing that tomato PP5 interacts with HSP90 using both

the yeast two-hybrid system and affinity purification from

tomato lysates. Two HSP90 isoforms, LeHSP90-1 and

LeHSP90-2, were identified as interactors of PP5. The

shortest clone isolated in the yeast two-hybrid screens

encodes residues 643–699 of LeHSP90-2, indicating that

this short C-terminal region is sufficient for binding PP5.

This finding suggests that the interaction with the TPR

domain of PP5 occurs via the C-terminus of HSP90 in

tomato. Indirect evidence indicates that the TPR domain

of rat PP5 binds to wheat HSP90 (Zhang et al., 2003). A

similar interaction has been found for PP5 and HSP90 in

animal cells (Chen et al., 1996; Russell et al., 1999).

Apparently, the interaction between PP5 and HSP90

observed in yeast and animal cells is conserved in tomato

and likely in other plant species.

The I-2 complex contains HSP90 and PP5 291

ª Blackwell Publishing Ltd, The Plant Journal, (2005), 43, 284–298

In animal cells, HSP90 is specialized in moulding pre-

folded signalling proteins such as the glucocorticoid recep-

tor (GR) into a mature state (Pratt and Toft, 2003). The folding

of clients by HSP90 is accompanied by dynamic heterocom-

plex assembly involving different proteins. Among these

proteins are immunophilin co-chaperones that can be

divided in FK506-binding proteins (FKBPs) and cyclosporin

A-binding proteins (cyclophilins). Characteristic of these

immunophilins is that they interact with the C-terminus of

HSP90 via their TPR domain, like PP5. They compete with

each other, and PP5, for entering HSP90 heterocomplexes at

a late stage of client folding. In plants, a few homologues of

mammalian FKBPs have been identified and these have

been shown to interact with HSP90 (Kamphausen et al.,

2002; Reddy et al., 1998). Mammalian PP5 is a co-chaperone

present in heterocomplexes containing HSP90 and either the

GR, haem-regulated eIF2a kinase (HRI) or the OR, which are

all HSP90 clients (Chen et al., 1996; Ikeda et al., 2004; Shao

et al., 2002; Silverstein et al., 1997). Although the exact

mode of action is unclear, PP5 suppresses responses

mediated by the GR, HRI and the OR, possibly by dephospho-

rylation of these proteins (Chinkers, 2001; Ikeda et al., 2004;

Shao et al., 2002). A functional HSP90 co-chaperone machin-

ery capable of folding the mammalian GR is present in plant

cells (Pratt et al., 2001). This observation, together with our

results and the previous identification of other TPR co-

chaperones in plants, strongly suggest that there is an

HSP90 co-chaperone machinery in plants that is functionally

conserved and involves similar components as in animal

cells (Pratt et al., 2001; Zhang et al., 2003).

Possible role of HSP90 in R protein function

Specific Rx mutants trigger cell death in the absence of its

elicitor (Bendahmane et al., 2002). We have made one of

these mutations in the I-2 protein. As I-2 confers resistance to

a root-invading pathogen, it was somewhat surprising that

the D495V mutation in I-2 was able to activate a cell death

response in N. benthamiana leaves. This mutation is located

in the MHD motif in the ARC region, which is conserved in the

NB-ARC-LRR class of R proteins. The fact that this mutation in

both Rx and I-2 leads to activation of the proteins suggests

that the function of the MHD motif is conserved in R proteins.

Our results show that HSP90 is required for I-2-mediated

cell death in N. benthamiana. All other plant R proteins

belonging to the NB-ARC-LRR class that have been tested so

far depend on HSP90 function as well (Hubert et al., 2003; Liu

et al., 2004; Lu et al., 2003; Takahashi et al., 2003). Disruption

of HSP90 function reduces the level of different R proteins,

suggesting that the accumulation of R proteins is dependent

on HSP90 function (Hubert et al., 2003; Lu et al., 2003).

Unfortunately, we have neither been able to detect the native

I-2 protein in N. benthamiana nor has either N-terminal or

C-terminal tagging been successful. Thus, we are unable to

test whether the I-2 protein level is decreased in HSP90-

silenced plants. Like the cell death response mediated by

other R proteins tested in N. benthamiana (Peart et al., 2002),

cell death triggered by I-2 D495V depends on SGT1.

Furthermore, we show that RAR1 is also required for full

HR triggered by I-2 D495V. The partial dependence of I-2 on

RAR1 function is similar to the partial reduction of Pto/Prf-

mediated resistance by RAR1 silencing (Ekengren et al.,

2003). RAR1 is involved in stabilization/accumulation of R

proteins (Bieri et al., 2004; Tornero et al., 2002), and over-

expression of I-2 in N. benthamiana might cause a partial

bypass of RAR1 dependence in comparison with the in vivo

situation.

PP5 and HSP90 are potentially general components of R

protein complexes

Our data show that inhibition or complete abrogation of PP5

expression in tomato and Arabidopsis, respectively, does

not lead to an obvious phenotype (data not shown). Like-

wise, a knock-out of the PP5 homologue Ppt1 in yeast does

not give a phenotype either (Chen et al., 1994). This dem-

onstrates that the single-copy PP5 gene is dispensable in

yeast and plants under normal growth conditions. Silencing

of PP5 did not significantly inhibit I-2-mediated disease

resistance nor cell death. Simultaneous overexpression of

PP5 and I-2 D495V did not clearly affect severity or timing of

cell death in N. benthamiana (data not shown). So, the effect

of PP5 on the regulation of I-2 by HSP90 appears to be minor

or redundant. Also, RPM1 resistance in Arabidopsis was not

significantly altered in an Arabidopsis pp5 knock-out plant.

However, we do not rule out the possibility that PP5 could

have an effect on specific R protein-mediated resistance in

plants. In animal cells, both immunophilin co-chaperones

and PP5 are dispensable for the folding of clients by HSP90

into a signalling-competent state (Pratt and Toft, 2003).

Although HSP90 is essential in yeast, a yeast strain in which

all immunophilin TPR co-chaperones are deleted is still

viable, indicating that the total capacity of HSP90 is not

entirely dependent on these immunophilins (Dolinski et al.,

1997). Apparently, PP5 and each of the immunophilins in

HSP90-client heterocomplexes are involved in fine-tuning of

the maturation of the client. As immunophilin co-chaper-

ones compete with PP5 for binding to HSP90, reduction in

PP5 levels could be compensated for by other immunophi-

lins. Mammalian PP5 and immunophilin TPR co-chaperones

are present in distinct HSP90-client pools through a direct

interaction with the client after entering the complex via

binding to HSP90 (Ikeda et al., 2004; Pratt and Toft, 2003;

Shao et al., 2002). In analogy with mammalian systems,

their interactions with R proteins suggest that HSP90 and

PP5 are components of a chaperone heterocomplex

involved in the transformation of R proteins into a signal-

ling-competent state. Interestingly, sequences located in

292 Sergio de la Fuente van Bentem et al.

ª Blackwell Publishing Ltd, The Plant Journal, (2005), 43, 284–298

between the three TPR motifs and the catalytic domain of

PP5 are essential for the interaction with I-2 but not with

HSP90, suggesting that this region is involved in selecting

the PP5 co-chaperone for its presence in R protein-HSP90

heterocomplexes. Because in animal cells PP5 seems to

target HSP90 clients, we speculate that PP5 in plants dep-

hosphorylates R proteins within HSP90-R protein hetero-

complexes.

Despite extensive attempts to identify a complex consist-

ing of I-2, HSP90 and PP5 in vivo and/or in vitro we have not

been able to show its presence. An explanation for this could

be that due to its dynamic behaviour the I-2/PP5/HSP90

protein complex only exists transiently (Pratt et al., 2004;

Shao et al., 2002), resulting in too low amounts for detection

on Western blots. An alternative explanation is the selective

binding of the interactors to the distinct parts of the LRR. In

our pull-down experiments we have used specific parts of

the LRR that either bind HSP90 or PP5. Possibly by using the

full-length I-2 LRR domain (LRR 1–29) that contains both

binding domains, the presence of all three proteins in one

complex can be shown. Unfortunately, however, we could

not do this experiment as we are unable to express the full-

length LRR domain in E. coli.

As LRR domains of R proteins are so important for the

regulation of signalling, it is interesting that we have found

two proteins that bind to the LRR domain. We speculate that

both are potentially involved in the regulatory function of

this domain. Our data and previous findings point at

similarities between the maturation of R proteins by HSP90

and GR folding by the HSP90 chaperone machinery in

mammalian cells, and thus we might learn from the mech-

anism of HSP90-mediated GR folding (Hubert et al., 2003;

Schulze-Lefert, 2004). The identification of PP5 as a protein

interacting with both R proteins and HSP90 provides

evidence for the idea that R proteins are clients of an

HSP90 folding machinery related to that present in animal

cells. The interactions of plant R proteins with HSP90 and its

co-factors such as PP5, RAR1 and SGT1 suggest that R

proteins are regulated by dynamically assembling and

disassembling HSP90 heterocomplexes to achieve optimal

control of both R protein maturation and signalling.

Experimental procedures

cDNA library construction

To generate cDNA libraries, we used RNA isolated from tomatoplants (Lycopersicon esculentum cultivar GCR161; Kroon andElgersma, 1993) that had been infected with F. oxysporum formaspecialis lycopersici race 2 isolate Fol007. RNA was isolated fromroot and stem material of 23-day-old seedlings 13 days post-inoculation. Total RNA was isolated as described (Rep et al., 2003).Poly(A)þ RNA was isolated from 5 mg of total RNA using oligo-dTagarose beads (Boehringer Mannheim, Mannheim, Germany).Twenty micrograms poly(A)þ RNA was used to synthesize cDNA

using the ZAP cDNA synthesis kit (Stratagene, La Jolla, CA, USA)and Superscript reverse transcriptase (Promega, Madison, WI,USA). Gel filtration fractions representing large (>1.5 kb) and small(<1.5 kb) cDNAs were directionally cloned downstream of the GAL4transcription activation domain into the EcoRI and XhoI sites fromthe Lambda-ACT vector described by Elledge et al. (1991). Ligationproducts were packaged into phage particles using the Gigapack IIIgold cloning kit (Stratagene). This primary library was used to infectE. coli strain XL1-blue-MRA. The fraction of large cDNAs gave rise toa library that consisted of eight million independent clones with anaverage insert length of 2.1 kb. The fraction of smaller cDNAs gaverise to the library used in this study and consisted of two millionindependent clones with an average insert length of 1.1 kb. Thisprimary library was eluted from the plates giving rise to the sec-ondary library. Lambda phages of this secondary library were con-verted into yeast shuttle vectors using an ‘in vivo mass excision’protocol (Elledge et al., 1991). This protocol makes use of the Crerecombinase that is expressed in E. coli strain BNN132, whichrecognizes the lox sites flanking the pACT2 insert in LambdaACT2.

Bait construction

For the I-2 baits we used I-2 genomic DNA (cosmid A55; Simonset al., 1998) as starting material. Because introns are absent in theI-2 coding region its genomic DNA is suitable for expression inyeast. An NcoI site was engineered at the start codon of I-2, whichdid not result in amino acid changes. Bait I-2 LRR1–29 (amino acids520–1250) was constructed by cloning the NdeI–MscI fragment intopAS2-1 (Clontech Laboratories, Palo Alto, CA, USA). Bait CC-NB-ARC (amino acids 1–519) was constructed by cloning the NcoI–NdeIfragment of I-2 into pAS2-1. Bait LRR12–29 (amino acids 823–1250)was constructed by cloning the EcoRI–MscI fragment of I-2 intopAS2-1. Bait LRR12–22 (amino acids 823–1092) was constructed bycloning the EcoRI–BspHI fragment of I-2 into pAS2-1. Bait LRR17–29(amino acids 940–1250) was constructed by cloning the EcoRV–MscIfragment of I-2 into pAS2-1. Bait LRR21–29 (amino acids 1027–1250)was constructed by cloning the SspI–MscI fragment of I-2 into pAS2-1. Bait LRR1–15 (amino acids 520–939) was constructed by cloningthe NdeI–EcoRV fragment of I-2 into pAS2-1. Bait LRR12–27 (aminoacids 823–1206) was constructed by cloning the EcoRI–PvuII frag-ment of I-2 into pAS2-1. Bait LRR12–19 (amino acids 823–1026) wasconstructed by cloning the EcoRI–SspI fragment of I-2 into pAS2-1.Bait LRR17–22 (amino acids 940–1092) was constructed by cloningthe EcoRV–BspHI fragment of I-2 into pAS2-1.

For the I-2C-1 bait we used cosmid A29 (Simons et al., 1998). ADNA fragment encoding amino acids 831–1204 was amplified usingthe primers FP173 (GAAGAATTCTATGGCAGATTG) and FP172(GGGCCAGTATTCCCCCTTGTC) and cloned into the pAS2-1 vectorusing the same restriction sites as were used for the correspondingI-2 fragments (bait LRR12–29). For I-2C-2, a fragment was amplifiedfrom cosmid CC14 (Simons et al., 1998) with primers FP173 andFP172 encoding amino acids 822–1224 and cloned in pAS2-1identical to I-2C-1. For the RPM1 bait we used a plasmid containingthe RPM1 cDNA as starting material. Using the EaeI site flanking theinsert and the endogenous PstI site the C-terminal LRR domain ofRPM1, encoding amino acids 533–926, was cloned in-frame down-stream of the GAL4 DNA-binding domain from the pAS2-1 vector.For the Rx bait we used a plasmid containing the Rx cDNA. Afragment encoding the LRR domain was amplified using primersFP368 (TGCCATGGCCATGAATTTTGTGAATG) and FP369(TGCTCGAGCTACTCGACATTATTG). The PCR product was diges-ted using NcoI and XhoI. A fragment encoding amino acids 473–789from Rx was cloned in the pAS2-1 vector. For RPS5, a fragment wasamplified from a plasmid containing RPS5 with primers FP240

The I-2 complex contains HSP90 and PP5 293

ª Blackwell Publishing Ltd, The Plant Journal, (2005), 43, 284–298

(CACCCGGGAGTCAAGGATTGGAACACTG) and FP220 (CAGTCGA-CTTATGTTTCTCTCCACCGC), cut with SmaI and SalI and cloned inpAS2-1, resulting in a clone encoding amino acids 513–889 of RPS5.For the Mi-1.2 bait, a fragment was amplified from a plasmidcontaining Mi-1.2 ORF (pSE23) with FP210 (CATGCCATGGATGAC-TTTTGTTTGATAAAAGCAAG) and FP211 (CTGGTCGACCAACCA-TAATGCTACTTAAATAAGG), digested with NcoI and SalI andcloned into pAS2-1. This clone encoded amino acids 845–1257.The sequences of all inserts were verified to exclude PCR-inducederrors. Expression of bait constructs was verified by Westernblotting using antibodies against the Gal4 DNA-binding domain.

Yeast two-hybrid assays and cDNA library screens

The host strain for the two-hybrid assay used in this study is thePJ69-4a strain (James et al., 1996). This strain was grown on min-imal medium [MM: 2% glucose, 0.17% yeast nitrogen base, 0.5%(NH4)2SO4, 0.25% succinic acid, pH adjusted to 5.5 with NaOH]supplemented with 0.002% Uracil (U), 0.004% D/L-methionine (M),0.01% L-leucine (L), 0.002% L-tryptophane (W), 0.002% adeninesulphate (A) and 0.002% L-histidine (H). If selection for an auxotro-phy marker was desired, one or more of the above components wasomitted [MM ())A, H, W, L]. PJ69-4a harbours three reporter genes(ADE2, HIS3 and LacZ) that can be transcriptionally activated uponreconstitution of the Gal4 transcription factor. PJ69-4a was trans-formed with plasmids that contain bait (pAS2-1) and prey (pACT2)constructs using the ‘high efficiency transformation protocol’ des-cribed by Gietz et al. (1992). Tryptophane and leucine prototrophictransformants were selected on MM )WL plates.

In order to identify proteins that interact with I-2, a cDNA two-hybrid library was screened using the I-2 bait LRR1–29. PJ69-4atransformants expressing the bait construct were supertransformedwith library plasmids. Cells were plated on MM )HWL plates andgrown at 30�C for 3 days and an additional 5 days at roomtemperature. For screening for PP5 interactors, PJ69-4a containingthe pAS2-1 bait plasmid with a LePP5 TPR domain construct(encoding amino acids 1–151) or full-length LePP5 were trans-formed with the two-hybrid cDNA library. For the PP5 screens, cellswere plated on minimal medium MM )HWL and grown for 10–15 days at room temperature. All HWLþ colonies were transferredto MM )WL plates and subjected to the two-hybrid assays describedbelow. Prey plasmids were discarded as false positives when theyshowed activation of the two-hybrid markers when combined withan empty bait plasmid. From the screens with the I-2 LRR domain,6 · 106 WLþ colonies were obtained. From the screens with the PP5TPR domain and full-length PP5, 2 · 103 and 2 · 105 WLþ trans-formants were obtained respectively.

For yeast two-hybrid assays, yeast strains containing both thebait and the prey vector were spotted on MM )WL, MM )AWL andMM )HWL plates and after 2 days of growth at 30�C they weretransferred to room temperature. The MM )WL plate was subjectedto X-gal staining (Duttweiler, 1996) for detection of LacZ activity.Interaction strengths were scored visually based on growth onselective plates and are indicated as negative ()) or positive (from þto þþþþþ) depending on the growth rate.

GST and MBP fusion protein expression and purification

GST, GST-TPR (amino acids 1–151 of LePP5), GST-CN (amino acids1–519 of I-2), GST-LRR1–11 (amino acids 520–840 of I-2), GST-LRR12–29 (amino acids 823–1250 of I-2) fusion proteins wereexpressed from pGEX-KG (Guan and Dixon, 1991). MBP-I-2 LRR1–11and MBP-I-2 LRR12–29 were expressed from pMal-c2X (New Eng-

land Biolabs, Beverly, MA, USA). GST and MBP fusion proteinswere overexpressed at 37�C and isolated from E. coli strain BL21(DE3) by immobilization on glutathione Sepharose 4B beads(Amersham Biosciences, Uppsala, Sweden) and on amylose resin(New England Biolabs) respectively.

For in vitro pull-down assays, immobilized GST fusion proteinswere eluted with 50 mM Tris, pH 8.0, containing 20 mM reducedglutathione and MBP fusion proteins were eluted with PBS (2.7 mM

KCl, 6.5 mM Na2HPO4, 1.5 mM KH2PO4, 137 mM NaCl and completeprotease inhibitor cocktail; Roche, Basel, Switzerland) containing10 mM maltose. Fusion proteins were then dialysed against 50%glycerol (v/v), 50 mM Tris, pH 7.5, 100 mM NaCl, 5 mM DTT andstored at )20�C.

For pull-down assays from tomato lysate, fusion proteinsimmobilized on beads were washed two times in PBS. To removeE. coli proteins cross-reacting with anti-HSP90 antibodies, beadswere rotated for 30 min at 4�C in PBS supplemented with 500 mM

NaCl. After washing the beads twice with PBS supplemented with500 mM NaCl and three times in PBS without extra NaCl, SDS-PAGE and Coomassie brilliant blue staining was performed. Todetermine the concentration of the purified GST fusion proteins,samples were compared with a BSA concentration range on thesame gel.

In vitro GST pull-down assays

For pull-down assays, 1 lg of GST or GST fusion protein was usedand mixed with MBP fusion protein in interaction buffer (150 mM

NaCl and 50 mM Tris, pH 7.5, complete protease inhibitor cocktailand 0.2% v/v Tween-20). After addition of glutathione Sepharosebeads, immobilized proteins were washed three times with inter-action buffer. To determine protein interactions, samples weresubjected to SDS-PAGE and subsequent Western blotting.

GST pull-down from tomato lysates

The soluble fraction of tomato stem lysate was isolated as previ-ously described (de la Fuente van Bentem et al., 2003), except thatcentrifugation was performed at 13 000 g. For pull-down assaysusing PP5 as bait, control GST (15 lg) or GST-TPR (10 lg) immo-bilized on glutathione Sepharose beads was applied to the stemlysate (containing approximately 500 lg of protein) and rotatedovernight at 4�C in the presence of 0.05% Triton X-100. Beads werewashed five times with interaction buffer lacking Tween-20 andresuspended in SDS-PAGE loading mix [final concentrations:66 mM Tris-HCl pH 6.8, 3% (w/v) SDS, 5% (v/v) glycerol, 2% (v/v)beta-mercaptoethanol, and 0.001% (w/v) bromophenol blue]. Afterboiling for 5 min, proteins were separated by SDS-PAGE and sub-sequently stained with Coomassie brilliant blue. Isolation and pre-paration of proteins from gel for mass spectrometric analysis wereperformed as described by Rep et al. (2002).

For pull-down assays using I-2 fragments as baits, 15 lg of eachGST fusion protein was added to the tomato lysate containing500 lg of protein and rotated overnight at 4�C in the presence of0.5% (v/v) Tween-20. Protein complexes were washed four timeswith PBS containing 0.5% (v/v) Tween-20. To determine proteininteractions, samples were subjected to SDS-PAGE and subsequentWestern blotting.

Generation of antibodies against recombinant HSP90

A fragment encoding amino acids 1–351 of LeHSP90-2 was cloned inthe NcoI and XhoI sites of the pGEX-KG vector. Protein purification

294 Sergio de la Fuente van Bentem et al.

ª Blackwell Publishing Ltd, The Plant Journal, (2005), 43, 284–298

from E. coli and GST tag removal was performed as described be-fore (de la Fuente van Bentem et al., 2003) and used for immun-ization of rats. The obtained serum recognized the GST-LeHSP90-2fusion protein purified from E. coli. In lysates of tomato, Arabidopsisand N. benthamiana, the serum, but not the pre-immune serum,cross-reacted with proteins of about 80 kDa a molecular massexpected for cytosolic HSP90 (data not shown).

SDS-PAGE and Western blotting

Proteins were separated on 8% acrylamide gels by SDS-PAGE andtransferred to polyvinylidene difluoride (PVDF) membrane. Specificprimary anti-HSP90 (1:5000) antibodies (see above), anti-MBP(1:10 000) antibodies (New England Biolabs), peroxidase conju-gated secondary antibodies (Pierce, Rockford, IL, USA), and ECLplus Western Blotting Detection System (Amersham Biosciences)were used for detection.

Analysis of PP5 transcript levels in tomato and Arabidopsis

For Northern blot analysis,15 lg of total RNA was size fractionated byagarose gel electrophoresis. Preparation of the gels and transfer ofthe fractionated nucleotides to nylon membranes (Hybond Nþ;Amersham Biosciences) followed standard procedures. Probe DNAfragments were labelled with a[32P]-dATP using the Decamer labe-ling kit (MBI Fermentas, Vilnius, Lithuania). For detection of endog-enous LePP5 transcript in transgenic tomato plants we used a strand-specific labelling protocol in order to prevent detection of theheterologous antisense RNA. LePP5 cDNA fragment (25 ng) wasused in the labelling reaction. In addition, the labelling mixturecontained 5 pmol of FP80 (GTGGTACCAAATTCGACTTATCAAAGC),2.5 ll of 10x Taq buffer, 2.5 ll of dCTG (0.2 mM) and 3 ll of a[32P]-dATP. The mixture was adjusted to 25 ll and 0.5 ll of Taq polym-erase (2.5 U) was added. LePP5 probe DNA was labelled by primerextension at 94�C for 4 min, 55�C for 1 min and 94�C for 45 sec. Thelast two steps were repeated 16 times. As a control, a petuniaGAPDHprobe was used that is able to hybridize to LeGAPDH.

For analysis of AtPP5 transcript, RT-PCR was performed withprimers FP (atctcgagTCCTCTCCCGCTGTCGCCAAGATTGAATC) andFP239 (TTGGATCCTCGAGTTTTTTTTTTTTTTTTTTV; V represents A,C or G) that amplify the complete ORF. Control RT-PCRs wereperformed with AtGAPDH primers FP360 (GTCCAATGGCTGACAA-GAAGATTAGGATCGGA) and FP361 (CTTAGGCCTTTGACATGTG-GACGATCAA).

Silencing of the PP5 gene in tomato

The tomato I-2-containing cultivar Mogeor was stably transformedby A. tumefaciens strain EHA105 with binary vector pMOG402 inwhich a cassette was cloned that contained the silencing construct[the cauliflower mosaic virus 35S promoter, antisense PP5 cDNA(part of the 5¢ UTR, the complete ORF and the 3¢ UTR), NOS tran-scriptional terminator] between the T-DNA left and right border.Transgenic, kanamycin-resistant callus was selected and regener-ated to tomato plants (T0), essentially as described (Fillatti et al.,1987). Approximately 30 independent diploid transformed plantswere selected and tested for the expression of the endogenous PP5gene. Eighteen of 31 Mogeor T0 plants showed <50% expression ofPP5 (relative to GAPDH expression levels to correct for loadingdifferences) when compared with non-transgenic plants. Seedswere collected from only 10 plants with PP5 silencing, as eightplants were sterile. Lines 03 and 39 showed the best silencing.

Infection assays of tomato and Arabidopsis

Twenty tomato seedlings (11-day old) were inoculated by eitherF. oxysporum forma specialis race 2 (isolate 007) or water by theroot-dip method (Wellman, 1939). Three weeks after inoculation,plants were cut off at the cotyledons and their weight was deter-mined. Infection assays were performed at least twice.

Mutants in the Arabidopsis PP5 gene were identified by a PCR-based screen of ecotype Wassilewskija-2 (Ws-2) T-DNA insertionmutants generated at the University of Wisconsin Arabidopsisknock-out facility (Krysan et al., 1996). Arabidopsis Ws-2 was usedas a wild-type control. Plants were grown and assayed in acontrolled-environment chamber at 21�C, 70% relative humidityon a 13 h light/11 h dark cycle.

Pseudomonas syringae pv. tomato strain DC3000, either carryingavrRpm1 (on a plasmid) or not (kindly provided by Justin Lee,Leibniz Institute of Plant Biochemistry, Halle, Germany), wascultured in King’s B medium (KB). Leaves were infiltrated with asuspension of bacteria in 10 mM MgSO4 at an OD600 of 0.005 byusing a needle-less syringe to force the suspension through thestomata on the abaxial side of leaves. Bacterial titres were deter-mined by cutting leaf discs with a cork borer, grinding them in10 mM MgSO4, and plating serial dilutions on KB agar platescontaining 25 lg ml)1 rifampicin.

VIGS in N. benthamiana

A 502-bp SspI–StuI fragment of LeHSP90-2 (positions 1055–1556downstream of the ATG translation initiation codon) was clonedinto the pTV00 silencing vector (Ratcliff et al., 2001). For silencing ofNbPP5, a fragment of 344 bp (corresponding to nucleotides 1–344 ofaccession no. AY569436) was amplified with the PP5 primers FP351(5¢-GCTATTGATCTGTACACA-3¢) and FP162 (5¢-ATGGATCCC-TAAGAGTCAGCTACTGAACG-3¢) using cDNA synthesized fromN. benthamiana leaves as a template. The PCR fragment was clonedinto the pGEM-T easy vector (Promega). An ApaI–SpeI fragmentwas excised and cloned into the same sites of the pTV00 vector.VIGS was performed as described before (Ratcliff et al., 2001). Forsilencing of NbSGT1, the TRV:SGT1 vector was used (Peart et al.,2002). NbRAR1 was silenced with the TRV:RAR1 vector (kindlyprovided by G. Martin) that has been described (Liu et al., 2002).

To determine silencing of NbPP5, total RNA isolated from upperleaves 3 weeks after VIGS was used to make cDNA as describedbefore (de la Fuente van Bentem et al., 2003). RT-PCR on cDNA wasperformed with NbPP5 primers FP351 and FP352 (5¢-AAA-GACCCTCTATCAACA-3¢), which anneal outside the region usedfor silencing. Control RT-PCRs were carried out with Actin primersFP237 (5¢-GGGATGATATGGAGAAGATC-3¢) and FP239 (5¢-TTGGATCCTCGAGTTTTTTTTTTTTTTTTTTV-3¢; V ¼ A, C or G).Silencing of NbHSP90 was determined on extracts made from theupper leaves 3 weeks after VIGS. For this purpose, soluble fractionsof extracts were isolated by centrifugation at 10 000 g as described(de la Fuente van Bentem et al., 2003), and subjected to Western blotanalysis. Quantification of transcript and protein levels was per-formed in Adobe Photoshop by histogram analysis.

Leaf infiltration assays

To construct binary vectors containing full-length I-2 or I-2 D495V,an SpeI–SalI linker consisting of FP778 (TCGAAGATC-TCTTACTCGAGGGCCCATGGA) and FP779 (CTAGTCCATGG-GCCCTCGAGTAAGAGATCT) was cloned in the SpeI and XhoI sitesof pBS KSþ to yield plasmid WP33. An NcoI–XhoI fragment

The I-2 complex contains HSP90 and PP5 295

ª Blackwell Publishing Ltd, The Plant Journal, (2005), 43, 284–298

containing the complete I-2 ORF was cloned into the same sites ofWP33 to yield WP36. An I-2 SpeI–BglII fragment was excised fromWP36 and cloned into the XbaI–BamHI sites of pGreen 1K to yield abinary vector encoding wild-type I-2 (WP42). pGreen 1K contains aCaMV 35S promoter and a NOS transcription terminator (a kind giftof K. Spelt, Free University of Amsterdam, the Netherlands). Tointroduce the D495V mutation in I-2, two PCRs were performed withPfu polymerase (Stratagene) using primer set 1, FP 515(GTCTACCGAAGCCTCTTGG) and FP516 (CATTGACAAGAACA-TGCATAAG), and primer set 2, FP517 (CTTATGCATGTTCTTGT-CAATG) and FP518 (CTGCTCCAATTTGTAGAGGG). The obtainedfragments were pooled and reamplified with FP515 and FP518. Theresulting PCR product was cut with BamHI and NdeI and used toreplace the corresponding fragment in WP36. After sequence veri-fication, the I-2 D495V construct was cut out with SpeI and BglII andcloned into the XbaI and BamHI sites of pGreen 1K to yield a binaryvector encoding I-2 D495V (WP45).

To express I-2 and I-2 D495V, pGreen 1K constructs weretransformed to A. tumefaciens GV3101 and infiltrated into N. bent-hamiana leaves as previously described (van der Hoorn et al., 2000).For silencing experiments, the I-2 D495V construct was agroinfil-trated into the upper three leafs 3 weeks after initiation of silencing.Seven days after infiltration, leafs were photographed. Experimentswere repeated at least three times with at least four independentplants.

Acknowledgements

We want to thank Martijn Rep and Dave Speijer for analysis of massspectrometry data, Thijs Hendrix for preparing anti-HSP90 anti-bodies, Sandra Elzinga for the Mi-1.2 LRR clone and Julian Verdonkfor his help during the isolation of HSP17. Ludek Tikovsky andHarold Lemereis are acknowledged for taking care of the plants. Weare grateful to Jack Peart, Peter Moffett and David Baulcombe forthe Rx clone, TRV silencing vectors and their expertise on VIGS.RPM1 and RPS5 clones were kindly provided by Douglas Boyes andRoger Innes respectively. J.H.V. was supported by Keygene NV,Wageningen, the Netherlands and K.J. de V. by The Centre forBioSystems Genomics, the Netherlands.

Supplementary Material

The following supplementary material is available for this articleonline:Figure S1. Updated annotation of the LRR domain of I-2.(Jones, D.A. and Jones, J.D.G. (1997) The role of leucine-rich repeatproteins in plant defences. Adv. Bot. Res. 24, 90–167).Figure S2. PP5 is not essential for I-2 nor RPM1-mediated resist-ance.Figure S3. Silencing of HSP90 in N. benthamiana affects plantdevelopment.

References

Austin, M.J., Muskett, P., Kahn, K., Feys, B.J., Jones, J.D.G. and

Parker, J.E. (2002) Regulatory role of SGT1 in early R gene-mediated plant defenses. Science, 295, 2077–2080.

Axtell, M.J. and Staskawicz, B.J. (2003) Initiation of RPS2-specifieddisease resistance in Arabidopsis is coupled to the AvrRpt2-directed elimination of RIN4. Cell, 112, 369–377.

Axtell, M.J., McNellis, T.W., Mudgett, M.B., Hsu, C.S. and Stas-

kawicz, B.J. (2001) Mutational analysis of the Arabidopsis RPS2disease resistance gene and the corresponding Pseudomonas

syringae avrRpt2 avirulence gene. Mol. Plant Microbe Interact. 14,181–188.

Azevedo, C., Sadanandom, A., Kitagawa, K., Freialdenhoven, A.,

Shirasu, K. and Schulze-Lefert, P. (2002) The RAR1 interactorSGT1, an essential component of R gene-triggered diseaseresistance. Science, 295, 2073–2077.

Bendahmane, A., Farnham, G., Moffett, P. and Baulcombe, D.C.

(2002) Constitutive gain-of-function mutants in a nucleotidebinding site-leucine rich repeat protein encoded at the Rx locus ofpotato. Plant J. 32, 195–204.

Bieri, S., Mauch, S., Shen, Q.-H. et al. (2004) RAR1 positively con-trols steady state levels of barley MLA resistance proteins andenables sufficient MLA6 accumulation for effective resistance.Plant Cell, 16, 3480–3495.

van der Biezen, E.A. and Jones, J.D.G. (1998) Plant disease-resist-ance proteins and the gene-for-gene concept. Trends Biochem.Sci. 23, 454–456.

Chen, M.X., McPartlin, A.E., Brown, L., Chen, Y.H., Barker, H.M. and

Cohen, P.T.W. (1994) A novel human protein serine/threoninephosphatase, which possesses four tetratricopeptide repeatmotifs and localizes to the nucleus. EMBO J. 13, 4278–4290.

Chen, M.-S., Silverstein, A.M., Pratt, W.B. and Chinkers, M. (1996)The tetratricopeptide repeat domain of protein phosphatase 5mediates binding to glucocorticoid receptor heterocomplexesand acts as a dominant negative mutant. J. Biol. Chem. 271,32315–32320.

Chinkers, M. (2001) Protein phosphatase 5 in signal transduction.Trends Endocrinol. Metab. 12, 28–32.

D’Andrea, L.D. and Regan, L. (2003) TPR proteins: the versatile helix.Trends Biochem. Sci. 28, 655–662.

Dangl, J.L. and Jones, J.D.G. (2001) Plant pathogens and integrateddefence responses to infection. Nature, 411, 826–833.

Dobson, S., Kar, B., Kumar, R., Adams, B. and Barik, S. (2001) Anovel tetratricopeptide repeat (TPR) containing PP5 serine/thre-onine protein phosphatase in the malaria parasite, Plasmodiumfalciparum. BMC Microbiol. 1, 31.

Dolinski, K., Muir, S., Cardenas, M. and Heitman, J. (1997) Allcyclophilins and FK506 binding proteins are, individuallyand collectively, dispensable for viability in Saccharomycescerevisiae. Proc. Natl Acad. Sci. USA, 94, 13093–13098.

Duttweiler, H.M. (1996) A highly sensitive and non-lethal beta-galactosidase plate assay for yeast. Trends Genet. 12, 340–341.

Ekengren, S.K., Liu, Y., Schiff, M., Dinesh-Kumar, S.P. and Martin,

G.B. (2003) Two MAPK cascades, NPR1, and TGA transcriptionfactors play a role in Pto-mediated disease resistance in tomato.Plant J. 36, 905–917.

Elledge, S.J., Mulligan, J.T., Ramer, S.W., Spottswood, M. and

Davis, R.W. (1991) Lambda YES: a multifunctional cDNA expres-sion vector for the isolation of genes by complementation ofyeast and Escherichia coli mutations. Proc. Natl Acad. Sci. USA,88, 1731–1735.

Fillatti, J.J., Kiser, J., Rose, R. and Comai, L. (1987) Efficient transferof a glyphosate tolerance gene into tomato using a binary Agro-bacterium tumefaciens vector. Biotechnology, 5, 726–730.

Flor, H.H. (1942) Inheritance of pathogenicity in Melampsora lini.Phytopathology, 32, 653–669.

de la Fuente van Bentem, S., Vossen, J.H., Vermeer, J.E.M., de

Vroomen, M.J., Gadella, T.W.J., Jr, Haring, M.A. and Cornelissen,

B.J.C. (2003) The subcellular localization of plant protein phos-phatase 5 isoforms is determined by alternative splicing. PlantPhysiol. 133, 702–712.

Gavin, A.C., Bosche, M., Krause, R. et al. (2002) Functional organ-ization of the yeast proteome by systematic analysis of proteincomplexes. Nature, 415, 141–147.

296 Sergio de la Fuente van Bentem et al.

ª Blackwell Publishing Ltd, The Plant Journal, (2005), 43, 284–298

Gietz, D., St Jean, A., Woods, R.A. and Schiestl, R.H. (1992) Im-proved method for high efficiency transformation of intact yeastcells. Nucleic Acids Res. 20, 1425.

Guan, K.L. and Dixon, J.E. (1991) Eukaryotic proteins expressed inEscherichia coli: an improved thrombin cleavage and purificationprocedure of fusion proteins with glutathione S-transferase. Anal.Biochem. 192, 262–267.

Holt, B.F., III, Boyes, D.C., Ellerstrom, M., Siefers, N., Wiig, A., Ka-

uffman, S., Grant, M.R. and Dangl, J.L. (2002) An evolutionarilyconserved mediator of plant disease resistance gene function isrequired for normal Arabidopsis development. Dev. Cell, 2, 807–817.

van der Hoorn, R.A.L., Laurent, F., Roth, R. and De Wit, P.J.G.M.

(2000) Agroinfiltration is a versatile tool that facilitates compar-ative analyses of Avr9/Cf-9-induced and Avr4/Cf-4-induced nec-rosis. Mol. Plant Microbe Interact. 13, 439–446.

van der Hoorn, R.A.L., De Wit, P.J.G.M. and Joosten, M.H.A.J. (2002)Balancing selection favors guarding resistance proteins. TrendsPlant Sci. 7, 67–71.

Hubert, D.A., Tornero, P., Belkhadir, Y., Krishna, P., Takahashi, A.,

Shirasu, K. and Dangl, J.L. (2003) Cytosolic HSP90 associates withand modulates the Arabidopsis RPM1 disease resistance protein.EMBO J. 22, 5679–5689.

Hwang, C.F. and Williamson, V.M. (2003) Leucine-rich repeat-mediated intramolecular interactions in nematode recognitionand cell death signaling by the tomato resistance protein Mi.Plant J. 34, 585–593.

Hwang, C.F., Bhakta, A.V., Truesdell, G.M., Pudlo, W.M. and Will-

iamson, V.M. (2000) Evidence for a role of the N terminus andleucine-rich repeat region of the Mi gene product in regulation oflocalized cell death. Plant Cell, 12, 1319–1330.

Ikeda, K., Ogawa, S., Tsukui, T., Horie-Inoue, K., Ouchi, Y., Kato, S.,

Muramatsu, M. and Inoue, S. (2004) Protein phosphatase 5 is anegative regulator of estrogen receptor-mediated transcription.Mol. Endocrinol. 18, 1131–1143.

James, P., Halladay, J. and Craig, E.A. (1996) Genomic libraries anda host strain designed for highly efficient two-hybrid selection inyeast. Genetics, 144, 1425–1436.

Kamphausen, T., Fanghanel, J., Neumann, D., Schulz, B. and

Rahfeld, J.-U. (2002) Characterization of Arabidopsis thalianaAtFKBP42 that is membrane-bound and interacts with Hsp90.Plant J. 32, 263–276.

Kobe, B. and Deisenhofer, J. (1994) The leucine-rich repeat: a ver-satile binding motif. Trends Biochem. Sci. 19, 415–421.

Koning, A.J., Rose, R. and Comai, L. (1992) Developmental expres-sion of tomato heat-shock cognate protein 80. Plant Physiol. 100,801–811.

Kroon, B.A.M. and Elgersma, D.M. (1993) Interaction between race 2of Fusarium oxysporum f. sp. lycopersici and near-isogenicresistant and susceptible lines of intact plants or callus of tomato.J. Phytopathol. 137, 1–9.

Krysan, P.J., Young, J.C., Tax, F. and Sussman, M.R. (1996) Identi-fication of transferred DNA insertions within Arabidopsis genesinvolved in signal transduction and ion transport. Proc. Natl Acad.Sci. USA, 93, 8145–8150.

Liu, Y., Schiff, M., Marathe, R. and Dinesh-Kumar, S.P. (2002)Tobacco Rar1, EDS1 and NPR1/NIM1 genes are required forN-mediated resistance to tobacco mosaic virus. Plant J. 30,415–429.

Liu, Y., Burch-Smith, T., Schiff, M., Feng, S. and Dinesh-Kumar, S.P.

(2004) Molecular chaperone Hsp90 associates with resistanceprotein N and its signaling proteins SGT1 and Rar1 to modulatean innate immune response in plants. J. Biol. Chem. 279, 2101–2108.

Lu, R., Malcuit, I., Moffett, P., Ruiz, M.T., Peart, J., Wu, A.J., Rathjen,

J.P., Bendahmane, A., Day, L. and Baulcombe, D.C. (2003) Highthroughput virus-induced gene silencing implicates heat shockprotein 90 in plant disease resistance. EMBO J. 22, 5690–5699.

Mackey, D., Holt, B.F., III, Wiig, A. and Dangl, J.L. (2002) RIN4interacts with Pseudomonas syringae type III effector moleculesand is required for RPM1-mediated resistance in Arabidopsis.Cell, 108, 743–754.

Mackey, D., Belkhadir, Y., Alonso, J.M., Ecker, J.R. and Dangl, J.L.

(2003) Arabidopsis RIN4 is a target of the type III virulence effectorAvrRpt2 and modulates RPS2-mediated resistance. Cell, 112, 379–389.

Martin, G.B., Bogdanove, A.J. and Sessa, G. (2003) Understandingthe functions of plant disease resistance proteins. Annu. Rev.Plant Biol. 54, 23–61.

Maurer-Stroh, S., Dickens, N.J., Hughes-Davies, L., Kouzarides, T.,

Eisenhaber, F. and Ponting, C.P. (2003) The Tudor domain ‘RoyalFamily’: Tudor, plant Agenet, Chromo, PWWP and MBT domains.Trends Biochem. Sci. 28, 69–74.

Meyers, B.C., Kozik, A., Griego, A., Kuang, H. and Michelmore, R.W.

(2003). Genome-wide analysis of NBS-LRR-encoding genes inArabidopsis. Plant Cell, 15, 809–834.

Ori, N., Eshed, Y., Paran, I., Presting, G., Aviv, D., Tanksley, S., Zamir,

D. and Fluhr, R. (1997) The I2C family from the wilt disease resist-ance locus I2belongs to the nucleotide binding, leucine-rich repeatsuperfamily of plant resistance genes. Plant Cell, 9, 521–532.

Peart, J.R., Lu, R., Sadanandom, A. et al. (2002) Ubiquitin ligase-associated protein SGT1 is required for host and nonhost dis-ease resistance in plants. Proc. Natl Acad. Sci. USA, 15, 10865–10869.

Pratt, W.B. and Toft, D.O. (2003) Regulation of signaling proteinfunction and trafficking by the hsp90/hsp70-based chaperonemachinery. Exp. Biol. Med. 228, 111–133.

Pratt, W.B., Krishna, P. and Olsen, L.J. (2001) Hsp90-binding im-munophilins in plants: the protein movers. Trends Plant Sci. 6,54–58.

Pratt, W.B., Galigniana, M.D., Harrell, J.M. and DeFranco, D.B.

(2004) Role of hsp90 and the hsp90-binding immunophilins insignalling protein movement. Cell. Signal. 16, 857–872.

Ratcliff, F., Martin-Hernandez, A.M. and Baulcombe, D.C. (2001)Tobacco rattle virus as a vector for analysis of gene function bysilencing. Plant J. 25, 237–245.

Reddy, R.K., Kurek, I., Silverstein, A.M., Chinkers, M., Breiman, A.

and Krishna, P. (1998) High-molecular-weight FK506-bindingproteins are components of heat-shock protein 90 heterocom-plexes in wheat germ lysate. Plant Physiol. 118, 1395–1401.

Rep, M., Dekker, H.L., Vossen, J.H., de Boer, A.D., Houterman, P.M.,

Speijer, D., Back, J.W., de Koster, C.G. and Cornelissen, B.J.C.

(2002) Mass spectrometric identification of isoforms of PR pro-teins in xylem sap of fungus-infected tomato. Plant Physiol. 130,904–917.

Rep, M., Dekker, H.L., Vossen, J.H., de Boer, A.D., Houterman, P.M.,

de Koster, C.G. and Cornelissen, B.J.C. (2003) A tomato xylem sapprotein represents a new family of small cysteine-rich proteinswith structural similarity to lipid transfer proteins. FEBS Lett. 534,82–86.

Russell, L.C., Whitt, S.R., Chen, M.-S. and Chinkers, M. (1999)Identification of conserved residues required for the binding of atetratricopeptide repeat domain to heat shock protein 90. J. Biol.Chem. 274, 20060–20063.

Schulze-Lefert, P. (2004) Plant immunity: the origami of receptoractivation. Curr. Biol. 14, R22–R24.

Shao, J., Hartson, S.D. and Matts, R.L. (2002) Evidence that proteinphosphatase 5 functions to negatively modulate the maturation

The I-2 complex contains HSP90 and PP5 297

ª Blackwell Publishing Ltd, The Plant Journal, (2005), 43, 284–298

of the Hsp90-dependent heme-regulated eIF2a kinase. Biochem-istry, 41, 6770–6779.

Shirasu, K. and Schulze-Lefert, P. (2003) Complex formation, pro-miscuity and multi-functionality: protein interactions in disease-resistance pathways. Trends Plant Sci. 8, 252–258.

Silverstein, A.M., Galigniana, M.D., Chen, M.-S., Owens-Grillo, J.K.,

Chinkers, M. and Pratt, W.B. (1997) Protein phosphatase 5 is amajor component of glucocorticoid receptorhsp90 complexeswith properties of an FK506-binding immunophilin. J. Biol. Chem.272, 16224–16230.

Simons, G., Groenendijk, J., Wijbrandi, J. et al. (1998) Dissection ofthe Fusarium I2 gene cluster in tomato reveals six homologs andone active gene copy. Plant Cell, 10, 1055–1068.

Takahashi, A., Casais, C., Ichimura, K. and Shirasu, K. (2003) HSP90interacts with RAR1 and SGT1 and is essential for RPS2-mediateddisease resistance in Arabidopsis. Proc. Natl Acad. Sci. USA, 100,11777–11782.

Tornero, P., Merritt, P., Sadanandom, A., Shirasu, K., Innes, R.W.

and Dangl, J.L. (2002) RAR1 and NDR1 contribute quantitativelyto disease resistance in Arabidopsis, and their relative contri-butions are dependent on the R gene assayed. Plant Cell, 14,1005–1015.

Warren, R.F., Henk, A., Mowery, P., Holub, E. and Innes, R.W. (1998)A mutation within the leucine-rich repeat domain of the Arabid-opsis disease resistance gene RPS5 partially suppresses multiplebacterial and downy mildew resistance genes. Plant Cell, 10,1439–1452.

Wellman, F.L. (1939) A technique for studying host resistance andpathogenicity in tomato Fusarium wilt. Phytopathology, 29, 945–956.

Zhang, Z., Quick, M.K., Kanelakis, K.C., Gijzen, M. and Krishna, P.

(2003) Characterization of a plant homolog of hop, a cochaperoneof hsp90. Plant Physiol. 131, 525–535.

Sequence data from this article have been deposited with the GenBank data libraries under accession numbers AY150040

(LeHSP17), AY569436 (NbPP5).

298 Sergio de la Fuente van Bentem et al.

ª Blackwell Publishing Ltd, The Plant Journal, (2005), 43, 284–298