Constitutive gain-of-function mutants in a nucleotide binding site

Constitutive gain-of-function mutants in a nucleotidebinding site–leucine rich repeat protein encoded at the Rxlocus of potato

Abdelhafid Bendahmaney Garry Farnham, Peter Moffett and David C. Baulcombe�

The Sainsbury Laboratory, John Innes Centre, Norwich Research Park, Colney, Norwich NR4 7UH, UK

Received 25 April 2002; revised 13 June 2002; accepted 14 June 2002.�For correspondence (fax þ44 1603 250024; e-mail [email protected]).yPresent address: INRA-URGV, 2 Rue Gaston Cremieux CP 5708, 91057 Evry Cedex, France.

Summary

Rx in potato encodes a protein with a nucleotide binding site (NBS) and leucine-rich repeats (LRR) that

confers resistance against Potato virus X. The NBS and LRR domains in Rx are present in many disease

resistance proteins in plants and in regulators of apoptosis in animals. To investigate structure-function

relationships of NBS-LRR proteins we exploited the potential of Rx to mediate a cell death response. With

wild-type Rx cell death is elicited only in the presence of the viral coat protein. However, following random

mutagenesis of Rx, we identified mutants in which cell death is activated in the absence of viral coat

protein. Out of 2500 Rx clones tested there were seven constitutive gain-of-function mutants carrying eight

independent mutations. The mutations encoded changes in the LRR or in conserved RNBS-D and MHD

motifs of the NBS. Based on these findings we propose that there are inhibitory domains in the NBS and

LRR. The constitutive gain-of-function phenotypes would be due to deletion or modification of these

inhibitory domains. However activation of Rx is not simply release of negative regulation by the LRR and

adjacent sequence because deleted forms of Rx that lack constitutive gain of function mutations are not

active unless the protein is overexpressed.

Keywords: potato virus X; coat protein; hypersensitive response; cell death.

Introduction

Many disease resistance (R) genes in plants encode pro-

teins with nucleotide binding site (NBS) and leucine rich

repeat (LRR) domains (Ellis et al., 2000; Hammond-Kosack

and Jones, 1997). Genetic evidence indicates that these

NBS-LRR proteins are receptors that interact, directly or

indirectly, with elicitors produced by the pathogen. Follow-

ing recognition there is activation of downstream signalling

pathways leading to disease resistance and a hypersensi-

tive response (HR) that is manifested as cell death (Shirasu

and Schulze-Lefert, 2000).

In common with other receptors it is generally considered

that NBS-LRR proteins have a modular structure with sepa-

rate recognition and signalling domains. The LRR is a

candidate recognition domain because it is the most vari-

able region in closely related NBS-LRR proteins and is

under selection to diverge (Meyers et al., 1998; Noel et al.,

1999). Functional analysis of recombinant R proteins also

indicates that recognition specificity resides in the LRR (Ellis

et al., 1999). However, there is indirect evidence that the

LRR can contribute to signalling as well as recognition.

(Banerjee et al., 2001; Warren et al., 1998)

The amino terminal region in NBS-LRR R proteins is

thought to be the major signalling domain. In most NBS-

LRR proteins this region includes either a Toll-interleukin

receptor (TIR)-like domain or a putative coiled-coil (CC)

structure (Ellis et al., 2000). In Arabidopsis, TIR-NBS-LRR

proteins apparently signal through the EDS1 lipase-like

protein, whereas CC-NBS-LRR proteins require pathways

involving NDR1 (Aarts et al., 1998). RPS7 and RPP8 are NBS-

LRR proteins but do not have absolute dependency on

EDS1 or NDR1 and it is thought that they signal through

a novel, as yet uncharacterized, pathway (McDowell et al.,

2000). In some NBS-LRR proteins the amino terminal

domain may be involved in recognition as well as signal-

ling. For example, from the novel resistance specificity of

recombinant L6, L2 and LH flax rust R genes, it seems likely

The Plant Journal (2002) 32, 195–204

� 2002 Blackwell Publishing Ltd 195

that recognition can be influenced by a TIR domain (Ellis

et al., 1999; Luck et al., 2000).

The NBS is positioned between the putative signalling

domain and the LRR. Its precise function is not known but

there are clues from sequence motifs that are also present

in the NBS of CED-4 (C. elegans) and APAF-1 (human)

regulators of apoptosis (Aravind et al., 1999; van der Biezen

and Jones, 1998). The common sequences include the P

loop (kinase 1), kinase 2 and kinase 3a and other motifs

found in ATPase proteins from both prokaryotes and eukar-

yotes (Aravind et al., 1999). Mutation analysis indicates that

at least some of these motifs are required for the function of

R proteins (Axtell et al., 2001; Dinesh-Kumar et al., 2000; Tao

et al., 2000; Tornero et al., 2002). By analogy to APAF-1 and

similar proteins, it seems likely that ATPase activity in NBS-

LRR proteins mediates conformational changes necessary

for downstream signalling and disease resistance (Hu et al.,

1999).

Indirect evidence from the nematode resistance gene Mi

indicates how the activity of NBS-LRR proteins could be

controlled (Hwang et al., 2000). Recombinant Mi proteins

containing elements of a non-functional Mi homologue

activate an HR in the absence of elicitor. To explain this

result it was proposed that there are intra-molecular inter-

actions between domains of Mi. In the wild-type protein in

the absence of elicitor these interactions would prevent

signalling of disease resistance and cell death. Modifica-

tion of the interaction, either following elicitor recognition

or by mismatch of interacting residues in the recombinant

proteins, would allow the signalling domains of Mi to

activate the response pathways. The lesion mimic pheno-

type of a natural recombinant of the Rp1 resistance gene in

maize (Sun et al., 2001) is also consistent with this inter-

pretation.

Here we investigate structure-function relationships in

the CC-NBS-LRR protein encoded by Rx. Rx confers

resistance against Potato virus X (PVX) in potato (Cocker-

ham, 1970) and the elicitor is the viral coat protein (CP)

(Bendahmane et al., 1995). Rx-mediated resistance is unu-

sual in that it is not normally associated with an HR either in

potato or as a transgene in Nicotiana species (Bendahmane

et al., 1999). However transgenic expression of the CP elicits

an Rx-dependent HR in potato and in Nicotiana species

(Bendahmane et al., 1999). Our approach to the analysis of

Rx involved screening for Rx mutants that mediate an HR in

the absence of the elicitor CP. We also tested the over-

expression of wild type and deleted forms of Rx for an

elicitor-independent HR. Our results imply that conserved

motifs in the NBS and LRR regions of Rx regulate a signal-

ling domain in the amino terminal region. As deletion or

mutation of these motifs leads to gain of Rx function it is

likely that they have an inhibitory role in the wild-type

protein. Presumably this inhibitory function is abrogated

in the presence of the coat protein elicitor.

Results

Random mutagenesis of Rx

To address the possible role of the NBS and other domains

in Rx we screened for Rx mutants in which an HR was

activated in the absence of the elicitor PVX CP. The mutants

were generated by PCR of Rx cDNA under conditions

leading to misincorporation of nucleotides. The PCR

products were then inserted between the promoter and

transcriptional terminator of Rx (Figure 1a, pR-AT) and

the constructs were assembled in the T-DNA region of an

Agrobacterium binary plasmid vector. A library of Rx

mutants was generated and individual clones were assayed

for HR induction by infiltration of liquid Agrobacterium

cultures into leaves of Nicotiana tabacum. All experiments

were repeated at least three times with several replicates in

each experiment and the data, as presented here, were

consistent and reproducible in all instances.

Agrobacterium cultures carrying a wild-type Rx cDNA

construct (pR-Rx; Figure 1a) did not induce an HR when

infiltrated into non-transformed N. tabacum leaves. How-

ever, when this construct was infiltrated into transgenic N.

tabacum producing the CP of PVX (Spillane et al., 1997),

there was necrosis after 48 h and death of the infiltrated

region by 72 h (Figure 1b). Presumably the transiently

expressed Rx had recognized the elicitor CP and activated

signalling leading to cell death.

Out of 2500 mutant Rx clones tested there were seven that

induced an HR in the leaves of non-transformed N. tabacum.

Figure 1(b) illustrates the HR phenotype of one of these con-

stitutive gain-of-function mutants (AT25). Four of the

mutants (AT39, AT193, AT25, and AT7) induced an HR within

48 h of Agrobacterium infiltration. The others (AT32, AT72

and AT28) induced an HR that was delayed until 48–72 h post

infiltration.Thesegain-of-functionmutantsallshowedanHR

in the presence of the coat protein (Figure 1b).

Constitutive gain-of-function mutants are modified in the

NBS and LRR domains

DNA sequence analysis revealed that each of the constitu-

tive gain-of-function forms of Rx carries between three and

11 amino acid substitutions relative to the wild type. To

identify the amino acid substitutions implicated in the

activation of Rx we constructed a series of recombinant

molecules incorporating elements of the wild type and

constitutive gain-of-function mutants. For each of the

gain-of-function mutants we identified recombinant mole-

cules with a single amino acid mutation that induced an

elicitor-independent HR on non-transformed N. tabacum

using the Agrobacterium infiltration assay.

In each of the pR-AT39, pR-AT193, pR-AT25, pR-AT32,

pR-AT72 and pR-AT28 clones a single mutation was

196 Abdelhafid Bendahmane et al.

� Blackwell Publishing Ltd, The Plant Journal, (2002), 32, 195–204

responsible for the HR (Figure 2a). In each instance the

timing of the HR was the same with the single amino acid

mutant and the corresponding progenitor clone. The pR-

AT7 clone carried two mutations that were independently

responsible for activation of an HR (Figure 2a). The pR-AT7-

derived recombinants with single amino acid mutations

caused an HR that was slower than with the progenitor

pR-AT7. Presumably the faster HR of the progenitor pR-AT7

was due to an additive effect of the two mutations.

Three out of these eight mutations (F393I, D399V, E400K)

are in an eight amino acid interval close to the RNBS-D

(CFLY) motif that is conserved in R proteins (Hammond-

Kosack and Jones, 1997; Meyers et al., 1999). Two other

mutations led to the same amino acid change (D460V) in the

conserved MHD motif (Hammond-Kosack and Jones, 1997)

(Figure 2a,b) but are derived independently because, in the

Figure 1. Induction of HR by a constitutive gain-of-function mutant of Rx.(a) Schematic representation of T-DNA constructs for expression of RxcDNA. The cDNA inserts of wild-type or mutant forms of Rx were insertedbetween Rx promoter (pR) and transcriptional terminator (ter). The blackbox indicates the cDNA or either wild-type (wt) Rx cDNA or the cDNA ofmutant (AT) forms of Rx. LB and RB indicate the left and right border of the T-DNA.(b) Expression of wild type Rx and the constitutive gain-of-function mutantpR-AT25 in N. tabacum. The Rx constructs were the wild type Rx (pR-Rx)or the constitutive gain-of-function mutant (pR-AT25). Agrobacterium cul-tures carrying pR-Rx or pR-AT25 was infiltrated into leaves of either-non-transformed (NT) or transgenic N. tabacum (CP) expressing the PVXCP from the 35S promoter. The leaves were photographed 4 days afterinfiltration.

Figure 2. The sequence of the Rx constitutive gain-of-function mutants.(a) Sequence changes responsible for constitutive gain-of-function pheno-types. The diagram shows the changes at the nucleotide and protein levelthat are responsible for the constitutive gain-of-function phenotype.(b) The predicted protein sequence of Rx. The sequence is divided into threedomains corresponding to the amino terminal CC domain, the NBS domainand the LRR. The NBS domain includes the P loop, kinase2 and kinase 3amotifs of the NBS (Aravind et al., 1999) and the RNBS-C, GLPL, RNBS-D andMHD motifs that are conserved in NBS-LRR proteins (Meyers et al., 1999; vander Biezen and Jones, 1998). The residues modified in the constitutive gain-of-function mutants are shown with a black background and the mutantresidues are in bold above the main sequence. The LRR and C terminusregion includes the LRR and the amide rich and acidic tail regions at thecarboxy terminus of Rx (Bendahmane et al., 1999). The LRR consensusmotifs are shown in blue and are aligned. The conserved VLDL motif in LRR3is underlined.


Gain of Rx function by mutation and overexpression 197

pR-AT39 and pR-AT193 progenitors, they were associated

with different mutations that did not affect the Rx response

(AB, data not shown). The remaining three constitutive

gain-of-function mutations were in LRR2 (H519R), LRR3

(D543E) and LRR11 (H738R) (Figure 2a,b). The LRR3 muta-

tion is in a motif, VLDL, that is present in LRR3 of many NBS-

LRR R proteins (Axtell et al., 2001; Banerjee et al., 2001) (G.

Farnham, unpublished data).

The Rx gain-of-function phenotype is dependent on a

wild-type P loop motif and on SGT1

In principle the constitutive gain-of-function mutants could

elicit an HR because the encoded proteins are toxic in the

Agrobacterium-infiltrated leaves.Alternatively thesemutant

proteins could have activated the downstream pathway

that is involved in the Rx-mediated response to the PVX

CP. To test these possibilities we introduced alanine sub-

stitutions (G175aþK176A) in the P loop (K1) motif of the

NBS (Aravind et al., 1999) (Figure 2) and assayed the HR

phenotype with the Agrobacterium infiltration assay in non-

transgenic and CP transgenic plants. We predicted that an

HR related to Rx-mediated resistance would depend on in-

tegrity of the conserved P loop, whereas an HR due to toxic

effectsof protein over-expression in the Agrobacterium infil-

tration assay would be unaffected by disruption of the NBS.

As predicted the P loop Rx mutants failed to show an HR

in the infiltrated region of the CP transgenic plants (Figure

3a) indicating that activation of downstream signalling

requires integrity of the NBS. Similarly, when the P loop

mutation was introduced into Rx gain-of-function mutants,

there was no HR in non-transformed N. tabacum leaves.

Identical results were obtained with HA epitope tagged

versions of Rx and the constitutive gain-of-function

mutants. Thus, Rx(HA) was fully functional in an Agrobac-

terium infiltration HR assay in the CP transgenic N. tabacum

and was readily detected by Western blot analysis with HA

antibody in non-transgenic plants (Figure 3b). The P loop

mutants of Rx(HA), D543E(HA) and D460V(HA) were also

non-functional in the HR assay although the encoded pro-

teins were detected by Western analysis (Figure 3b). From

these results we can rule out that the lack of an HR with the P

loop mutants was due to protein destabilization. A more

likely explanation is that the P loop has the same function in

wild type and gain-of-function mutants of Rx. This inter-

pretation is reinforced by our finding (AB, data not shown)

that mutations in residues D244AþD245A (kinase 2) or

G330AþP332A (GLPL) abrogated the ability of both wild-

type and constitutive gain-of-function mutants to produce

an HR in the infiltration assay.

Further indication that the gain-of-function phenotype is

relevant to Rx mediated resistance was based on an HR

assay in N. benthamiana following virus induced gene

silencing (VIGS) of SGT1 loci. SGT1 proteins are required

for many types of disease resistance responses (Austin

et al., 2002; Azevedo et al., 2002) including the HR asso-

ciated with Rx (Figure 3c)(Peart et al., 2002). However

expression of SGT1 loci is not required for cell death

induced by ethanol, sodium azide or sodium chloride (Peart

et al., 2002). Based on these findings we predicted that the

constitutive gain-of-function HR would only be compro-

mised by silencing of SGT1 if it were a bona fide resistance

response. A toxic response would not be compromised by

SGT1 silencing. The HR tests in SGT1 silenced plants were

carried out with D460V and H519RþD543EþH738R in

which three LRR gain-of-function mutations were com-

bined in one construct. The SGT1 silencing was induced

by infection of N. benthamiana with a tobacco rattle virus

vector (Ratcliff et al., 2001) construct carrying an SGT1

cDNA insert.

The results shown in Figure 3(c) show that SGT1 expres-

sion is required for an HR induced by the mutant Rx

proteins. This requirement is not because SGT1 silencing

causes reduced levels of Rx (P. Moffett, unpublished data).

It is likely therefore that the constitutive gain-of-function

phenotype represents Rx-mediated resistance rather than a

toxic response. The SGT1 requirement for other gain-of-

function mutants could not be tested because, for reasons

that we do not understand, they did not produce an elicitor-

independent HR in N. benthamiana unless they accumulate

at very high levels due to co-expression in the presence of a

suppressor of gene silencing (A. Bendahmane, G. Farnham

and P. Moffett unpublished observations). In our hands

virus-induced silencing is much more effective in N.

benthamiana than in N. tabacum (Ana Montserrat Martin

Hernandez, unpublished observations).

Overexpression of Rx

As a second approach to structure-function analysis of Rx

we used the Agrobacterium infiltration assay to over-

express full length and deleted Rx constructs in the absence

of elicitor (Figure 4a). We predicted that overexpression of

the full length Rx in N. tabacum would activate resistance in

the absence of elicitor, as with other NBS-LRR proteins

(Oldroyd and Staskawicz, 1998; Tao et al., 2000). We also

predicted that deletion of domains required for down-

stream signalling would prevent the overexpression phe-

notype. However, deletions in inhibitory domains would

either have no effect or would enhance the overexpression

phenotype. The overexpression constructs all included a

carboxy terminal HA tag so that the level of the Rx proteins

could be monitored by immuno-blotting with HA antibody.

As predicted, expression of full length Rx(HA) under the

strong constitutive 35S RNA promoter of CaMV (Figure 4a)

induced an HR that developed by 72 h post infiltration in the

absence of PVX coat protein. The same protein, either with

or without the HA tag did not induce an HR when expressed



from the genomic Rx promoter (Figure 1b). The level of

Rx(HA) was higher from the 35S construct than with the Rx

promoter (Figure 4b) and we conclude that the HR is caused

by overexpression of Rx.

Overexpression of three deleted versions of Rx(HA)

(382D, 318D and 293D) resulted in an HR that developed

more rapidly (< 48 h) than with the full length protein (72 h).

The 282D protein induced a very slow HR (> 96 h) and the

1D138 amino terminal deletant was inactive. These deleted

proteins were HA tagged but we could not detect the 382D,

318D and 293D proteins in N. tabacum because rapid HR

occurred before they had accumulated to detectable levels.

Figure 3. Constitutive gain-of-function phenotypes require P loop-dependent signalling and SGT1.(a) Rx constructs were infiltrated into N. tabacum that was either non-transgenic or that expressed the PVX CP. Production of an HR was monitored over a periodof 4 days. The Rx constructs were either in a wild-type background or with constitutive gain-of-function mutations E440K, D460V or H519RþD543EþH738R. TheP loop motif in these constructs was either wild type or had GK residues at position 175 and 176 modified to AA (K1). HR indicates an HR within 48 h of infiltrationand ‘-’indicates no HR.(b) Western blot analysis of HA tagged versions of Rx and constitutive gain-of-function mutants with P loop (K1) mutations. The constructs were either the wild-type Rx with an HA tag (RxHA), K1 mutant Rx with an HA tag (RxHAK1) and RxHA constructs with double constitutive gain-of-function and K1 mutations((D543E)K1 and (D460V)K1). The constructs were transiently expressed using Agrobacterium infiltration (O.D.600¼0.2). Samples were taken at 3 days postinfiltration, fractionated by SDS-PAGE and the HA tagged proteins were detected with anti-HA antibody. The HR or ‘-’symbols indicate whether the proteins wereable to mediate an HR within 48 h either in the presence or absence of CP.(c) N. benthamiana plants were infected with TRV:00 or TRV:SGT to silence SGT1 expression. After 21d Agrobacterium infiltration was used to introduce Rxconstructs into the upper infected leaves of the plants either individually or in combination with constructs carrying a 35S CP construct. The PVX constructs wereeither wild type Rx in the presence (1) or absence (2) of CP, D460V (3) or H519RþD543EþH738R (4). The images show that Rx mediated an HR in TRV:00-infectedplants either as a wild-type Rx in the presence of the CP construct or as a constitutive gain-of-function mutant (right leaf) but that SGT1 silencing compromisedthe resistance-associated HR (left leaf).



However in N. benthamiana none of these proteins induced

an HR and they could be detected at 48 h post infiltration by

western analysis (Figure 4c). All of the deleted forms of Rx

were substantially less abundant than full length Rx.

Based on the assumption that accumulation in N.

benthamiana is indicative of levels in N. tabacum these

results show that HR-inducing activity is enhanced if the

carboxy terminal part of the protein is removed. Thus 382D,

318D and 293D elicited a rapid HR in N. tabacum, despite

their accumulation at a lower level than the full length

protein. This interpretation is consistent with the proposal,

based on the gain-of-function mutant phenotype described

above, that there are negative regulatory functions in Rx

between residues 382 and the carboxy terminus.

The removal of sequence on the amino terminal side of

residue 293 either reduced (282D) or abolished (1D138) HR-

inducing activity, even though these proteins were at least

as abundant as those with a rapid HR phenotype (Figure 4c).

The 282D construct differs from the 293D construct that

elicited a more rapid HR in that it lacks a complete kinase 3

motif and the completely inactive 1D138 is deleted in the CC

domain. It is likely therefore that the CC domain and the

kinase 3 motif of the NBS are essential for downstream

signaling capability of Rx.

Figure 4. Over expression of Rx deletion mu-tants.(a) A scale diagram of Rx showing the CC andLRR regions, as in Figure 2. In the NBS domainthe locations are shown of the P loop(K1), kinase2 kinase 3a and other motifs that are conservedin NBS-LRR proteins including APAF-1 andCED4. The lines indicate the extent of deletionsin HA tagged constructs relative to these con-served motifs. The figures in the right hand sidecolumn indicate the timing of an HR with theseconstructs when transiently expressed from a35S promoter in the absence of CP (NR¼noresponse).(b) Western analysis of RxHA expressed in N.tabacum in an Agrobacterium transient assay(O.D.600 ¼0.2) either from the Rx promoter(pB1)or from the 35S promoter (35S). The Rxconstructs encoded either the wild-type proteinof the P loop (K1) mutants. Protein extracts wereprepared 40 h post-infiltration, prior to the onsetof HR, separated by SDS-PAGE and detected byanti-HA immuno-blotting. An asterisk denotesthat in these samples, five-fold less proteinextract was loaded.(c) Western analysis of RxHA and deletion con-structs expressed in an Agrobacterium transientassay (O.D. 600¼ 0.5)from the 35S promoter ofCaMV in N. benthamiana. Protein extracts fromduplicate experiments were prepared 48 h post-infiltration, separated by SDS-PAGE and de-tected by anti-HA immuno-blotting.



Discussion

Regulatory domains in Rx and other NBS-LRR proteins

In this paper we report two findings related to the function

of Rx and, presumably, other NBS-LRR proteins. First we

describe constitutive gain-of-function mutations in the cen-

tral and carboxy terminal regions of the Rx protein and,

second, we show that an HR induced by overexpression of

Rx can be accelerated by deletion of conserved sequence

motifs in the NBS. We conclude that these Rx phenotypes

are related to signalling rather than recognition because

they involve an elicitor-independent HR.

In one scenario a constitutive gain-of-function mutation

would have fortuitously generated specific structures of Rx

that normally exist only in the presence of elicitor. Alter-

natively, the gain-of-function mutations could have an

indirect effect by, for example, disruption of inhibitory

domains in Rx that prevent signalling in the absence of

elicitor. Formally, from the data presented here, neither

interpretation can be ruled out. However it is unlikely that

random mutagenesis would repeatedly generate new spe-

cific structures involved in downstream signalling. A more

likely scenario is that the mutation has a disruptive effect

and that the constitutive gain-of-function mutations affect

inhibitory domains in Rx. A similar interpretation, invoking

inactivation of inhibitory domains, was used to explain a

constitutive gain-of-function phenotype in a mutant Pto

protein (Rathjen et al., 1999).

Other NBS-LRR proteins that, like Rx, may be subject to

negative regulation include Nod1 in humans. This protein

exhibits enhanced signalling activity when over-expressed

with a deletion in the LRR (Inohara et al., 1999) and it was

concluded that the LRR prevents signalling by amino term-

inal domains. The analysis of the nematode resistance

protein Mi is also consistent with negative regulatory

domains in NBS-LRR proteins. Recombinants of Mi and

an inactive homologue induced an elicitor-independent HR

(Hwang et al., 2000) like the constitutive gain of function Rx

mutants described here. To explain this phenotype, it was

proposed that there are intramolecular interactions in Mi

(Hwang et al., 2000). In the recombinants or, presumably in

the presence of the Mi elicitor, these inhibitory interactions

would be disrupted so that signalling domains would

become exposed and available for interaction with compo-

nents of the downstream pathway.

Conserved sequence motifs in NBS-LRR proteins

Most of the constitutive gain-of-function mutations in Rx

are in or close to sequence motifs that are common to many

NBS-LRR proteins. For example the D543E (Figure 2) muta-

tion is in the VLDL motif that is present in the third LRR of

many NBS-LRR proteins (Axtell et al., 2001) (G. Farnham,

unpublished data) including RPS5 in Arabidopsis. RPS5

encodes an NBS-LRR protein and confers resistance

against Pseudomonas syringae. However a VLDL motif

mutant allele, rps5-1, has a complex phenotype involving

loss of RPS5-mediated resistance and suppression of resis-

tance conferred by other NBS-LRR proteins (Warren et al.,

1998).

To reconcile the interference phenotype of rps5-1 and the

constitutive gain-of-function of D543E we propose that

these mutations both disrupt a negative regulatory domain

in the respective NBS-LRR proteins. Complete inactivation

of this domain, as in D543E, would release the negative

regulation so that the Rx protein could interact with down-

stream components of the signalling pathway and activate

an HR in the absence of elicitor. Partial inactivation of this

domain could allow the NBS-LRR protein to interact with

downstream signalling components but, as may be the

case with rps5-1, without activation of signalling. In that

scenario, if the downstream signalling components are

required by other NBS-LRR proteins, a partial loss-of-func-

tion would interfere with the respective signalling path-

ways.

Other NBS-LRR motifs, including RNBS-D (Meyers et al.,

1999) and MHD (van der Biezen and Jones, 1998) are

implicated in Rx inhibitory domains by constitutive gain-

of-function mutant phenotypes of D399V, E400K and D460V

(Figure 2). The MHD motif had not been targeted previously

in mutational analysis of NBS-LRR proteins (Axtell et al.,

2001; Dinesh-Kumar et al., 2000; Tao et al., 2000; Tornero

et al., 2002) and so the data presented here provide the first

direct analysis of its function.

Other than Rx, the most extensively mutagenised NBS-

LRR protein is RPM1 (Tornero et al., 2002). A total of 53

missense loss-of-function alleles were identified at this

locus of which only one is at or adjacent to the conserved

motifs that were modified in the Rx gain-of-function alleles.

The one exception (rpm1-12) affected the RNBS-D motif

that was modified in the AT72 and AT25 alleles of Rx (Figure

2). However the mutated amino acid in rpm1-12 is not

conserved in Rx and therefore is not directly comparable

with our constitutive gain-of-function alleles. The other

RPM1 mutants, like the gain-of-function mutants of Rx,

were predominantly either within or on the carboxy term-

inal side of the NBS. However, it is not possible to draw

additional conclusions by cross referencing between the

two sets of data because the gain- or loss-of-function

screens would identify different functional domains.

Amino terminal domains

Our constitutive gain-of-function mutants are not directly

informative about signalling domains in Rx. However, since

the HR would require signalling, the phenotypes of the Rx

deletion mutants (1D138, 382D, 318D and 293D; Figure 4) are



indirectly informative. They indicate, consistent with pre-

vious suggestions (Aarts et al., 1998), that signalling

involves the amino terminal domain in combination with

the P loop, kinase 2 and kinase 3a motifs. It is striking that

none of the constitutive gain-of-function mutations was

within this region of the Rx protein. Perhaps, as suggested

above, random mutagenesis leads primarily to disruption

of functional domains rather than the generation of specific

structures that are required for interaction with compo-

nents of the signalling pathway.

A gain of function analysis of the nematode resistance

gene Mi led to the conclusion that the amino terminal

domain was a regulator of signalling mediated by the

LRR. This interpretation differs from that made here about

Rx and it may be that activation of Mi- and Rx-mediated

resistance involves different mechanisms. However, if as

seems likely, the NBS-LRR domains have multiple functions

in recognition, activation and response signalling these two

sets of data are not necessarily incompatible. The artificial

Mi recombinants would have been informative about some

but not all of these functions. Similarly the Rx point mutants

would have provided information about a different set of

functions. For more complete understanding of recognition

and response in disease resistance, it will be necessary to

generate high resolution secondary and tertiary structures

of Rx and other NBS-LRR proteins and their elicitors.

Experimental procedures

Rx constructs

pB1 is a modified pBIN19 plasmid (Bevan, 1984) that carries atranscription cassette comprising 3 kb of the Rx promoter and a1.5-kb Rx terminator separated by an XbaI and a SacI cloning sites(Bendahmane et al., 2000). All Rx derivative mutants were clonedbetween the XbaI and the SacI cloning sites. All constructs weresequenced to confirm the absence of PCR-derived mutations.To construct pR-Rx, Rx cDNA was PCR amplified and the PCRproduct was cloned into the XbaI and SacI sites of pB1 to createpR-Rx.

The 35S promoter constructs were all in the pBIN61 binaryvector. pBIN61 is a modified pBIN19 binary vector that carries atranscription cassette comprising the CaMV 35S promoter andterminator. To construct the pBIN61 binary vector, the transcrip-tion cassette containing the CaMV 35S promoter and terminatorwas released by digestion with KpnI and XhoI from the plasmidpJIT61 (kindly provided by P. Mullineaux, JIC, Norwich, UK). Thetranscription cassette was then ligated to the pBIN19 plasmidvector digested with KpnI and SalI to create pBIN61. To construct35S-Rx, Rx cDNA was PCR amplified and cloned into the XbaI andSmaI sites of pBIN61.

RxHA was constructed via PCR, using a primer that introduced aBamHI site at the extreme carboxyl terminus of the Rx openreading frame followed by the HA epitope tag. Similarly thetruncated constructs were generated by PCR from the Rx cDNAwith 30 primers introducing a BamHI site. The PCR products werecloned into the Xba1/BamH1 sites of PBin61-RxHA and subse-quently tranferred to pB1.

Full details of all constructs are available on request and bio-materials are available through our website (http://www.sains-bury-laboratory.ac.uk).

PCR mutagenesis

Random mutagenesis of the Rx gene was performed under con-ditions similar to those previously described (Shafikhani et al.,1997). The PCR was carried out using the primers RxP1 and Rxac4,which flank the Rx ORF. The PCR reaction contained (100 ml finalvolume) 10 mM Tris (pH 8.3), 50 mM KCl, 0.05% Nonidet P-40, 7 mM

MgCl2, 0.15 mM MnCl2, 0.2 mM dGTP, 0.2 mM dATP, 1 mM dCTP,1 mM dTTP, 0.3 mM of both primers, 50 ng of template and 5 U TaqDNApolymerase(Gibco-BRL).PCRwasperformedfor35cycles:15 sat 948C, 15 s at 558C, and 2 min at 728C. The PCR products weredigested with XbaI and SacI, gel purified and cloned in the binaryvector pB1 in E. coli. Plasmid DNA was purified from 10000 coloniesand electroporated into A. tumefaciens strain C58C1 carrying thevirulence helper plasmid pCH32 (Hamilton et al., 1996).

Site-directed mutagenesis

Oligonucleotide-directed mutagenesis was used to introduce spe-cific mutations into the Rx cDNA or into the mutants E400K, D460Vand D543EþH738R. The site directed mutations were confirmedby sequence analysis. The mutations of the P loop were at Rxcodon 175 from GGG(G) to GCG(A) and at codon 176 from AAA(K)to GCA(A). The mutations of the kinase Za motif were at Rx codon244 from GAT(D) to GCT(A) and at codon 245 from GAC(D) toGCC(A). The mutations of the GLPL motif were at codons 330 fromGGA(G) to GCA(A) and 332 from CCT(P) to GCT(A).

Agrobacterium-mediated transient expression

Agrobacterium-mediated transient expression was performedunder conditions similar to those described previously (Bendah-mane et al., 1999). The binary Ti-plasmid vector constructs weretransformed into A. tumefaciens strain C58C1 carrying the viru-lence helper plasmid pCH32 (Hamilton et al., 1996). The transfor-mants were inoculated into 5 ml L-broth medium supplementedwith 50 mg ml-1 kanamycin and 5 mg ml-1 tetracycline and grown at288C overnight. Cells were precipitated and resuspended to the ODof 0.5 in solution containing 10 mM MgCl2, 10 mM MES pH 5.6 and150 mM acetosyringone. The cells were left at room temperature onthe bench for 2 h before infiltration into N. tabacum leaves. Theinfiltrations were either into-non-transformed N. tabacum or intotransgenic N. tabacum expressing the PVX CP (Spillane et al.,1997). The transient expression was more efficient in the first trueleaf of young N. tabacum seedlings.

DNA sequencing and analysis

The sequencing reactions were performed using a dye terminatorcycle sequencing reaction kit (Perkin-Elmer, Cambridge, UK).Sequence reactions were resolved on ABI377 automated sequen-cer (Applied Biosystems ABI, La Jolla, CA, USA). Sequence contigswere assembled using UNIX versions of the Staden programspackage (Staden, 1996).

Immunodetection

To prepare protein samples, two leaf discs (approximately 35 mg)were ground in 150 ml of 8 M urea, followed by addition of 75 ml of



3X SDS loading buffer and boiling for 10 min Fifty ml of proteinextract was separated by 10% SDS-PAGE followed by transfer toImmun-blot PVDF membrane (Bio-Rad, Hemel Hempstead, UK).Membranes were incubated successively in 5% skim milk powderin TBST (25 mM Tris–HCl (pH 7.5), 150 mM NaCl, 0.1% Tween-20),0.1 ng ml�1 of 3F10 anti-HA antibody (Roche, Lewes, UK) andfinally with goat anti rat HRP conjugate (Sigma, Poole, UK). Pro-teins were visualised using the chemiluminesence kit ECL Plus(Pharmacia, Little Chalfont, UK).

Virus induced gene silencing (VIGS)

VIGS of the N. benthamiana homologue of SGT1 using a tobaccorattle virus vector (Ratcliff et al., 2001) has been described else-where (Peart et al., 2002). The procedures involved inoculation of14-day-old N. benthamiana plants with TRV vectors either withoutan insert (TRV:00) or carrying an N. benthamiana SGT1 insert(TRV:SGT). The ability of these infected plants to support an Rx-mediated HR was then tested after a further 14–21 days in theleaves that were systemically infected with TRV.

Acknowledgements

We thank Erik Van der Biezen for comments on the manuscript.The work was supported by the Gatsby Charitable Foundation andthe Biotechnology and Biological Sciences Research Council. P.Moffett is supported by an EMBO long-term fellowship. Recombi-nant viruses and transgenic plants were contained in greenhousesunder DEFRA license PHL 161/4080.

References

Aarts, N., Metz, M., Holub, E., Staskawicz, B.J., Daniels, M.J. andParker, J.E. (1998) Different requirements for EDS1 and NDR1 bydisease resistance genes define at least two R. gene-mediatedsignaling pathways in Arabidopsis. Proc. Natl Acad. Sci. USA,95, 10306–10311.

Aravind, L., Dixit, V.M. and Koonin, E.V. (1999) The domains ofdeath: evolution of the apoptosis machinery. Trends Biochem.Sci. 24, 47–53.

Austin, M.J., Muskett, P., Kahn, K., Feys, B.J., Jones, J.D.G. andParker, J.E. (2002) Regulatory role of SGT1. early R. gene-mediated plant defenses. Science, 295, 2077–2080.

Axtell, M.J., McNellis, T.W., Mudgett, M.B., Hsu, C.S. and Staska-wicz, B.J. (2001) Mutational analysis of the Arabidopsis RPS2disease resistance gene and the corresponding Pseudomonassyringae 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, 2073–2076.

Banerjee, D., Zhang, X.C. and Bent, A.F. (2001) The leucine-richrepeat domain can determine effective interaction betweenRPS2 and other host factors in Arabidopsis RPS2-mediateddisease resistance. Genetics, 158, 439–450.

Bendahmane, A., Kanyuka, K. and Baulcombe, D.C. (1999) The Rxgene from potato controls separate virus resistance and celldeath responses. Plant Cell, 11, 781–791.

Bendahmane, A., Kohm, B.A., Dedi, C. and Baulcombe, D.C. (1995)The coat protein of potato virus X is a strain-specific elicitor ofRx1-mediated virus resistance in potato. Plant J. 8, 933–941.

Bendahmane, A., Querci, M., Kanyuka, K. and Baulcombe, D.C.(2000) Agrobacterium transient expression system as a tool forisolation of disease resistance genes: application to the Rx2locus in potato. Plant J. 21, 73–81.

Bevan, M.W. (1984) Binary Agrobacterium vectors for plant trans-formation. Nucl Acids Res. 12, 8711–8721.

van der Biezen, E.A. and Jones, J.D.G. (1998) The NB-ARC domain:a novel signalling motif shared by plant resistance geneproducts and regulators of cell death in animals. Curr. Biol. 8,226–227.

Cockerham, G. (1970) Genetical studies on resistance to potatoviruses X and Y. Heredity, 25, 309–348.

Dinesh-Kumar, S.P., Tham, W.H. and Baker, B.J. (2000) Structure-function analysis of the tobacco mosaic virus resistance gene N.Proc. Natl Acad. Sci. USA, 97, 14789–14794.

Ellis, J., Dodds, P. and Pryor, T. (2000) Structure, function andevolution of plant disease resistance genes. Curr. Opin. Plant.Biol. 3, 278–284.

Ellis, J.G., Lawrence, G.J., Luck, J.E. and Dodds, P.N. (1999)Identification of regions in alleles of the flax rust resistance geneL that determine differences in gene-for-gene specificity. PlantCell, 11, 495–506.

Hamilton, C.M., Frary, A., Lewis, C. and Tanksley, S.D. (1996)Stable transfer of intact high molecular weight DNA into plantchromosomes. Proc. Natl Acad. Sci. USA, 93, 9975–9979.

Hammond-Kosack, K.E. and Jones, J.D.G. (1997) Plant diseaseresistance genes. Annu. Rev. Plant Physiol. Plant Mol. Biol.48, 575–607.

Hu, Y., Benedict, M., Ding, L. and Nunez, G. (1999) Role of cyto-chrome c and dATP/ATP hydrolysis in Apaf-1-mediated caspase-9 activation and apoptosis. EMBO J. 18, 3586–3595.

Hwang, C.-F., Bhakta, A.V., Truesdell, G.M., Waclawa, M.P. andWilliamson, V.M. (2000) Evidence for a role of the N terminusand leucine-rich repeat region of the Mi gene product in regula-tion of localized cell death. Plant Cell, 12, 1319–1329.

Inohara, N., Koseki, T., del Peso, L., Hu, Y.M., Yee, C., Chen, S.,Carrio, R., Merino, J., Liu, D., Ni, J. and Nunez, G. (1999) Nod1, anApaf-1-like activator of caspase-9 and nuclear factor- kappa B. J.Biol. Chem. 274, 14560–14567.

Luck, J.E., Lawrence, G.J., Dodds, P., Shepherd, K.W. and Ellis,J.G. (2000) Regions outside of the leucine-rich repeats of flaxrust resistance proteins play a role in specificity determination.Plant Cell, 12, 1367–1377.

McDowell, J.M., Cuzick, A., Can, C., Beynon, J., Dangl, J.L. andHolub, E.B. (2000) Downy mildew (Perenospora prasitica) resis-tance genes in Arabidopsis vary in functional requirements forNDR1, EDS1, NPR1 and salicylic acid accumulation. Plant J. 22,523–529.

Meyers, B.C., Dickerman, A.W., Michelmore, R.W., Sivaramakrish-nan, S., Sobral, B.W. and Young, N.D. (1999) Plant diseaseresistance genes encode members of an ancient and diverseprotein family within the nucleotide-binding superfamily. PlantJ. 20, 317–332.

Meyers, B.C., Shen, K.A., Rohani, P., Gaut, B.S. and Michelmore,R.W. (1998) Receptor-like genes in the major resistance locusof lettuce are subject to divergent selection. Plant Cell, 10,1833–1846.

Noel, L., Moores, T.L., van der Biezen, E.A., Parniske, M., Daniels,M.J., Parker, J.E. and Jones, J.D.G. (1999) Pronounced intras-pecific haplotype divergence at the RPP5 complex disease resis-tance locus of Arabidopsis. Plant Cell, 11, 2099–2111.

Oldroyd, G.E.D. and Staskawicz, B.J. (1998) Genetically engi-neered broad-spectrum disease resistance in tomato. Proc. NatlAcad. Sci. USA, 95, 10300–10305.



Peart, J.R., Lui, R., Sadanandom, A. et al. (2002) The ubiquitinligase-associated protein SGT1 is required for host and nonhostdisease resistance in plants. Proceedings of the Natl Acad. Sci.USA, 99, 10865–10869.

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.

Rathjen, J.P., Chang, J.H., Staskawicz, B.J. and Michelmore, R.W.(1999) Constitutively active Pto induces a Prf-dependent hyper-sensitive response in the absence of avrPto. EMBO J. 18, 3232–3240.

Shafikhani, S., Siegel, R.A., Ferrari, E. and Schellenberger, V.(1997) Generation of large libraries of random mutants in Bacil-lus subtilis by PCR-based plasmid multimerization. Biotechni-ques, 23, 304–310.

Shirasu, K. and Schulze-Lefert, P. (2000) Regulators of cell death indisease resistance. Plant Mol. Biol. 44, 371–385.

Spillane, C., Verchot, J., Kavanagh, T.A. and Baulcombe, D.C.(1997) Concurrent suppression of virus replication and rescue

of movement-defective virus in transgenic plants expressing thecoat protein of potato virus X. Virology, 236, 76–84.

Staden, R. (1996) The Staden sequence analysis package. MolBiotechnol. 5, 233–241.

Sun, Q., Collins, N.C., Ayliffe, M., Smith, S.M., Drake, J., Pryor, T.and Hulbert, S.H. (2001) Recombination between paralogues atthe rp1 rust resistance locus in maize. Genetics, 158, 423–438.

Tao, Y., Yuan, F.H., Leister, R.T., Ausubel, F.M. and Katagiri, F.(2000) Mutational analysis of the Arabidopsis nucleotide bindingsite- leucine-rich repeat resistance gene RPS2. Plant Cell, 12,2541–2554.

Tornero, P., Chao, R.A., Luthin, W.N., Goff, S.A. and Dangl, J.(2002) Large-scale structure-function analysis of the ArabidopsisRPM1 disease resistance protein. Plant Cell, 14, 435–450.

Warren, R.F., Henk, A., Mowery, P., Holub, E. and Innes, R.W.(1998) A mutation within the leucine-rich repeat domain ofthe arabidopsis disease resistance gene RPS5 partially sup-presses multiple bacterial and downy mildew resistance genes.Plant Cell, 10, 1439–1452.



Documents

Constitutive gain-of-function mutants in a nucleotide binding site