Extent of Variation of the Bacillus thuringiensis Toxin Reservoir: the Case of the Geranium Bronze, Cacyreus marshalli Butler (Lepidoptera: Lycaenidae)

Despite the fact that around 200 cry genes from Bacillus thuringiensishave already been cloned, only a few Cry proteins are toxictowards a given pest. A crucial step in the mode of action ofCry proteins is binding to specific sites in the midgut of susceptibleinsects. Binding studies in insects that have developed cross-resistancediscourage the combined use of Cry proteins sharing the samebinding site. If resistance management strategies are to beimplemented, the arsenal of Cry proteins suitable to controla given pest may be not so vast as it might seem at first. Thepresent study evaluates the potential of B. thuringiensis forthe control of a new pest, the geranium bronze (Cacyreus marshalliButler), a butterfly that is threatening the popularity of geraniumsin Spain. Eleven of the most common Cry proteins from the threelepidopteran-active Cry families (Cry1, Cry2, and Cry9) weretested against the geranium bronze for their toxicity and bindingsite relationships. Using ¹²⁵I-labeled Cry1A proteins we foundthat, of the seven most active Cry proteins, six competed forbinding to the same site. For the long-term control of the geraniumbronze with B. thuringiensis-based insecticides it would beadvisable to combine any of the Cry proteins sharing the bindingsite (preferably Cry1Ab, since it is the most toxic) with thosenot competing for the same site. Cry1Ba would be the best choiceof these proteins, since it is significantly more toxic thanthe others not binding to the common site.

INTRODUCTION

Top

Introduction

Insecticidal products based on the bacterium Bacillus thuringiensisare an environmentally friendly alternative to chemical insecticidesfor biological control. The use of B. thuringiensis has increasedin the last decade due to its low toxicity to the environmentand to nontarget insects. The toxicity of B. thuringiensis isdue to proteins contained in the crystal inclusion producedduring the sporulation stage (Cry proteins) (21). In susceptibleinsects, crystals are solubilized in the midgut and the releasedproteins are activated by midgut proteases. Activated toxinscross the peritrophic membrane and bind to specific bindingsites in the apical membrane of the midgut cells, producingpores that lead to the lysis of the cell and finally to thedeath of the insect. The role of membrane binding sites is ofspecial relevance for the specificity of action of the Cry proteins(13). Studies carried out on insects that have evolved resistanceto Cry proteins have shown that the alteration of a bindingsite shared by multiple Cry proteins can lead to the loss ofeffectiveness of all of them (1, 17).

So far, approximately 200 cry genes have been cloned from awide range of B. thuringiensis strains (5) (http://www.biols.susx.ac.uk/home/Neil_Crickmore/Bt/index.html).Cry proteins have been grouped into different families basedupon sequence similarity. Despite the large number of Cry proteinsisolated, only a few families are toxic towards a given insectorder. For the Lepidoptera, only Cry1, Cry2, and Cry9 familiesare toxic (3), and for a given lepidopteran pest, only someof the Cry proteins from these families are effective. Furthermore,if resistance management strategies are to be implemented topreserve the long-term use of B. thuringiensis-based insecticides,combination of Cry proteins sharing the same binding site isdiscouraged (8).

Geraniums (Pelargonium spp.) are universally popular, with 106,350,000plants having been produced and distributed commercially in1998 in the United States (19) and an even larger productionand distribution in Europe. Recently, a lepidopteran pest, thegeranium bronze (Cacyreus marshalli Butler), is threateningthe popularity of geraniums in Spain, since these plants areno longer easy to grow because of the biweekly insecticidaltreatments that have to be conducted in order to keep the plantsfree of the pest.

The geranium bronze is a butterfly from the family Lycaenidae.Originally from southern Africa, it was introduced into Europethrough the Balearic Islands (Spain) early in the 1990s (6).The butterfly lays its eggs in the flower buds or in the leavesof the geranium. When the larva hatches, it tunnels inside thebud and later starts to burrow a gallery in the stem, whereit remains until the last instar. Galleries become infectedwith fungi and bacteria, which eventually leads to the deathof the plant (Fig. 1). The absence of natural enemies outsideits place of origin has allowed it to spread quickly, and ithas become the most important pest of cultivated geraniums inSpain. Currently, this pest is included in the EPPO A2 quarantinelist (http://www.eppo.org/QUARANTINE/lists.html). The costsof insecticide applications in geranium production have increasedby sixfold in the affected areas; while the insecticide costs$0.01 per 10-cm pot in the United States (4), the equivalentcost in Spain after the introduction of the pest is $0.06 per10-cm pot (Maribel Rodriguez, La Veguilla, personal communication).

View larger version (120K):
[in this window]
[in a new window]
FIG. 1. Effects of natural infestation of a parterre of Pelargonium plants by the geranium bronze. (A) Before infestation. (B) After infestation (two months later).

In the present study we have evaluated the potential of B. thuringiensisfor the control of this new pest, the geranium bronze. We havechosen 11 of the most common Cry proteins, distributed amongthe three lepidopteran-active Cry families. Their binding siterelationships have also been analyzed to determine their suitabilityfor use in combinations in resistance management programs.

MATERIALS AND METHODS

Top

Materials and Methods

Insects.
Eggs and larvae of the geranium bronze were collected from non-pesticide-treatedPelargonium x hortorum plants from parterres in the Campus ofBurjassot of the University of Valencia. Eggs were placed onflower petals in a petri dish containing solid 2% agar. Four-day-old,laboratory-raised larvae were transferred to geranium flowerbuds until they were required. Field-collected larvae were alsomaintained on flower buds. Insects were maintained in a rearingchamber at 25°C with 60% relative humidity and an 18-h/6-h(light/dark) photoperiod. Insects to be used for binding assayswere allowed to grow until the last instar, and then they werefrozen in liquid nitrogen and then kept at -80°C.

B. thuringiensis Cry proteins.
Cry1 proteins were prepared from recombinant B. thuringiensisstrains expressing a single toxin: Cry1Aa (EG1273), Cry1Ab (EG7077),Cry1Ac (EG11070), Cry1Ba (EG11916), Cry1Ca (EG1081), Cry1Da(EG7300), Cry1Ea (EG11901), Cry1Fa (EG11096), or Cry1Ja (EG7279).Cry2Aa was kindly supplied by William Moar (Auburn University,Auburn, Ala.) from recombinant Escherichia coli. Purified andactivated Cry9Ca was obtained from Jeroen Van Rie (Aventis CropScience,Ghent, Belgium). Cry1 and Cry2Aa proteins were purified andactivated as described elsewhere (20). For bioassay tests, Cryproteins were used in their activated form. Cry proteins usedfor binding experiments were purified by chromatography by usinga MonoQ HR 5/5 anion-exchange column (fast-protein liquid chromatographysystem from Pharmacia, Uppsala, Sweden) (20). The protein concentrationwas determined by the Bradford method (2).

Toxicity tests.
Toxicity of Cry proteins was measured as their ability to inhibitlarval growth. Growth inhibition was chosen instead of mortalityto determine the effect of the Cry proteins because, workingwith a single dose and with a reduced number of insects, growthinhibition is more sensitive for detection of differences intoxicity (10) among the less active toxins. Tests were performedat 25°C with 60% relative humidity and an 18-h/6-h (light/dark)photoperiod. Geranium flower buds were dipped in 1-mg/litertoxin solutions in 50 mM carbonate buffer (pH 10.5) with 0.02%Triton AG-98, and then they were allowed to air dry. Preweighedthird-instar larvae were allowed to feed on these buds for 48h, and then they were weighed again. Every test was performedwith 10 larvae in two independent replicates. Buds dipped inbuffer alone were used as controls. The percentage of growthinhibition (%GI) was calculated as %GI = [1-(RG_t/RG_c)]x100,where RG_t and RG_c represent the relative growth in the presenceof toxin (RG_t) or in the control (RG_c), respectively. Relativegrowth (RG) was calculated as RG = [(W₁-W₀)/W₀], where W₀ andW₁ are the initial and final weight of the larva, respectively.

Binding assays.
Cry1Ab and Cry1Ac were labeled with ¹²⁵I by the chloramine-Tmethod (24), with final specific radioactivities of 6.4 and0.72 mCi/mg, respectively. Brush border membrane vesicles (BBMV)were prepared by using dissected midguts from last-instar larvaefollowing the method described by Wolfersberger et al. (25).The protein concentration in the BBMV was determined by theBradford method (2).

Binding competition experiments were performed as describedelsewhere (26) using conditions adapted for BBMV from the geraniumbronze with both labeled Cry1A proteins, which were determinedbeforehand. Incubations were performed at room temperature for45 min with 40 µg of BBMV/ml, with either 5 ng of labeledCry1Ab/ml or 50 ng of labeled Cry1Ac/ml. Values for the dissociationconstant (K_com) and binding site concentrations (R_t) were calculatedfrom competition binding data by the procedure described byMunson and Rodbard (18). K_com is used instead of K_d (the trueequilibrium dissociation constant) to indicate that bindingof Cry proteins to BBMV is a two-step process involving reversibleplus irreversible binding.

RESULTS

Top

Results

Toxicity tests.
Growth inhibition assays with third-instar larvae of the geraniumbronze showed various degrees of toxicity for the 11 Cry proteinstested (Fig. 2). Cry1Ab was the most toxic, with a growth inhibitionclose to 100% ± standard error of the mean (99.5% ±0.5%) at the tested concentration. The toxicities of the otherCry proteins, in decreasing order, were as follows: Cry1Ac (87%± 4%), Cry1Ja (85% ± 11%), Cry1Ea (72% ±4%), Cry1Ba (70% ± 4%), Cry1Aa (64% ± 10%), Cry2Aa(59% ± 7%), Cry1Da (48% ± 7%), Cry1Fa (43% ±1%), Cry1Ca (32% ± 12%), and Cry9Ca (28% ± 7%).

View larger version (28K):
[in this window]
[in a new window]
FIG. 2. Toxicity of B. thuringiensis Cry proteins (1 mg/liter) to third-instar larvae of the geranium bronze. Bars show the means ± standard errors of two independent replicates.

Binding of Cry proteins to BBMV.
Competition assays with ¹²⁵I-Cry1Ab versus unlabeled Cry proteinswere performed to find out which Cry proteins have binding sitesin common with Cry1Ab (Fig. 3). Besides Cry1Ab (which confirmsthat binding was specific), the structurally related Cry1Aaand Cry1Ac proteins competed completely for the Cry1Ab bindingsites (Fig. 3A). Cry1Ea and Cry1Ja were also able to competecompletely for the Cry1Ab binding sites (Fig. 3B). Cry2Aa competedonly at high concentrations, and it was unable to compete completelyat the highest concentration tested (Fig. 3B). No competitionwas found with Cry1Ba, Cry1Ca, Cry1Da, Cry1Fa, or Cry9Ca (Fig.3B).

View larger version (27K):
[in this window]
[in a new window]
FIG. 3. Binding of ¹²⁵I-Cry1Ab to BBMV from the geranium bronze at increasing concentrations of unlabeled competitor. (A) Cry1Ab (solid line and •), Cry1Ac (broken line and ${blacksquare}$ ), and Cry1Aa (dotted line and ${diamondsuit}$ ). (B) Cry1Ea (broken line and •), Cry1Ja (dotted line and ${blacksquare}$ ), Cry2Aa (dashed-dotted line and ${diamondsuit}$ ), Cry1Ca ( ${triangleup}$ ), Cry1Da ( ${square}$ ), Cry1Fa ( ${triangleup}$ ), Cry1Ba ( ${circ}$ ), and Cry9Ca ( ${diamond}$ ).

Similarly, competition of ¹²⁵I-Cry1Ac with unlabeled Cry proteinsshowed that, besides Cry1Ac, Cry1Ab and Cry1Ja competed completelyfor Cry1Ac binding sites (Fig. 4). As with ¹²⁵I-Cry1Ab, Cry2Aacompeted partially with ¹²⁵I-Cry1Ac and only at high concentrations.Competition of ¹²⁵I-Cry1Ac with unlabeled Cry1Aa, Cry1Ba, Cry1Ca,Cry1Da, Cry1Ea, Cry1Fa, and Cry9Ca was not tested.

View larger version (20K):
[in this window]
[in a new window]
FIG. 4. Binding of ¹²⁵I-Cry1Ac to BBMV from the geranium bronze at increasing concentrations of unlabeled competitor. Symbols: solid line and •, Cry1Ab; broken line and ${blacksquare}$ , Cry1Ac; dotted line and ${diamondsuit}$ , Cry1Ja; dashed-dotted line and ${blacktriangleup}$ , Cry2Aa.

Quantitative estimates of the dissociation constants (K_com)and the concentrations of binding sites (R_t) were obtained forCry1Ab and Cry1Ac from the homologous competition data (whenthe labeled toxin and the unlabeled toxin were the same). Thejoint analysis of these data gave a K_com value of 0.16 nM forCry1Ab and of 0.93 nM for Cry1Ac, with an R_t value of 4.4 ±0.5 pmol/mg of BBMV.

DISCUSSION

Top

Discussion

Most commercial B. thuringiensis insecticides used for the controlof lepidopteran pests contain proteins of the Cry1A family (14).According to our results, two Cry1A proteins (Cry1Ab and Cry1Ac)are among the most toxic ones against the geranium bronze. Thisresult is in agreement with the observation that commercialB. thuringiensis insecticides (Dipel2X [Abbot Laboratories,Chicago, Ill.], Delfin [Sandoz Agro, Basel, Switzerland], andXentari [Bayer]) are toxic against this pest (Herrero et al.,unpublished data). Other Cry proteins contained in relativelycommon B. thuringiensis commercial products are Cry1Aa, Cry1Ca,Cry1Da, and Cry2Aa, which are significantly less active againstthis pest.

Homologous and heterologous competition experiments using Cry1Aband Cry1Ac revealed that these two proteins share the same bindingsites, since both compete completely against each other in reciprocaltests. In a previous study, reciprocal heterologous bindingusing biotin-labeled Cry1Ja and ¹²⁵I-labeled Cry1Ac indicatedthat these two proteins share the same binding sites (12). Therefore,there seems to be just one binding site for Cry1Ab, Cry1Ac,and Cry1Ja. Because Cry1Ab and Cry1Ac shared the same bindingsites and because the heterologous binding pattern was essentiallyidentical for both toxins, competition experiments using ¹²⁵I-Cry1Acwith Cry1Ba, Cry1Ca, Cry1Da, Cry1Fa, and Cry9Ca were omitted.Despite the fact that the adjustment of the competition databy the LIGAND program indicated a single binding site for Cry1Aband Cry1Ac, the observation that Cry2Aa did not compete completelyat the highest concentration tested indicates the occurrenceof two distinct populations of binding sites instead of one,with Cry2Aa competing for just one of them. Additional experimentsincluding the use of higher concentrations of Cry2Aa would berequired to prove or disprove this fact.

In summary, 6 of the 11 Cry proteins tested (Cry1Aa, Cry1Ab,Cry1Ac, Cry1Ea, Cry1Ja, and Cry2Aa) bind to a common site, andnotably, these Cry proteins happen to be the most toxic ones,along with Cry1Ba. Nevertheless, heterologous competition usinglabeled Cry1Aa, Cry1Ea, and Cry2Aa was not performed, and thus,the possibility that these proteins can bind to an additionalsite not shared with Cry1Ab and Cry1Ac remains open. Furthermore,although Cry2Aa competes for saturable sites for Cry1 toxins,the unique mode of action proposed for Cry2Aa (7) makes it possiblethat a hypothetical loss of binding of this toxin to the commonbinding site would have only a small effect on its toxicity.

It is important to understand that even if two Cry proteinsapparently compete for the same binding site they might reallybe binding to different epitopes. These may be located eitherin the same molecule or in a nearby molecule (that could beforming part of a cluster) but in such a way that binding ofone Cry protein hinders binding of the other. However, evidencewith other insect pests indicates that mutations in a singlegene can confer resistance to Cry proteins sharing a same bindingsite, most likely by an alteration in a membrane molecule (8).In Plutella xylostella, Cry1Aa, Cry1Ab, Cry1Ac, Cry1Fa, andCry1Ja share a common binding site (1, 11, 12) and a singlegene is responsible for conferring resistance to all five Cryproteins (22, 23). In Heliothis virescens, resistance to fourCry proteins (Cry1Aa, Cry1Ab, Cry1Ac, and Cry1Fa) seems to beconferred by a single gene (9, 10) and also seems to be associatedwith the alteration of a common binding site (15, 16).

Our toxicity data indicate that B. thuringiensis can be usedto control the geranium bronze, with some Cry proteins beingmore active than others. However, the binding data show thatdevelopment of resistance to several of these toxins simultaneouslyshould not be surprising. Since the most toxic Cry proteinsfound in some of the widespread B. thuringiensis-based insecticidesshare the same binding site, its use could favor the evolutionof simultaneous resistance to all components or to the maincomponents of the mixture. For the long-term use of B. thuringiensis-basedinsecticides to control the geranium bronze it would be advisableto combine any of the Cry proteins sharing the binding site(preferably Cry1Ab, since it is the most toxic) with those notcompeting for the same site. Of these latter proteins, Cry1Bawould be the best choice, since it is significantly more toxicthan other Cry proteins not binding to the common site. Despitethe fact that our tests have been performed with a representativefraction of the lepidopteran-active Cry proteins, for effectivecontrol of the geranium bronze we are apparently left with aquite restricted number of toxin combinations that seem advisablefrom a resistance management standpoint.

It is important to note that the arguments given above are basedon our binding site model, which does not take into accountunexpected synergistic effects arising from combinations ofCry proteins or even other mechanisms of resistance not relatedto binding site modification. Despite this limitation, the bindingsite model among Cry proteins is the only way to predict cross-resistanceso far.

It is widely accepted that, despite the high number of B. thuringiensistoxins available, only a small number are useful for effectivecontrol of a determined insect pest. The results of the presentstudy show that this number can be severely diminished if resistancemanagement is to be considered to preserve the long-term useof B. thuringiensis-based products. We therefore encourage thestudy of the binding relationships of Cry proteins before anew pest is targeted. Moreover, despite the large number ofB. thuringiensis toxins that have been described, the need tokeep searching for new toxins with different modes of actionor that use binding sites different from the most commonly utilizedCry proteins is evident.

ACKNOWLEDGMENTS

We thank L. Calzada Grau for technical assistance, Ecogen, Inc.,for providing the recombinant strains used to prepare toxins,Jeroen Van Rie for providing Cry9Ca toxin, and William Moarfor providing Cry2Aa toxin.

This work was supported by grant 1FD1997-0917-C02-01 from theMinisterio de Ciencia y Tecnología, Spain.

FOOTNOTES

* Corresponding author. Mailing address: Department of Genetics, University of Valencia, Dr. Moliner 50, 46100 Burjassot (Valencia), Spain. Phone: (34) 96 386 4506. Fax: (34) 96 398 3029. E-mail: Juan.Ferre{at}uv.es.

REFERENCES

Top