Cellular Microbiology (2005)

7

(3), 373–382 doi:10.1111/j.1462-5822.2004.00467.x

© 2004 Blackwell Publishing Ltd

Blackwell Science, LtdOxford, UKCMICellular Microbiology 1462-5814Blackwell Publishing Ltd, 20047

3373382

Original Article

N. Waterfield et al.Potentiation of the Toxin complexes

Received 9 July, 2004; revised 3 August, 2004; accepted 3 Septem-ber, 2004. *For correspondence. E-mail bssrfc@bath.ac.uk; Tel.(

+

44) 1225 386261; Fax (

+

44) 1225 386779.

Potentiation and cellular phenotypes of the insecticidal Toxin complexes of

Photorhabdus

bacteria

N. Waterfield, M. Hares, G. Yang, A. Dowling and R. ffrench-Constant*

Center for Molecular Microbiology and Department of Biology, University of Bath, Bath BA2 7AY, UK.

Summary

The

toxin complex

(

tc

) genes of bacteria comprise alarge and growing family whose mode of actionremains obscure. In the insect pathogen

Photor-habdus

,

tc

genes encode high molecular weightinsecticidal toxins with oral activity against caterpil-lar pests. One protein, TcdA, has recently beenexpressed in transgenic plants and shown to conferinsect resistance. These toxins therefore representalternatives to toxins from

Bacillus thuringiensis

(Bt)for deployment in transgenic crops. Levels of TcdAexpression in transgenic plants were, however, lowand the full toxicity associated with the native toxinwas not reconstituted. Here we show that increasedactivity of the toxin TcdA1 requires potentiation byeither of two pairs of gene products, TcdB1 andTccC1 or TcdB2 and TccC3. Moreover, these samepairs of proteins can also cross-potentiate a secondtoxin, TcaA1B1. To elucidate the likely functionaldomains present in these large proteins, we ex-pressed fragments of each ‘toxin’ or ‘potentiator’gene within mammalian cells. Several domains pro-duced abnormal cellular morphologies leading to celldeath, while others showed specific phenotypes suchas nuclear translocation. Our results prove that the Tctoxins are complex proteins with multiple functionaldomains. They also show that both toxin genes andtheir potentiator pairs will need to be expressed toreconstitute full activity in insect-resistant transgenicplants. Moreover, they suggest that the same potenti-ator pair will be able to cross-potentiate more thanone toxin in a single plant.

Introduction

The

toxin complex

(

tc

) genes of

Photorhabdus

bacteria

encode high molecular weight toxins with unexpectedoral activity against insects (Bowen

et al

., 1998). Oralactivity of these toxins is not anticipated as

Photorhabdus

bacteria are normally delivered directly into the openinsect blood system (haemocoel) via their nematodevectors (ffrench-Constant

et al

., 2003). Therefore duringinfection of an insect by

Photorhabdus

, any gut activetoxins would be expected to work from the haemocoelside of the gut rather than via the gut lumen. However,recent surveys of

tc

-like homologues in sequenced bac-terial genomes (Waterfield

et al

., 2001a; 2002) show that

tc

-like genes are widespread among Gram-negative bac-teria and that oral activity may therefore be an ancestralphenotype. Thus

tc

-like genes in another insect pathogen

Serratia entomophila

(termed

SepA, SepB

and

SepC

)are carried on a large conjugative plasmid and are bothnecessary and sufficient for bacteria to infect and killNew Zealand grass grubs following ingestion of recombi-nant

Escherichia coli

(Hurst

et al

., 2000). This suggeststhat

tc

genes with ancestral oral activity may have beenacquired by

Photorhabdus

and re-deployed as toxinsalso active from the haemolymph side of the gut (Black-burn

et al

., 1998).Despite the lack of knowledge of the normal biology

and mode of action of Tc proteins, one gene

tcdA

, from

Photorhabdus luminescens

ssp.

akhurstii

strain W14(here termed simply ‘W14’ for brevity), has recently beenexpressed in transgenic

Arabidopsis

plants and shown toconfer insect resistance against caterpillars of the modelinsect the Tobacco hornworm moth,

Manduca sexta

(Liu

et al

., 2003). The demonstration that these large toxingenes can be engineered and expressed in transgenicplants makes them candidate alternative toxins to thoseof

Bacillus thuringiensis

(Bt) (Schnepf

et al

., 1998). Levelsof TcdA protein within transgenic plants were, however,low and recombinant protein was not associated with thefull levels of oral activity displayed by native TcdA contain-ing complexes purified from

Photorhabdus

W14 superna-tants (Liu

et al

., 2003). This failure to restore full activitysuggests that expression of one gene alone in plants isnot sufficient to restore maximum oral activity and raisesthe question as to which elements actually synergize or‘potentiate’ the TcdA toxin to give full activity.

Our previous experiments on expression of recombi-nant

tc

genes in

E. coli

(Waterfield

et al

., 2001b), and

374

N. Waterfield

et al.

© 2004 Blackwell Publishing Ltd,

Cellular Microbiology

,

7

, 373–382

parallel experiments in the

Xenorhabdus tc

-like

xpt

genes(Sergeant

et al

., 2003), suggested that three

tc

gene com-ponents are necessary for full oral activity. Thus coexpres-sion of

tcdA1

with

tcdB1

and

tccC1

, all on the sameplasmid, confers high levels of oral activity on superna-tants of recombinant

E. coli

(Waterfield

et al

., 2001b).Similarly, for the

tc

-like genes from

Xenorhabdus

,although expression of a single gene (

xptA

) is toxic (Mor-gan

et al

., 2001), coexpression of three genes (

xptA

,

xptB

and

xptC

) was again necessary to reconstitute full oraltoxicity (Sergeant

et al

., 2003). Moreover, potentiation ofthe toxin XptA was best achieved by coexpression of bothXptB and XptC in the same bacterial cytoplasm (Sergeant

et al

., 2003). These studies of

tc

genes in both

Photorhab-dus

and

Xenorhabdus

suggest that toxins like TcdA needto be potentiated by two components – a TcdB-like ele-ment and a TccC-like element, but they tell us little abouthow this potentiation is achieved and what the likelymodes of action of the individual Tc components are. Toaddress this question in

Photorhabdus

here we examinethe potentiation of TcdA1 by TcdB1 and TccC1 in detail.We show that, like the

Xenorhabdus

toxins, the toxin TcdAis best enhanced by coexpression of both potentiatorsTcdB1 and TccC1 in the same

E. coli

cytoplasm. Surpris-ingly, we also show that TcdB1 and TccC1 can ‘cross-potentiate’ a second toxin, TcaAB, which unlike TcdAshows no toxicity on its own. This suggests that one setof potentiators (TcdB1 and TccC1) could be deployed ina transgenic plant to potentiate two different toxins (eitherTcdA or TcaAB). We investigate the potential mode of

action of the different Tc components via expression ofpotential functional domains in tissue culture cells andderive a working model for the interaction of the threedifferent subunits in reconstituting full oral activity againstinsects.

Results

Toxins and potentiators

Searches of current nucleotide, protein and motif data-bases (last search date 1 June 2004) reveal little furtherinformation, beyond the RGD (integrin-binding)-like and

Salmonella

plasmid bourne virulence factor B (SpvB)-likedomains already noted, as to the likely mode of action ofthe Tc-like toxin family. However, as previously suggested(Bowen

et al

., 1998), the two proteins encoded by the twoopen reading frames (ORFs)

tcaA

and

tcaB (

termed

tcaAB

) are clearly homologous to the toxin encoded bythe single ORF

tcdA1

(Fig. 1). Given the similarity ofTcaAB and TcdA, we investigated the hypothesis that bothof these elements are ‘toxins’ and that they can both be‘potentiated’ by the single combination of TcdB1 andTccC1. We also test the activity of a second pair of poten-tiator genes,

tcdB2

and

tccC3

, also encoded on the

P.luminescens

strain W14

tcd

pathogenicity island(Waterfield

et al

., 2002).Expression of TcdA1 alone confers oral toxicity on

E.coli

supernatants as measured by a reduction in growthweight relative to control caterpillars (Fig. 2A), confirming

Fig. 1.

Diagram of the

toxin complex

(

tc

) gene loci used in this study. The positions of regions of predicted amino acid similarity between the different toxin (

tcdA1

and

tcaAB

) and potentiator (

tcdB1, tccC1

,

tcdB2, tccC3

and

tcaC

) genes are shown by broken lines and the percentage amino acid identity/similarity is indicated. Note the shared presence of predicted RGD-like and coiled-coil motifs within the C-terminus of TcdA1 and TcaB, the similarity of the N-terminus of TcdB1, TcdB2 and TcaC to SpvB and the presence of a RCC1-like motif in all three predicted proteins (see text for discussion). Numbers refer to amino acid positions.

Toxins Potentiators

32/49 21/39 40/5632/54 59/73

57/72

74/86

tcdA1

1863-1866 2038-2247

751-934

tcaA

tcdB1

tcaC

572-582

569-579

9-371

563-5739-394

9-371

tccC1

Core 1-680

Tail 681-1043

1-669 669-961

Coil–coil RCC1 SpvBRGD

557-560

tcaB

tcdB2 tccC3

Potentiation of the Toxin complexes

375

© 2004 Blackwell Publishing Ltd,

Cellular Microbiology

,

7

, 373–382

Fig. 2.

Oral bioassay of recombinant

E. coli

expressing different combinations of ‘toxin’ (

tcdA1

and

tcaAB

) and ‘potentiator’ (

tcdB1, tcdB2

and

tccC1

,

tccC3

) genes against

Manduca

. Histograms of the relative weight gain of caterpillars fed with one in 10 dilutions of toxin added to artificial diet. Weight gain of the different treatments is expressed relative to a pBAD30 only control (always given a value of 1.0) and standard errors of each mean are shown.A. Larvae fed on

E. coli

extracts containing TcdA1 alone showed reduced weight relative to the control at a 10

-

1

dilution.B. The same TcdA1 preparation from A showed similar toxicity at the greater dilution of 10

-

2

following mixture with TcdB1 and TccC1, when they were coexpressed in the same bacterial cytoplasm, showing that TcdB1 and TccC1 can potentiate the activity of the toxin TcdA1.C. In contrast, the same TcdA1 preparation from A was not potentiated by TcdB1 and TccC1 when they were expressed in different

E. coli

cytoplasms and then mixed subsequently. This shows that coexpression of TcdB1 and TccC1 in the same bacterial cytoplasm is necessary for them to potentiate the original TcdA1 preparation.D. Control

E. coli

carrying pBAD30 only. This control treatment was set to 1.0 and treatments A–C were expressed relative to this value.E. Larvae fed on

E. coli

coexpressing the toxin genes

tcaA1B1

and the potentiator pair

tcdB1

and

tccC1

show reduced weight gain, whereas

tcaA1B1

expressed on its own is not toxic (data not shown). This shows that TcdB1 and TccC1 can cross-potentiate a different toxin TcaA1B1.F. Positive control showing levels of toxicity achieved using TcdA1 with the same TcdB1 and TccC1 preparation as in E.G. The toxin TcaA1B1 can also be cross-potentiated by a second gene pair,

tcdB2 and tccC3, again only when the potentiator pair are expressed in the same E. coli cytoplasm (data not shown).H. Finally, the toxin TcdA1 can also be cross-potentiated by the second potentiator gene pair tcdB2 and tccC3, showing that more than one potentiator pair can potentiate both toxins.I. Control E. coli carrying pBAD30 only. This control treatment was set to 1.0 and treatments E-H were expressed relative to this value.

Toxin

Rel

ativ

e w

eigh

tR

elat

ive

wei

ght

Potentiators

tcaA1 tcaB1tcaA1 tcaB1

tcdA1

tccC1

tcdA1

tcdB1 tccC1

tcdA1

tcdB1 tccC1 tcdB2 tccC3 tcdB2 tccC3

pBAD

pBAD

tcdA1

tcdB1 tccC1 tcdB1

tcdA1

Toxin

Potentiators

11

10-110-210-3

10-1 10-2 10-3 10-1 10-2 10-3 10-1 10-2 10-3

10-110-210-3 10-110-210-3 10-110-210-31 1 1 1 1

1 1

0

0.2

0.4

0.6

0.8

1

1.2

0

0.2

0.4

0.6

0.8

1

1.2

tcdA1

A

E F G H I

B C D

376 N. Waterfield et al.

© 2004 Blackwell Publishing Ltd, Cellular Microbiology, 7, 373–382

that TcdA1 is indeed toxic on its own, as anticipated byits proven ability to confer insect resistance to transgenicplants (Liu et al., 2003). However this toxicity is increased(potentiated) fivefold when the TcdA1 protein is premixedwith extracts of an E. coli strain coexpressing two furtherproteins TcdB1 and TccC1 in the same bacterial cyto-plasm (Fig. 2B). In contrast, when the potentiators TcdB1and TccC1 are expressed separately in different bacterialcytoplasms, and then mixed with TcdA1 prior to feedingto caterpillars, the toxicity of TcdA1 is not enhanced(Fig. 2C). Interestingly the same potentiators TcdB1 andTccC1 can also cross-potentiate a different toxin TcaAB(Fig. 2E) which, unlike TcdA1, shows little or no activitywhen fed to caterpillars on its own (data not shown). Thetoxin TcaAB can also be cross-potentiated by the secondpotentiator pair, TcdB2 and TccC3 (Fig. 2G). SDS-PAGEanalysis of the resulting proteins shows that all of thepresumptive Tc components are produced from theseplasmids (Fig. 3), with the exception of TccC, which canonly be visualized on a native gel (data not shown).Moreover, close examination of TcdB1 expressed in thepresence of TccC1 shows a reduction in its apparentmolecular weight (Fig. 3B). This suggests not only thatTcdB1 and TccC1 are potentiators of oral toxicity but alsothat TccC1 may cleave, or otherwise modify, the TcdB1component when coexpressed.

Expression of toxin components in transfected cells

To examine the potential mode of action of this complexthree-component system, we transfected individualdomains of each of the two toxins, and two potentiators,

into mammalian tissue culture cells and co-transfected thecells with a construct expressing EGFP-actin in order toexamine their resulting morphology. Control cells, trans-fected with pEGFP-actin alone, or co-transfected withpEGFP-actin and an empty pRK5myc vector, showed nor-mal EGFP visible actin cytoskeletons (Figs 4A, D and 5A,D). In contrast, cells expressing either the large N-terminaldomain of TcdA1, or the second protein from the tca locus,TcaB, showed massive re-arrangement of their actincytoskeleton. Thus 24 h after transfection most of thetransfected cells have detached from the substrate andremaining cells showed a heavily condensed cytoskeleton(Fig. 4B and F). In contrast, cells transfected with the 3¢end of tcdA1 (Fig. 4C) or the first tca gene, tcaA (Fig. 4E),were unaffected and showed normal morphology.Although the actin cytoskeleton and gross morphology ofcells transfected with the 3¢ end of tcdA1 were unaffected,particulate aggregations of cMyc positive protein can beobserved in the cytoplasm of transfected cells (arrow inFig. 4C). Similar aggregations were also seen within therecombinant E. coli expressing an N-terminal fusionbetween GFP and this same C-terminus (data not shown).We suggest these visible aggregations, both within thehost bacteria and within transfected cells, are protein–protein interactions mediated by the predicted coiled-coilspresent in this region of TcdA (Fig. 1), which may beimportant for the formation of the mature Tc complex.

To test if the potentiator genes themselves also hadtoxic effects when transfected into tissue culture cells, weperformed similar experiments with constructs expressingsections of the genes tcdB1 and tccC1. Surprisingly,despite their absence of oral toxicity when expressed on

Fig. 3. Coomassie strained SDS-PAGE gels showing the production of the appropriate polypeptides by recombinant E. coli expressing different combinations of toxins (TcdA1 and TcaA1B1) and potentiators (TcdB1 and TccC1).A. A total of 10% PAGE gel showing the presence of full-length polypeptides for the toxins TcdA1 and TcaA1B1 and the potentiator TcdB1. Note that TccC1 cannot be detected on an SDS-PAGE gel but is visible on native gels (data not shown).B. A total of 6% PAGE gel showing that the apparent size of TcdB1 is reduced (lower molecular weight species indicated by *) upon coexpression of TccC1 within the same bacterial cytoplasm. Note that the molecular basis of this change has not been identified (see text).

tcdB1 tccC1 tcdB1 tccC1 tcdB1 tccC1 tcdB1 tcdB1 tccC1 tccC1

TcaB1

TcdB1TcdA1

TcdB1TcdB1+TccC1

TcaA1

tcaA1 tcaB1tcaA1 tcaB1

pBAD30

tcdA1

PotentiatorsToxins

83

116

62

175180

tcdA1

A B

*

Potentiation of the Toxin complexes 377

© 2004 Blackwell Publishing Ltd, Cellular Microbiology, 7, 373–382

their own, the C-terminal domain of TcdB1 also causessevere contraction of the actin cytoskeleton in transfectedcells (Fig. 5C). Moreover the N-terminus of TcdB1, whichcarries similarity to SpvB (Bowen et al., 1998), was trans-located into the nucleus of transfected cells, as confirmedby colocalization of the protein with DAPI stained nuclei(Fig. 5B). Finally, although transfection of cells with con-structs containing either the entire tccC1 gene from Pho-torhabdus W14 (Fig. 5E) or a tccC-like homologue fromYersinia pestis caused an unusual punctate pattern ofactin staining (Fig. 5F), no dramatic toxicity was observedin tccC1 or tccC-like transfected cells.

Discussion

Photorhabdus is an insect pathogen vectored by ento-mopathogenic nematodes (ffrench-Constant et al., 2003).Recent sequencing of Photorhabdus genomes has shown

that this group of species contain a wide range of toxinsand hydrolytic enzymes, including proteases, presumablyused to kill their insect hosts (Waterfield et al., 2002;Duchaud et al., 2003; Brugirard-Ricaud et al., 2004). Onesuch class of proteins are the Toxin complexes (Tcs),discovered via their oral toxicity to caterpillar pests(Bowen et al., 1998). Our previous work suggested that toconfer full tc-associated oral activity on E. coli superna-tants, that coexpression of three elements tcdA1, tcdB1and tccC1 (here termed A, B and C for simplicity) on thesame plasmid was necessary (Waterfield et al., 2001b).Parallel work in Tc-like toxins in Xenorhabdus bacteria,however, showed that tcdB-like and tccC-like (or simply Band C) elements could be expressed separately from thetcdA-like toxin, but that they had to be expressed in thesame bacterial cytoplasm (Sergeant et al., 2003). A singlemore recent study suggests that expression of a singleXenorhabdus tccC-like gene alone can confer injectable

Fig. 4. Within cell expression of different components of the two tc toxin encoding loci tcdA1 and tcaAB.A and D are representative examples of control cells transfected with pEGFP-actin alone (green) or co-transfected with pEGFP-actin (green) and pRK5myc (red).B. Cells transfected with the N-terminal two-thirds of tcdA1 show marked retraction of the actin cytoskeleton after 24 h. Note that the intense yellow staining corresponds to the colocalization of both myc-tagged TcdA1 (red) and actin-EGFP (green).C. Cells transfected with the remaining C-terminal third of TcdA1 show particulate aggregations of expressed protein in their cytoplasms (arrow)E. Expression of the first open reading frame (tcaA1) in the tcaAB locus appears to have no effect on transfected cells, which show normal morphology.F. In contrast, expression of the second open reading frame (tcaB1) again causes massive contraction of the actin cytoskeleton (see text).

tcdA1 tcdA1

tcaA1 tcaB1 tcaA1 tcaB1

A B C

D E F

Toxins

TcdA

TcaAB

pEGFP-actin

pRK5myc+pEGFP-actin

378 N. Waterfield et al.

© 2004 Blackwell Publishing Ltd, Cellular Microbiology, 7, 373–382

Fig. 5. Within cell expression of different components of the potentiator encoding genes tcdB1 and tccC.A and D are representative examples of control cells transfected with pEGFP-actin alone (green) or co-transfected with pEGFP-actin (green) and pRK5myc (red).B. Cells transfected with the N-terminus of TcdB1 show translocation of the myc-tagged fusion protein (red) to the nucleus, whereas their cellular morphology is normal.C. Cells transfected with the C-terminus of TcdB1 show massive contraction of the actin cytoskeleton, similar to the morphology achieved from transfection with the N-terminus of TcdA1 and all of TcaB. This suggests that this ‘potentiator’ gene product is toxic to cells when expressed internally.E and F. Expression of either tccC1 from Photorhabdus or a tccC-like gene from Y. pestis within cells has no affect visible effect on their morphology.

tcdB1 tcdB1

A B C

D

Potentiators

TcdB

pEGFP-actin

pRK5myc+pEGFP-actin

TccC

E F

tccC1

yp1

YPO3674

yp1

activity against the wax moth Galleria (Joo Lee et al.,2004). However, this study did not examine oral activity tothe insects tested and also used very few test insects andto date we have not observed any oral activity of E. colipreparations expressing Photorhabdus TccC proteins ontheir own (data not shown).

Here we have shown that, like the restoration of full oralactivity in Xenorhabdus tc-like toxins (Sergeant et al.,2003), the same in true for Photorhabdus toxins. Thus thetoxin TcdA1 can be ‘potentiated’ by TcdB1 and TccC1 butonly when the latter are expressed together in the samebacterial cytoplasm. Moreover, and unexpectedly, thesame two potentiators (TcdB1 and TccC1) can also poten-tiate (i.e. ‘cross-potentiate’) a second, homologous Pho-torhabdus toxin TcaAB. Further, when PhotorhabdustcdB1 and tccC genes are transcribed in the same bacte-rial cytoplasm, the expressed TcdB1 peptide shows alower than expected molecular weight (i.e. migrates fasteron an SDS-PAGE gel). These findings suggest that TcdB1and TccC1 may be useful as potentiators for increasing

the toxicity of more than one toxin (in this case both TcdA1and TcaAB) at the same time and therefore that they maybe able to act as common potentiators for two toxinsexpressed in the same transgenic plant. This strategy isgenerically termed toxin ‘stacking’ and may be useful formanagement of resistance in treated insect populations ifthe two different toxins are shown to have a different modeof action. Importantly, we stress that the current observa-tions on Photorhabdus W14 Tc toxins only relate to oraltoxicity to a single caterpillar pest, M. sexta. Thus,although TcdB1 and TccC1 act as ‘potentiators’ of TcdA1mediated toxicity against M. sexta specifically, we cannotexclude the possibility that they are toxins in their ownright. For example, TcdB1 may be orally toxic to other,more sensitive, insect pests. This parallel hypothesisgains some support from the observation that parts ofTcdB1 are themselves toxic when expressed in tissueculture cells (Fig. 4). Further, paired homologues of TcdBand TccC can be found in the genomes of a diverse arrayof other bacteria (Waterfield et al., 2001a), supporting the

Potentiation of the Toxin complexes 379

© 2004 Blackwell Publishing Ltd, Cellular Microbiology, 7, 373–382

suggestion that they may play a role independent of TcdAhomologues in other species.

The expression of tc gene fragments within tissue cul-ture cells provides a first indication of what the potentialbiological role of the different domains of these large andcomplex toxins. Although the Tc toxins are expressed onthe outer surface of Photorhabdus bacteria within infectedinsects (Silva et al., 2002), application of orally toxic highmolecular weight preparations of TcdA1 show no cellularphenotypes when applied to NIH3T3 cells, suggestingthat topical application of Tc toxins to cell cultures is nota viable method of dissecting their functionality. Theresults presented here on expression of gene fragmentswithin tissue culture cells support the working hypothesesderived for the functionality of the different toxin domains.Thus expression of the N-terminus of TcdA1 is toxic tocells, whereas the C-terminus containing the predictedcoil–coil domains appears to promote protein aggrega-tion. For the TcaA1B1 toxin, it is expression of TcaB1 thatis toxic to cells. As TcdA1 and TcaB1 have a region ofsignificant amino acid similarity (Fig. 1) this may be theregion promoting cell toxicity. Similarly with the potentiatorTcdB1, expression of the N-terminus within cells promotesnuclear translocation of the associated fusion protein,potentially suggesting a functional role for the RCC1domain (see below). In contrast, expression of TccC1within cells appears to leave their cellular morphologyunaffected, with the associated cMyc-labelled fusion pro-tein adopting an unexplained punctate pattern of staining(Fig. 5E and F)

To provide a framework for discussion of the complexinteraction between the three different toxin elements wepresent a simple working model (Fig. 6). This model incor-porates our current observations and will provide a basis

from which to test further hypotheses for the interactionof the three toxin components, A, B and C. In this model,the A component (TcdA pictured) can interact directly withthe insect midgut epithelium and cause toxicity (Fig. 6A),further when the B and C (TcdB1 and TccC1) potentiatorcomponents are expressed in the same bacterial cyto-plasm and then added to the A component toxin, they canenhance its toxicity (Fig. 6C). Alternately, adding either ofthe B or C potentiator components produced alone failsto give a productive interaction (Fig. 6B), in other words Band C can only potentiate the activity of the toxin whenthey are expressed in the same bacterial cytoplasm aseach other. This observation supports parallel work in theTc-like Xpt toxins from Xenorhabdus (Sergeant et al.,2003). These observations suggest that B and C mayinteract with one another in the cytoplasm of the bacterialcell within which they are expressed, a hypothesis that issupported by the observation that TcdB1 shows an appar-ent reduction in molecular weight when expressed in thesame cytoplasm as TccC1 (Fig. 3B). In our model(Fig. 6C), we therefore infer that a modification of the Belement by the C element is necessary for full potentiationof the A toxin. However, the precise nature of this modifi-cation remains unclear.

This data raises several new questions about themode of action of the complex, multidomain, Tc proteins.First, what are the relative levels of toxicity of the twodifferent A toxins, TcdA and TcaAB? This question ishard to answer with the experiments presented herewithout further quantification of the amount of each toxinproduced in recombinant E. coli. However, much like theP. luminescens W14 bacteria themselves, recombinantE. coli make higher levels of the toxin TcaAB than TcdA,suggesting that TcaAB may also be useful in making

AA B

Ab

b

b? ? ?

Toxic

A B C

Toxic Highly toxic

no productive interaction productive interaction

modification?

tcdB1 tccC1tcdA1 tcdA1 tcdB1tcdA1

BC

Fig. 6. Working model of the interaction of the different toxin and potentiator gene products, where boxes represent separate bacterial expression strains.A. Recombinant TcdA1 expressed in E. coli can bind to insect gut cells and cause low levels of toxicity. Note that the precise mechanisms whereby the Tc toxins exert their effects on cellular components are still unclear (?).B. Recombinant TcdA1 and TcdB1 made in dif-ferent E. coli strains and then mixed prior to insect bioassay do not produce a productive interaction, i.e. recombinant TcdB1 alone can-not potentiate the toxicity of TcdA1.C. When recombinant TcdB1 is made in the same bacterial cytoplasm as TccC1, then sub-sequent mixing with TcdA1 produces a produc-tive interaction, i.e. the toxicity of TcdA1 is potentiated. Note that the modification of TcdB1 when produced in the presence of TccC1 can be visualized as an apparent truncation, or other modification, of TcdB1 (see Fig. 3B). Note that function of the inferred translocation (arrow) of TcdB1 to the nucleus (see Fig. 4B) is unclear.

380 N. Waterfield et al.

© 2004 Blackwell Publishing Ltd, Cellular Microbiology, 7, 373–382

insect-resistant transgenic plants. Second, what is theexact role of the potentiators? Although we can docu-ment an apparent change in the molecular weight ofTcdB1 when produced in the same bacterial cytoplasmas TccC1, we have not determined the nature of thischange and therefore cannot determine if TcdB1 needsto be cleaved, or otherwise modified, in order to carryout its potentiating role. Third, what is the role of theapparent nuclear localization of the N-terminus ofTcdB1? The N-terminus of TcdB1 shows predictedamino acid similarity to the N-terminus of SpvB (Bowenet al., 1998) and the remainder of the SpvB protein itselfencodes a mono(ADP-ribosyl) transferase (Otto et al.,2000). This suggests that the N-termini of both SpvBand TcdB1 are targeting domains while the remainder ofthe these two proteins carry the active sites of theseeffectors (Otto et al., 2000). In this context it is interest-ing to note that the remainder of TcdB1 carries anRCC1-like motif (Fig. 1). RCC1 is a regulator of chroma-tin condensation, which mediates nucleocytoplasmictransport, cell cycle progression, microtubule organiza-tion and nuclear envelope assembly in eukaryotic cells(Moore, 2001). The fact that the N-terminus of TcdB1 istranslocated to the nucleus and the observation that itcarries an RCC1-like motif, may suggest a potentialmode of action for TcdB1 and TcaC. However, the exactmechanisms whereby mature Tc particles, or derivedfragments, might enter cells and take advantage ofsuch a nuclear translocation mechanism remain unclear(Fig. 6). In conclusion, these questions illustrate thatthere is much still to learn about the large, complex andwidespread Tc toxins. Moreover, further studies of theirmode of action may also shed light on other virulencefactors like SpvB, as the Tc proteins appear to be mosa-ics of toxin domains present, or potentially still to befound, in other bacterial virulence factors.

Experimental procedures

Cloning and recombinant expression of tc genes

The tc genes and corresponding subdomains were amplifiedfrom P. luminescens strain W14 genomic DNA using rTthDNApolymerase (Applied Biosystems). Polymerase chain reac-tion (PCR) conditions were 1.2–1.6 mM magnesium acetate,2 mM each dNTP and 1 mM each primer. Thermocycling wasperformed as follows: 93∞C for 30 s; 50–55∞C for 30 s and 68∞Cfor 10 min, for 35 cycles and a final 68∞C incubation for 20 minPCR primers, used for cloning into the prokaryotic arabinose-inducible expression plasmid pBAD30 and the CMV-promoterbased eukaryotic expression vector pRK5myc, were designed toinclude unique restriction sites for subsequent ligation.

The primer sequences (5’ to 3’) used for cloning whole ORFsinto pBAD30 were as follows: for the ORF tcdA1, primerstcdAFsacI CTTATTGAgCtcTGATGTAAAAGCAACAAGGCT and

tcdARxma ACAGGCCCCGggTCGGAGCCTGTTTTAGTA; forTcdB1 and tccC1, tcdBFxma: CGAccCGGGGCCTGTAAGGAGTTTTTATGCA and tccCRxba; CGTCCTcTaGATCCGGTAATGCTTATTGCA, for tcaA, tcaB and tcaAB, tcaAFxba: ACCTTtCtAGATGTTCTAATTCTA, tcaARsph CTTgcatgCCGGTTACTGACGATTGCT, tcaBFxba GCAATCtagAATCGTCAGTAACCGGA and tca-BRsph AAAgcAtgcTTACGCGTCTCGAATGA.

For the subcloning of specific domains into pRK5myc forexpression within mammalian cells: for subclone tcdA208 prim-ers, tcdA208RK5Fbam CAATggatccCTGATAGGCTATAACA andtcdA208RK5Rpst AATctgcagttaTGCCGGTGTAGGTATATT; forsubclone tcdA2ii, tcdAiiRK5Fbam CCTggatccGCACCTTTATCATTGCGCA and tcdAiiRK5Rpst ACGTTGctgCaGTTATTTAATGGTGTA; for TcdB1N-terminus, tcdBNRK5Fbam GTTggatccCAGAATTCACAAACATTCA and tcdBNRK5Rpst GGCctgcagttaTGGTGGCAGCGCGGTAA; for TcdB1C-terminus, tcdBCRK5FbamACCggatccCCACCACTGGAACTGGCCT and tcdBCRK5RpstTTGctgcagTTACACCAGCGCATCAG; for tccC1, tccCRK5FbamATggatccAGTCCGTCTGAGACTACTCTTT and tccCRK5RecoTAATGCTTATTGCAAGAatTcAATTAC; for tcaA, tcaARK5FbamCATggatccGTGACTGTTATGCAAAAT and tcaARK5Rpst mMCTTctgcagTTACTGACGATTGCTGAT; for tcaB, tcaBRK5FbamTTGggatccTCTGAATCTTTATTTACA and tcaBRK5Rpst GAAActgcagTTACGCGTCTCGAATGAT and for yp1, yp1RK5FbamATggatccCCCAATATATTACCCACTGATC and yp1RK5Reco =TAgaATTCTTAATTATCGATTCTGTT.

Following PCR, products were purified (Millipore Montage PCRcolumn), cut with the appropriate restriction enzyme, and thenre-purified prior to cloning. Cloning vector DNA from pBAD30 andpRK5myc was prepared (Qiagen miniprep kit), co-digested withthe relevant restriction enzymes and dephosphorylated with NEBshrimp alkaline phosphotase. Ligations were performed at a 3:1molar excess of insert to vector using the Promega T4 DNAligase rapid ligation system. Aliquots of the ligation reaction wereelectroporated into Epicenter Transformax EC100 E. coli andrecovered on Luria–Broth (LB) agar containing 100 mg ml-1 ampi-cillin. Ligation products were selected by restriction digest and atcandidate clones re-sequenced using the Epicentre EZ::Tn<tet-1> transposon-based template generation system to 3–4¥redundancy using an ABI 3700 automated DNA sequencer. If aclone could not be identified that exactly matched the corre-sponding published sequence, we tested multiple clones in trans-fection experiments and oral bioassays to rule out variationarising from PCR-induced point mutations.

The resulting clones and subclones were subsequentlyinduced for Tc protein expression and insect bioassays. Briefly,-80∞C glycerol stocks were used to inoculate 5 ml of fresh LBmedia supplemented with 0.2% glucose (w/v) and the appropri-ate antibiotic for selection. Bacteria were grown overnight at 30∞Cwith shaking, 1 ml of this culture was then harvested and re-suspended in 100 ml of the same media and incubated in anorbital incubator at 37∞C until an OD600 of 0.7–0.9 was achieved.Cells were then harvested at room temperature by centrifugationat 4000 r.p.m. for 10 min. The pellet was re-suspended in 100 mlof fresh LB, supplemented with the appropriate antibiotic and0.2% (w/v) of the inducer L-arabinose. Optimised time points forinduction were determined experimentally (specifically 2 h forTcdA1, 24 h for TcdB1 and TcdB1C2, 3 h for TccC1 and 2 h forthe pBAD30 control) and cells were then harvested. The bacterialcell pellet was re-suspended in 10 ml 1¥ PBS and sonicated (foursonications of 20 s at 45 mAmps using a Branson 450 digital

Potentiation of the Toxin complexes 381

© 2004 Blackwell Publishing Ltd, Cellular Microbiology, 7, 373–382

sonifier) fitted with a tapered probe. These freshly sonicatedsamples were used for oral toxicity bioassays and for SDS-PAGEanalysis to confirm expression of the target protein.

Oral bioassay against insects

Sonicated samples (200 ml total) from Tc-expressing strains werediluted in PBS and applied to 1 cm-3 disks of artificial wheat germdiet. Treated food blocks were allowed to dry for 20 min and thentwo first-instar M. sexta neonate larvae were placed on each foodblock and held at 25∞C for 7 days. Treated larvae were thenscored for mortality and weighed. Results are presented as rel-ative weight gain compared to the pBAD30 only control. Wheremixtures of expression Tc containing lysates were added, theywere mixed at room temperature for 30 min on the bench beforeapplication. The volume of TcdA1 or TcaAB ‘toxin’ componentwas kept constant to allow for comparisons between treatments.

Mammalian cell culture, transfection and immunofluorescense

The mammalian cell line NIH 3T3 (Swiss mouse fibroblasts) wasobtained from the European Collection of Animal Cell Cultures(ECACC, Porton Down, Salisbury, UK). Cells were cultured inDulbeccos Modified Eagle’s Media (DMEM) supplemented with10% fetal calf serum, 2% (10 ml) 1¥ penicillin and streptomycinand 1% non-essential amino acids (Sigma) and grown at 37∞C,95% air/5% carbon dioxide (v/v). Cell lines were grown as mono-layers and subcultured at 80–100% confluence. For transfection,the mammalian expression vector pRK5myc (a kind gift fromKaren Knox, MRC) was used to express Tc proteins as fusionproteins with an N-terminal Myc epitope tag (see above for detailsof cloning). For transfection, the NIH 3T3 cells were seeded at aconcentration of 1 ¥ 105 cells ml-1 into six-well plates containingethanol-sterilized borosilicate glass coverslips (BDH) and thedomain expression vectors were co-transfected with a GFP-actinconstruct using GeneJuice Transfection Reagent (Novagen), inaccordance with the manufacturer’s protocols. The GFP-actinconstruct chosen was pEGFP-actin, a mammalian expressionvector (Clontech) expressing the EGFP-actin fusion protein of thehuman codon optimized variant of green fluorescent protein(EGFP) and the gene encoding human cytoplasmic b-actin. Forimmunofluorescense, samples were fixed with 4% paraformalde-hyde (w/v) in PBS, permeabilized with 0.2% Triton X-100 andthen blocked with 10% normal donkey serum. Cells were stainedwith an anti-myc primary antibody (Invitrogen), followed with aCy3-conjugated donkey anti-mouse secondary to detect the Mycprotein. The cell nucleus was visualized with DAPI staining.Briefly, the coverslip was incubated in Hoechst 33258 DAPI stain(Sigma), diluted to a final concentration of 0.12 mg ml-1 in PBSeach time, and held for 15 min at room temperature in the dark.Slides were then rinsed five times with PBS and mounted forvisualisation. Fluorescence images were obtained using a ZeissLSM-510 confocal laser-scanning microscope (Zeiss LSM-510system with inverted Axiovert 100 M microscope).

Acknowledgements

This work was supported by a grant from the BBSRC Exploiting

Genomics initiative to R. ff-C. A.D. is supported by a BBSRCCASE studentship with Syngenta, Jealotts Hill, UK. We thank B.Reaves for provision of tissue culture facilities and practicaladvice with immunocytochemistry and M. Blight for provision ofclone tcdB1.

References

Blackburn, M., Golubeva, E., Bowen, D., and Ffrench-Constant, R.H. (1998) A novel insecticidal toxin fromPhotorhabdus luminescens, Toxin complex a (Tca), and itshistopathological effects on the midgut of Manduca sexta.Appl Environ Microbiol 64: 3036–3041.

Bowen, D., Rocheleau, T.A., Blackburn, M., Andreev, O.,Golubeva, E., Bhartia, R., and ffrench-Constant, R.H.(1998) Insecticidal toxins from the bacterium Photorhabdusluminescens. Science 280: 2129–2132.

Brugirard-Ricaud, K., Givaudan, A., Parkhill, J., Boemare, N.,Kunst, F., Zumbihl, R., and Duchaud, E. (2004) Variationin the effectors of the type III secretion system amongPhotorhabdus species as revealed by genomic analysis. JBacteriol 186: 4376–4381.

ffrench-Constant, R., Waterfield, N., Daborn, P., Joyce, S.,Bennett, H., Au, C., et al. (2003) Photorhabdus: towards afunctional genomic analysis of a symbiont and pathogen.FEMS Microbiol Rev 26: 433–456.

Duchaud, E., Rusniok, C., Frangeul, L., Buchrieser, C.,Givaudan, A., Taourit, S., et al. (2003) The genomesequence of the entomopathogenic bacterium Photorhab-dus luminescens. Nat Biotechnol 21: 1307–1313.

Hurst, M.R., Glare, T.R., Jackson, T.A., and Ronson, C.W.(2000) Plasmid-located pathogenicity determinants of Ser-ratia entomophila, the causal agent of amber disease ofgrass grub, show similarity to the insecticidal toxins ofPhotorhabdus luminescens. J Bacteriol 182: 5127–5138.

Joo Lee, P., Ahn, J.Y., Kim, Y.H., Wook Kim, S., Kim, J.Y.,Park, J.S., and Lee, J. (2004) Cloning and heterologousexpression of a novel insecticidal gene (tccC1) fromXenorhabdus nematophilus strain. Biochem Biophys ResCommun 319: 1110–1116.

Liu, D., Burton, S., Glancy, T., Li, Z.S., Hampton, R., Meade,T., and Merlo, D.J. (2003) Insect resistance conferred by283-kDa Photorhabdus luminescens protein TcdA in Ara-bidopsis thaliana. Nat Biotechnol 21: 1307–1313.

Moore, J.D. (2001) The Ran-GTPase and cell-cycle control.Bioessays 23: 77–85.

Morgan, J.A., Sergeant, M., Ellis, D., Ousley, M., and Jarrett,P. (2001) Sequence analysis of insecticidal genes fromXenorhabdus nematophilus PMFI296. Appl Environ Micro-biol 67: 2062–2069.

Otto, H., Tezcan-Merdol, D., Girisch, R., Haag, F., Rhen, M.,and Koch-Nolte, F. (2000) The spvB gene-product of theSalmonella enterica virulence plasmid is a mono (ADP-ribosyl) transferase. Mol Microbiol 37: 1106–1115.

Schnepf, E., Crickmore, N., Van Rie, J., Lereclus, D., Baum,J., Feitelson, J., et al. (1998) Bacillus thuringiensis and itspesticidal crystal proteins. Microbiol Mol Biol Rev 62: 775–806.

Sergeant, M., Jarrett, P., Ousley, M., and Morgan, J.A. (2003)Interactions of insecticidal toxin gene products fromXenorhabdus nematophilus PMFI296. Appl Environ Micro-biol 69: 3344–3349.

Silva, C.P., Waterfield, N.R., Daborn, P.J., Dean, P., Chilver,

382 N. Waterfield et al.

© 2004 Blackwell Publishing Ltd, Cellular Microbiology, 7, 373–382

T., Au, C.P., et al. (2002) Bacterial infection of a modelinsect: Photorhabdus luminescens and Manduca sexta.Cell Microbiol 4: 329–339.

Waterfield, N.R., Bowen, D.J., Fetherston, J.D., Perry, R.D.,and ffrench-Constant, R.H. (2001a) The toxin complexgenes of Photorhabdus: a growing gene family. TrendsMicrobiol 9: 185–191.

Waterfield, N., Dowling, A., Sharma, S., Daborn, P.J., Potter,U., and ffrench-Constant, R.H. (2001b) Oral toxicity of Pho-torhabdus luminescens W14 toxin complexes in Escheri-chia coli. Applied Environ Microbiol 67: 5017–5024.

Waterfield, N.R., Daborn, P.J., and ffrench-Constant, R.H.(2002) Genomic islands in Photorhabdus. Trends Microbiol10: 541–545.