77

Journal of Protein Chemistry, Vol. 22, No. 1, January 2003 (© 2003)

0277-8033/03/0100-0077/0 © 2003 Plenum Publishing Corporation

Molecular Cloning of a-Amylases from Cotton Boll Weevil,Anthonomus grandisand Structural Relations to PlantInhibitors: An Approach to Insect Resistance

Osmundo B. Oliveira-Neto,1,2 João A. N. Batista,1 Daniel J. Rigden,1 Octávio L. Franco,1,3

Rosana Falcão,1 Rodrigo R. Fragoso,1,2 Luciane V. Mello,1 Roseane C. dos Santos,1,4

and Maria F. Grossi-de-Sá1,5

Received September 23, 2002

Anthonomus grandis, the cotton boll weevil, causes severe cotton crop losses in North and SouthAmerica. Here we demonstrate the presence of starch in the cotton pollen grains and young ovules thatare the main A. grandisfood source. We further demonstrate the presence of a-amylase activity, anessential enzyme of carbohydrate metabolism for many crop pests, in A. grandismidgut. Twoa-amylase cDNAs from A. grandislarvae were isolated using RT-PCR followed by 59 and 39 RACEtechniques. These encode proteins with predicted molecular masses of 50.8 and 52.7 kDa, respectively,which share 58% amino acid identity. Expression of both genes is induced upon feeding and concen-trated in the midgut of adult insects. Several a-amylase inhibitors from plants were assayed againstA. grandisa-amylases but, unexpectedly, only the BIII inhibitor from rye kernels proved highly effec-tive, with inhibitors generally active against other insect amylases lacking effect. Structural modelingof Amylag1and Amylag2showed that different factors seem to be responsible for the lack of effect of0.19 and a-AI1 inhibitors on A. grandisa-amylase activity. This work suggests that genetic engi-neering of cotton to express a-amylase inhibitors may offer a novel route to A. grandisresistance.

KEY WORDS: a-Amylase inhibitors; cotton boll weevil; cDNA cloning; plant defense; Anthonomus grandis.

1 Embrapa Recursos Genéticos e Biotecnologia, Brasilia-DF, Brazil.2 Depto. de Biologia Celular, Universidade de Brasília, Brasília-DF,Brazil.

3 Pós-Graduação em Ciências Genômicas e Biotecnologia, Universi-dade Católica de Brasília, Brasília-DF, Brazil.

4 EMBRAPA-Algodão, Campina Grande, PB, Brazil.5 To whom correspondence should be addressed at Embrapa RecursosGenéticos e Biotecnologia, S.A.I.N. Parque Rural, Final W5, AsaNorte, 70770-900, Brasília, DF, Brazil. E-mail: fatimasa@cenargen.embrapa.br

1. INTRODUCTION

The cotton boll weevil, Anthonomus grandis(Boheman,1843), is a major insect pest of cultivated cotton, Gossyp-ium hirsutumL., being responsible for significant cottonlosses in South and North Americas. The adult femalemoves to the flower bud after alighting on the plant and

6 Abbreviations: a-AI, a-amylase inhibitor; Amylag1and 2, a-amylasefrom Anthonomus grandis1 and 2; Bt, Bacillus thuringiensis; RACE,rapid amplification of cDNA ends; RT-PCR, reverse transcriptase–polymerase chain reaction; PPA, porcine pancreatic a-amylase; SaAI5,sorghum a-amylase inhibitor 5; TMA, Tenebrio molitora-amylase.

then proceeds to construct an oviposition hole in whichthe egg is placed and which is then sealed. Floral budsand bolls are the main A. grandisfood, in which femaleinsects depend on nutrients from pollen grains andovules for reproduction and eggs development (Bottrell,1983). Larval behavior, such as movement into protectedplaces on the bud after egg hatching, has made this in-sect difficult to control using conventional insecticidesand management.

The use of chemical pesticides leads to high produc-tion costs as well as causing risks to human health. In thelight of these considerations, plant genetic transformation

78 Oliveira-Neto et al.

with exogenous genes encoding factors of resistance tophytophageous insects is a modern and attractive alter-native to synthetic chemical insecticides for the controlof several aggressive plant pests (Estruch et al., 1997).Efforts have been focused on the screening of differentproteins known to have insect control properties, such asBt-toxins, proteinase inhibitors, lectins, and a-amylase in-hibitors (Boulter, 1993; Gatehouse and Gatehouse, 1998;Carlini and Grossi-de-Sá, 2002). Cotton improvementthrough genetic engineering has become a reality withthe successful testing of insect-resistant transgenic cottonexpressing protease inhibitors (Thomas et al., 1995) andBt-toxin (Jenkins et al., 1995).

a-Amylase (a-1,4-glucan-4-glucanohydrolase, EC3.2.1.1)6 catalyzes the hydrolysis of the a-(1,4)-glycosidiclinkages of starch components, glycogen and variousoligosaccharides that are widespread in nature, beingfound in animals, microorganisms and plants. Insect andmammalian a-amylases have been characterized from abiochemical, molecular and structural point of view inconsiderable detail (Franco et al., 2000; MacGregor et al.,2001). Because of their important biochemical roles ininsect growth and development, when the action of thea-amylases is inhibited, insect nutrition is impaired. Asstrategies of control, inhibitors to insect amylase havebeen already demonstrated to be an important biotech-nology system in the control of insect-pests. Pea andazuki transgenic plants expressing a-amylase inhibitorsfrom common beans (a-AI) were completely resistant tothe Bruchus pisorumand Callosobruchus chinensiswee-vil (Morton et al., 2000).

Here we show the presence of starch in the floralbuds of the cotton plant. Subsequent investigationshowed that A. grandisexpresses two different amylases.These were both cloned and characterized. Additionally,a range of a-amylase inhibitors was assayed againstA. grandisa-amylases. Unusually, only an inhibitor pres-ent in rye kernels proved effective toward the digestiveamylases from A. grandis.

2. MATERIALS AND METHODS

2.1. Insect Rearing

A population of A. grandis(Coleoptera: Curculion-idae) originally obtained from CIRAD (Montpellier,France) was maintained at 27 6 1°C, 70 6 10% relativehumidity with 14-hr day length. Insects were routinelymaintained on standard rearing diet as described byMonnerat et al. (2000). All components were purchasedfrom Sigma (St. Louis, MO). Wild insects were collectedin Unaí, Brazil.

2.2. Extraction of Larval a-Amylase and a-AmylaseInhibitory Assays

Midguts from third instar larvae and 10-day-oldadults were excised from cold-anesthetized larvae andinsect adults and ground in cold 0.15 M NaCl using a1:10 w/v ratio. The homogenate was centrifuged twice at10,000 3 g for 20 min and stored at 220°C. The clearedhomogenates were used for the amylolytic inhibitionassays. a-Amylase activity was measured by the Bernfeldmethod (1955) using a standard concentration at 50 mg/ml.The enzyme was dissolved in 100 mM sodium phosphatebuffer (pH 5.8) containing 10 mM NaCl and 1 mM CaCl2.The enzymatic reaction occurred for 15 min at 37°C usingsoluble starch as substrate. Protein concentration wasdetermined as described by Bradford (1976). Each assaywas done in triplicate, with a maximum difference of 10%.

2.3. Plant a-Amylase Inhibitor Purification

The a-amylase inhibitors 0.19 and 0.53 were purifiedfrom BR35 wheat (Triticum aestivum) kernels accordingto Franco et al. (2000). The inhibitor BIII from rye (Se-cale cereale) kernels was purified according to Iulek et al.(2000). The inhibitor SaAI5 from Sorghum bicolorseedswas purified according to Bloch and Richardson (1991),and the common bean seed a-amylase inhibitors (a-AI1and a-AI2) were purified according to Grossi-de-Sá et al.(1997). The purification progress was monitored byMALDI-TOF and SDS-PAGE analyses (Laemmli, 1970;Franco et al., 2000).

2.4. RT-PCR, 59 and 39 RACE Amplifications

Reverse transcription of A. grandistotal RNA wasperformed using oligo(dT)-anchor primer and AMV-RT(Boehringer Mannheim) according to the manufacturer’sprotocol. For the RT-PCR amplifications, two primers,AmyA (59-GGGTGTCGTGGTTGTCGACGAA-39) andAmyB (59-GACGTTGTGTTCAACCATATGGCTG-39),were designed, which correspond to short sequences con-served in a number of insect a-amylases. Amplificationwas carried out in a PTC-100 Programmable ThermalController (MJ Research) using Taq DNA Polymerase(GIBCO) under the following conditions: 2 min at 94°Cthen 30 cycles of 30 s at 94°C, 45 s at 52°C and 1 min at72°C plus an extension step for 5 min at 72°C. To ob-tain the complete cDNA sequences, the 59 and 39 endswere amplified using a 59/39 RACE Kit (BoehringerMannheim) according to the manufacturer’s instructionsusing specific primers designed based on the previous am-plified sequence. The amplified cDNAs were cloned intothe plasmid vector pGEM-T Easy (Promega, Madison,

Molecular Cloning of a-Amylases and Structural Relations to Inhibitors 79

WI), and recombinant clones were sequenced in bothstrands in an automated DNA sequencer. Computer analy-sis of the DNA and amino acid sequences were performedusing the GCG package (Genetics Computer Group, Inc.).

2.5. Southern Blots

A. grandis, genomic DNA was isolated as describedpreviously. DNA digests (15 mg per lane) were separatedon 0.8% agarose gel and transferred to Hybond-N1

(Amersham) nylon membranes using standard proce-dures (Sambrook et al., 1989). The complete Amylag1and Amylag2cDNAs were labeled with [a-32P]dCTP tohigh specific activity using the Rediprime DNA-labelingkit (Amersham) and used to probe the blots. Filters werewashed with 0.1 3 SSC at 65°C.7

2.6. Northern Blots

Total RNA from A. grandisdevelopmental stagescorresponding to third instar larvae, third instar larvaemidguts, pupa, nonfeeding adult, 10-day-old adults and10-day-old adult midguts was prepared using the RNeasyRNA extraction kit (Qiagen). Gel electrophoresis (Fourneyet al., 1988) (8 mg per lane) and hybridizations (Sam-brook et al., 1989) were performed according to standardprocedures.

2.7. Light Microscopy

Fresh anthers and pollen grains of young buds (withdiameters of 0.6 and 1 cm, respectively) and their respec-tive ovules were collected from cotton plants (Gossypiumhirsutumcv. BRS 186). The tissues were fixed in 2% glu-taraldehyde in 0.05 M sodium cacodylate buffer, pH 7.3,for 2 hr under vacuum at room temperature. After wash-ing in the same buffer, the tissues were dehydrated usingincreasing concentrations of ethanol (10%, 20%, 35%,50% and 70%), 30 min at each concentration, followedby two 1 hr exposures to 100% ethanol. Samples wereembedded at room temperature in ethanol/LR White (1:1)overnight with two changes of 100% LR White 3 hr each.The embedding was done in plastic capsules containingfresh LR White, and polymerization occurred at 70°C for48 hr. Sections of 500 nm were staining with Lugol’ssolution. Anthers with pollen grains and their respectiveovules were also stained with 1% toluidine blue. Speci-mens were viewed on a Zeiss-Axiophot and photographedusing polarized light.

2.8. Sequence Analysis and Molecular Modeling

Sequence alignments were obtained using CLUSTALW (Thompson et al., 1994) and processed usingALSCRIPT (Barton, 1993). Protein modeling was car-ried out using MODELLER-6 (Sali and Blundell, 1993).Structural superpositions were obtained using the CEserver (cl.sdsc.edu/ce.html; Shindyalov and Bourne,1998) or LSQMAN (Kleywegt et al., 2001). Geometricanalysis of protein structures was made with PROCHECK(Laskowski et al., 1993). Experimentally determined andmodeled amylase-inhibitor complexes were analyzedusing the protein–protein Interaction Server (http://www.biochem.ucl.ac.uk/bsm/PP/server/; Jones and Thornton,1996). Particular attention was paid to the interface acces-sible surface area and the surface complementarity. Thislatter property is expressed as a gap volume index, beingdefined as the gap volume (empty space between the twointeracting molecules) divided by the interface surfacearea (Jones and Thornton, 1996). Hence high values indi-cate poorer interface complementarity. Structures werevisualized using O (Jones et al., 1991).

3. RESULTS

3.1. Starch Detection in Floral Buds

Anthers, pollens and ovules were collected fromyoung buds, the developmental stage favored for bollweevil feeding and oviposition. Details of anthers andovules damaged by insect feeding are shown in Fig. 1.Starch distribution in pollen grains and ovule is seen inFig. 2. Starch droplets were highly abundant in mi-crospores both stained with 1% toluidine blue (Fig. 2A)

Fig. 1. Details of young bud (0.6 cm) showing regions of anthers notdamaged (AND) or anthers (DA) and ovules damaged due boll weevilfeeding. Fresh bud collected from field of cotton plants.

7 Nucleotide sequences of clones Amylag1and Amylag2are availablein the EMBL, GenBank and DDJB data bases under the accessionnumbers AF527876 and AF527877.

80 Oliveira-Neto et al.

Fig. 2. Semi-thin sections (500 nm) of pollens (A, B, C and D) (PG-Pollen grains) and ovules (E and F)(EI-Extern integument and II-Intern integument) stained with toluidine blue (A and C) and iodinesolution (B, D, E and F) showing starch droplets (S) from fresh young buds collected from cotton plants.

and iodine dyes (Fig. 2B). Also pollen grains collectedfrom fully developed buds and before anthesis showedincreased starch droplets (Fig. 2C and 2D). At this stage,microspores showed a more internal double membrane,the intin, and another more external and dense mem-brane, the exin, which favors adherence of the pollen tostigma or to pollinator insects. Young ovules collectedfrom 0.6-cm-diameter floral buds and stained with iodinealso revealed several starch droplets, particularly in theinteguments (Fig. 2E and 2F).

3.2. Presence of a-Amylase Activity in A. grandis

Amylolytic activity was clearly present in crude ex-tracts of either larval or adult A. grandis. We observedthat essentially a-amylase activity was located in the

midgut through testing of extracts of isolated midguts.This clearly suggests a digestive role for the a-amylaseactivity, in agreement with the demonstrated presence ofstarch in cotton anthers, pollen and ovules. A compari-son of the a-amylase activity found in extracts of insectsreared either artificially or in field conditions revealed nosignificant difference (Fig. 3). All further analysis wasthen performed with laboratory-reared insects.

3.3. Cloning and Characterization of Two cDNAsEncoding Putative a-Amylases from A. grandis

To clone the a-amylases from boll weevil, twooligonucleotide primers (AmyA and AmyB), based onconserved regions found in insect a-amylases, were

Molecular Cloning of a-Amylases and Structural Relations to Inhibitors 81

Fig. 3. a-Amylase assays of midgut extract from A. grandisfed onartificial (black) and natural (white) diets. Natural diet corresponds tofloral buds and bolls, which are main A. grandisfood, in which femaleinsects depend on nutrients from pollen grains and ovules.

designed and used to amplify cDNA fragments of RNAobtained from third instar larvae by RT-PCR. Cloningand sequencing analysis yielded two cDNA fragmentsaround 570 bp with high sequence similarity witha-amylase sequences. The 59 end of both clones wasobtained by 59 RACE using antisense primers derivedfrom the RT-PCR sequences. The complete sequence ofboth genes was then obtained by 39 RACE using senseprimers located close to the 59 end of each cDNA. Thetwo complete cDNA clones were designated Amylag1and Amylag2.

The Amylag1cDNA contains an 1452 bp open read-ing frame encoding a predicted protein of 484 aminoacids, whereas the Amylag2cDNA contains an 1473 bpopen reading frame encoding a predicted protein of491 amino acids (Fig. 4). Both sequences contained pre-dicted signal peptides. By comparison with the a-amylasefrom Tenebrio molitor(TMA), for which the site of signalpeptide cleavage has been experimentally determined,the signal peptide lengths for Amylag1and Amylag2are17 and 18 residues, respectively. The pI values of thepredicted mature Amylag1and Amylag2enzymes are 5.1and 4.8, respectively; they have predicted molecularmasses 50,870 Da and 52,767 Da, and they share 58%identity at the amino acid level. Both Amylag1and Amy-lag2 cDNAs have short 59 and 39 untranslated regions,producing final transcripts ,1.5 kb long.

The deduced amino acid sequences of Amylag1andAmylag2are homologous to other a-amylase sequencesfrom insects and mammals (Fig. 5). Amylag1and Amy-

lag2 have, respectively, 57% and 50% sequence identitywith Callosobruchus chinensis, 62% and 50% iden-tity with Diabrotica virgifera virgifera, 58% and 51%identity with Zabrotes subfasciatusa-amylases and 58%and 53% sequence identity with TMA, whose three-dimensional structure is known (Strobl et al., 1998a, b).An aspartic acid residue and two glutamic acid residues(numbered Asp185, Glu219 and Glu284 in PPA) that areknown to form the active center of PPA (Qian et al.,1994) are conserved in both A. grandissequences. More-over, three histidine residues (His99, 185 and 283) thatare believed to be involved in substrate binding are alsoconserved, as are residues Asn98 and Asp155 that bindto the catalytically essential calcium ion.

To analyze gene copy number, genomic Southernblots were performed using the Amylag1and Amylag2cDNA as probes. The results for the hybridization withthe Amylag2probe are shown in Fig. 6. Hybridization ofthe labeled Amylag2probe with EcoRI and HindIII di-gestions, which cut internally to the cDNA, yielded twostronger bands, and hybridizations with XbaI and PstI,which do not cut the amplified cDNA, yielded onestronger band. The additional fainter bands present in thehybridizations were also observed in the hybridization ofthe same blot with the Amylag1probe, and correspond tothis other gene. The results of hybridization with theAmylag1probe were less clear but also indicated that itis present in, at most, a few copies.

Hybridizations of the labeled Amylag2cDNA withtotal RNA extracted from different developmental stagesof the A. grandislife cycle are shown in Fig. 7. A singleband of 1.5 kb was observed, which conforms to the sizeof the cloned cDNA. Expression of Amylag2is concen-trated mainly in the gut of 10-day-old adults, feedingadult insects and, to a lesser extent, the gut of feeding lar-vae. Expression was not detected in whole larvae, pupaand nonfeeding adults. These results indicate expressionof Amylag2is induced upon feeding and concentrated inthe gut of adult insects. A similar trend was observed forAmylag1(results not shown).

3.4. Inhibitory Assays Toward the a-Amylasesfrom A. grandisMidguts and StructuralExplanation of Inhibition Patterns

Six plant inhibitors were assayed for activity againstA. grandisa-amylases (Fig. 8). The a-amylase inhibitorsfrom wheat (0.19 and 0.53), which belong to the cereala-amylase inhibitor family, and a-AI1 from the commonbean showed negligible activity. a-AI2, also from com-mon bean, produced weak but discernible inhibition.Strong inhibition was only observed for SaAI5, extracted

82 Oliveira-Neto et al.

Fig. 4. Nucleotide sequence of the Amylag1(A) and Amylag2(B) cDNAs and deduced amino acid sequences. Untranslated regionsare shown in lower case letters, coding region in uppercase letters. These sequences have been submitted in the GenBank/EMBLdatabases. Signal peptide (bold) was predicted by SignalP V1.1 World Wide Web Server (http://www.cbs.dtu.dk/services/SignalP/#submission (Nielsen et al., 1997). O-b-GlcNAc attachment sites in a-amylase sequences (bold and underlined) were predicted byYinOYang 1.2 Prediction Server (http://www.cbs.dtu.dk/services/YinOYang/). The three residues in bold and italic indicated aminoacids highly conserved in a-amylases and are part of the active site.

from S. bicolorseeds, which inhibited around 50% of theinsect amylolytic activity, and for BIII, purified from ryekernels, which inhibited around 90% of activity. Of par-ticular note is the unexpected lack of effect of 0.19 in-hibitor on the amylase activity of A. grandis. This inhibitorwas hitherto active against all tested insect amylases, in-cluding A. obtectus, Z. subfasciatus, C. maculatus(Franco

et al., 2002), T. molitor and T. castaneum(Feng et al.,1996).

Encouraged by previous success in correlating thesequences and structures of amylase and their inhibitorswith the strength of amylase-inhibitor interaction (Francoet al., 2000), an attempt was made to understand the rea-sons for the inhibitor characteristics shown in Fig. 8.

Fig

. 5.

Alig

nmen

t of

sel

ecte

d a-a

myl

ase

sequ

ence

s. T

he a

lignm

ent

of P

PA

, T

MA

and

hum

an s

aliv

ary

amyl

ase

has

been

adj

uste

d to

refl

ect

thei

r cr

ysta

l str

uctu

res.

N

umbe

ring

and

seco

ndar

y st

ruct

ure

for

PP

A a

ndT

MA

are

sho

wn

abov

e an

d be

low

the

alig

nmen

t re

spec

tivel

y. S

hade

d re

gion

s in

dica

te r

esid

ues

pred

icte

d by

mod

elin

g to

for

m i

nter

fac

es w

ith t

he i

nhib

itor

0.19

(F

ranc

o et a

l., 2

000)

. B

oxed

res

idue

s ar

e th

ose

atth

e in

terf

aces

with

a-A

I1 in

the

cry

stal

str

uctu

res

of it

s co

mpl

exes

with

PP

A a

nd T

MA

(B

ompa

rd-G

illes

e

t a

l., 1

996;

Nah

oum

et a

l., 1

999)

.

84 Oliveira-Neto et al.

Fig. 6. Southern blot analysis of A. grandisgenomic DNA digestedwith EcoRI, HindIII, XbaI and PstI, and probed with the 32P-labeledAmylag2cDNA. DNA size markers are indicated in kb.

Fig. 8. A. grandis a-amylase inhibition assay was done accordingBernfeld (1955). Wheat (0.19 and 0.53), common bean (a-AI1 anda-AI2), rye (BIII) and sorghum (SaAI5) inhibitors were preincubatedwith digestive enzymes for 15 min. Each assay contained 2.0 units ofamylolytic activity and inhibitors were tested at standard concentrationof 300 mg/ml. One amylolytic unit was defined as the amount of thisenzyme that increase the absorbance at 532 nm by 0.1 absorbance unitduring 15 min of assay. Each treatment was carried out in triplicate.

Consideration was confined to 0.19, 0.53 and a-AI1,each of which resulted in negligible inhibition ofA. grandisa-amylase activity. In these cases it is reason-able to assume that neither Amylag1nor Amylag2is in-hibited by the respective inhibitor. Analysis of the a-AI2result is complicated by the fact that the partial inhibi-tion obtained against total activity could result equallyfrom incomplete inhibition of both Amylag1and Amy-lag2 or from complete inhibition of one combined withno inhibition of the other. Because structural considera-tions can enhance the results of simple sequence com-parisons, we combined sequence analysis with modelingof Amylag1and Amylag2. The structure of TMA (shar-ing 53%–58% sequence identity) in complex with a-AI1has been determined (Nahoum et al., 1999). No a-amylasehas yet been crystallized in complex with a member ofthe 0.19 family of cereal a-amylase inhibitors but, basedon the distant homology of the 0.19 family with bifunc-tional a-amylase/protease inhibitors, a convincing modelof TMA in complex with 0.19 has been obtained (Francoet al., 2000). The degree of sequence identity betweenAmylag1and Amylag2and TMA is sufficient for reliablemodeling.

Annotation of a sequence alignment with TMAresidues contacting a-AI1 (Fig. 5) revealed two deletions

Fig. 7. (A) Northern blot analysis of Amylag2mRNA expression indifferent A. grandisdevelopmental stages. RNA was extracted from:(1) whole third-instar larvae; (2) third-instar larvae midguts; (3) pupae;(4) unfed adults; (5) 10-day-old feeding adults; (6) 10-day-old feedingadult midguts. (B) Ethidium bromide staining of rRNA loaded per lanefor each time point. Size standards are indicated in kb.

Molecular Cloning of a-Amylases and Structural Relations to Inhibitors 85

(of residues 138 and 294, TMA numbering) in theA. grandisamylases, relative to TMA, near to the interfacewith inhibitor. Models of Amylag1and Amylag2weretherefore constructed based on the a-AI1–bound struc-ture of TMA and the a-AI1 structure overlaid onto each.Features of the amylases not allowing the formation ofthe inhibitor complex (Fig. 5) were sought. The lack ofresidue 138 (TMA numbering) seems unlikely to con-tribute to the lack of A. grandisamylase inhibition bya-AI1 because that portion of the TMA–a-AI1 interfaceis relatively loosely packed. However, the deletion ofresidue 294 has dramatic consequences. When this loopis shortened in the A. grandisamylases, severe stericclashes are produced when the a-AI1 is overlaid. A fur-ther difference is apparent in this loop between thea-AI1–inhibited TMA and the noninhibited A. grandisamylases. The two glycine residues, numbers 292 and293 of TMA, are replaced by non-glycines in the A. gran-dis a-amylases. It seems possible that the flexibility ofthe glycine residues is important for inhibition since thisloop adopts significantly different conformations in theunbound and a-AI1–bound states of TMA (Strobl et al.,1998a, b; Nahoum et al., 1999). Thus, as well as theshorter length of this loop in A. grandisamylases, itsgreater rigidity compared with TMA, due to the absenceof glycine residues, may not allow the loop movementsnecessary for the formation of the a-AI1 complex.The Z. subfasciatusamylase, similarly not inhibitedby a-AI1, also has a shorter loop here with fewer glycineresidues than TMA, consistent with this explanation.Z. subfasciatusamylase is inhibited by a-AI2 (Grossi-de-Sá et al., 1997) which has two fewer residues, com-pared with a-AI1, at the part of the interface in contactwith this enzyme loop, also in agreement with this idea.

Using the published model of TMA in complexwith 0.19 inhibitor (Franco et al., 2000), interface residueswere also plotted on a sequence alignment (Fig. 5).In conjunction with A. grandisa-amylase models con-structed on the basis of the 0.19-bound TMA modelstructure, explanations for the non-inhibition of A. gran-dis a-amylases were again sought. In this case, the looparound residue 294 (TMA numbering) does not partici-pate in the 0.19 interface but the loop around residue 138does, and the deletions in A. grandisa-amylases relativeto TMA lead loss of hydrogen bonds and a reduced in-terface surface area. A further loop around residue 50, isthree residues longer in Amylag2than in TMA. Model-ing shows that these residues can be readily accommo-dated without disrupting the interface. However, TMAresidue Ala51 is replaced by the much larger tryptophanin Amylag2, a change that leads to dramatic steric clashes

with a superposed inhibitor structure. This single differ-ence seems capable of explaining the non-inhibition ofAmylag2.In the case of Amylag1, no single obvious ex-planation is forthcoming. However, several other smallersequence changes have detrimental consequences for theinterface. For example, position 49 is occupied by Leu inTMA, which makes favorable hydrophobic contacts withthe inhibitor, but by Ala in Amylag1,which is too smallto make the same contacts. Similarly, a serine replacesVal151 in TMA. In A. grandisenzymes the alterationresults in the loss of favorable hydrophobic contacts. Toquantify the overall effect of these smaller changes amodel of the hypothetical Amylag1-0.19 complex wasconstructed and its characteristics compared with themodel of TMA-0.19. The comparison shows that inter-face surface area (1034 Å2) and complementarity (2.34,defined as gap volume index; Jones and Thornton, 1996)for the TMA-0.19 model compare well with the rangestypically seen for enzyme-inhibitor structures (Jones andThornton, 1996). Typical interface surface areas are inthe range 785 6 75 Å2 while typical surface comple-mentarity values are 2.2 6 0.5 (expressed as mean 6SD). The Amylag1-0.19 complex in contrast has a some-what reduced surface area (934 Å2) but a significantlyworse complementarity value of 2.76. This implies thatthe sum of sequence differences between Amylag1andTMA are sufficient to dramatically impair the surfacecomplementarity between enzyme and inhibitor, therebyexplaining why 0.19 does not inhibit Amylag1.

4. DISCUSSION

The a-amylases play a key role in carbohydrate me-tabolism of several insects, especially those like the seedweevils that feed on starchy seeds during larval and/oradult stages and depend on their a-amylases for survival.Bean bruchid pests, for example, are highly starch de-pendent, and can produce severe pre- and postharvestdamage to stored grains (Franco et al., 2002). In contrast,A. grandis, an insect pest that inflicts damage on culti-vated cotton, does not feed on cotton seeds and is not apest of stored grain. A. grandisonly feeds on cotton re-productive structures (Monnerat et al., 2000), especiallyanthers and ovules present in the floral bud and fruits(Fig. 1). In these organs were found numerous plastidsresponsible for the starch accumulation in floral buds,specifically in male and female gametophytes. As seen inFig. 2, starch accumulation in pollen grains and ovules in-creases at the final stage of anther development, near toanthesis (buds with diameters in the range of 0.6–1.0 cm).

86 Oliveira-Neto et al.

As these structures are preferred by boll weevils for feed-ing and oviposition, this can explain the synthesis ofdigestive a-amylases in this pest.

Midgut extract from A. grandisreared on natural orartificial diet was enzymatic assayed to determine any ef-fect of diet on amylase expression (Fig. 3). The inducedexpression of amylase in response to starch has been re-ported for the yellow mealworm (Strobl et al., 1998a, b),rice weevil and red flour beetle (Feng et al., 1996). Incontrast to the variations seen in Z. subfasciatusfed ondifferent diets (Silva et al., 2001), no difference wasobserved with A. grandis.

The presence of a-amylases in A. grandisalimen-tary tract and the reliance upon these enzymes for feed-ing may represent a new strategy to control this pest withproteinaceous a-amylase inhibitors. Two cDNAs (Amy-lag1 and Amylag2) encoding putative a-amylases thatshare 58% amino acid identity were therefore identi-fied and sequenced. Amylag1and Amylag2share highidentity with other insect a-amylases (Fig. 5) and se-quence comparisons show that both seem to possess allthe characteristics of active amylase enzymes as severalo-glycosylation sites (Fig. 4).

As for A. grandis, most strains of Drosophilahavetwo closely linked genes encoding two a-amylaseisozymes (Boer and Hickey, 1986; Gemmil et al., 1986).Some crop pests such as Sitophilus oryzae(Baker et al.,1990), Sitophilus zeamays(Baker and Halliday, 1989) andD. vigifera(Titarenko and Chrispeels, 2000) also are pres-ent at the least two a-amylase isozymes. In the case ofD. virgifera, although two genes were cloned, only oneis expressed as an active a-amylase (Titarenko andChrispeels, 2000). As yet, we do not known if both clonedA. grandis cDNAs express active enzymes. This willrequire partial amino acid sequencing of purified A. grandisamylases or expression of the recombinant proteins.

a-Amylase enzymes are present in the insect ali-mentary tract for the hydrolysis of starch. In yellowmealworm T. molitor, the a-amylases are synthesized inanterior midgut cells and packed in the Golgi area intosecretory vesicles that undergo fusion, as they migrate tothe cell apex and to gut lumen (Cristofoletti et al., 2001).On the boll weevil 10-day-old adults, a faint sign ofAmylag2transcript was found when whole tissue adultwas analyzed. But a stronger sign was observed in themidgut tissue showing a higher expression of a-amylasecDNA in the midgut (Fig. 7). It is probable that A. gran-dis adult regulates gut a-amylase expression in the sameway as most other insects in response to protein in theirfood (Félix et al., 1991; Borovsky et al., 1996).

With the aim to discover novel defense factors to-ward the cotton boll weevil A. grandis, several plant

a-amylase inhibitors from different classes were tested(Fig. 8). Inhibitors from cereal family (0.19 and 0.53) andlectin class (a-AI1 and a-AI2) were unable to stronglyinhibit the boll weevil a-amylases. The chosen pH andconcentration were based on previously results, where thosefour inhibitors were able to inhibit another a-amylases atthis same pH at a much lower concentration (2 mg/ml)(Grossi-de-Sá et al., 1997; Franco et al., 2000). Never-theless, the thionin-class inhibitor SaAI5 (Bloch andRichardson, 1991) and the bifunctional inhibitor-like BIIIinhibitor (Iulek et al., 2000) demonstrated inhibitory ac-tivity against these same a-amylases at the sameconcentration used (300 mg/ml). These two inhibitorsprobably have different modes of action. While SaAI5probably chelates catalytically essential calcium ions(Castro et al., 2002), BIII probably binds directly in theenzymatic active site as observed crystallographically forrelated inhibitors (Strobl et al., 1998a, b).

The availability of inhibition, sequence and struc-tural data for a variety of insect amylases has stimu-lated studies that attempt to explain, in structural terms,observed patterns of amylase inhibition (Le Berre-Antonet al., 2000; Silva et al., 2000; Franco et al., 2002). Insome cases clear explanations have not been forthcom-ing (Silva et al., 2000), illustrating perhaps the variousforces that contribute to amylase-inhibitor interaction. Inother cases, convincing explanations have been obtained(Franco et al., 2000). As recently reviewed (Franco et al.,2002), steric factors, electrostatic interactions, hydrogenbonding and the particular conformational properties ofproline are all of proven importance in at least one caseof amylase-inhibitor interaction. The explanation foundfor the noninhibition of A. grandisamylases by a-AI1(Fig. 5) expands still further the range of factors likely tobe involved in conferring specificity between amylaseand inhibitor. A striking correlation between the lengthand composition of a key enzyme loop located at theinterface with a-AI1 and the ability or not of a-AI1 toaffect enzyme activity was found. In comparison withTMA, inhibited by a-AI1, noninhibited insect amylasessuch as the Z. subfasciatusand A. grandisenzymes haveshorter loops that, containing fewer glycines, are lessflexible. Flexibility is likely to be important because thisloop adopts different conformations in unbound and a-AI1–bound TMA (Strobl et al., 1998a, b; Nahoumet al., 1999). The corresponding loop in mammaliana-amylases is longer (Fig. 5) and changes conformationto an even greater extent on binding to a-AI1 (Larsonet al., 1994; Bompard-Gilles et al., 1996). It is thereforeintriguing to note that in comparison with PPA, inhibitedby a-AI1, in the A. obtectusenzyme, not inhibited bya-AI1; this loop is one residue shorter and contains one

Molecular Cloning of a-Amylases and Structural Relations to Inhibitors 87

fewer glycine residue. Thus the hypothesis that lengthand flexibility are important characteristics to allow a-AI1 inhibition gains further support.

Several articles have now reported protectionagainst pests exhibited by transgenic plants expressinga-amylase inhibitors. For example, in transgenic peaplants, a-AI1 levels in the range of 0.8%–1.0% total pro-tein in seeds produce complete protection against thethree Old World bruchids: the pea weevil Bruchus piso-rum, the cowpea weevil and the adzuki bean weevilC. chinensis(Ishimoto et al., 1996), whereas a-AI2 isalso effective against pea weevil (Morton et al., 2000).Thus it seems reasonable to hope that transgenic cottonexpressing amylase inhibitors may soon join Bt-express-ing transgenic cotton (Estruch et al., 1997) in the fightagainst the cotton boll weevil.

ACKNOWLEDGMENTS

We thanks to Dr. Carlos Bloch, Jr. for kindly pro-viding the SaAI5 and to Dr. Jorge Iulek for kindly pro-viding the rye inhibitor BIII. This work was supportedby Capes, CNPq, FACUAL and Common Fund forCommodities/International Cotton Advisory Committee.

REFERENCES

Baker, J. E., and Halliday, W. R. (1989). J. Stored Prod. Res.25: 17–24.Baker, J. E., Halliday, W. R., and Lum, P. T. M. (1990). J. Stored

Prod. Res26: 7–10.Barton, G. J. (1993). Protein Eng.6: 37–40.Bernfeld, P. (1955). Amylase, a and b. Meth. Enzymol.1: 149–158.Bloch, C., Jr., and Richardson, M. (1991). FEBS Lett.279: 101–104.Boer, P. H., and Hickey, D. A. (1986). Nucleic Acids Res.14: 8309–8311.Bompard-Gilles, C., Rousseau, P., Rouge, P., and Payan, F. (1996).

Structure4: 1441–1452.Borovsky, D., Jansen, I., Vanden Broeck, J., Huybrechts, R., Verhaert, P.,

De Bondt, L., et al. (1996). Eur. J. Biochem.237: 279–287.Bottrell, D. G. (1983). Managment Agric. Ecosyst. Environ.20: 240–274.Boulter, D. (1993). Phytochemistry34(6): 1453–1466.Bradford, M. M. (1976). Anal. Biochem.72: 248–254.Carlini, C. R., and Grossi-de-Sá, M. F. (2002). Toxicon(submitted).Castro, C. S. P., Souza, J. R., and Bloch, C., Jr. (2002). Protein Pept.

Lett. 9(1): 45–52.Cristofoletti, P. T., Ribeiro, A. F., and Terra, W. R. (2001). J. Insect

Physiol.47: 143–155.Estruch, J. J., Carozzi, N. B., Desai, N., Duck, N. B., Warren, G. W.,

and Koziel, M. G. (1997). Nat. Biotechnol.15: 137–141.Félix, C. R., Betschart, B., Billingsley, P. F., and Freyvogel, T. A.

(1991). Insect Biochem.21: 197–203.Feng, G. H., Richardson, M., Chen, M. S., Kramer, K. J., Morgan,

T. D., and Reeck, G. R. (1996). Insect Biochem. Mol. Biol.26(5): 419–426.

Fourney, R. M., Miyakoshi, J., Day, R. S., III, and Paterson, M. C.(1988). Focus10: 5–7.

Franco, O. L., Rigden, D. J., Melo, F. R., Bloch, C., Jr., Silva, C. P., andGrossi-de-Sá, M. F. (2000). Eur. J. Biochem.267: 2166–2173.

Franco, O. L., Rigden, D. J., Melo, F. R., and Grossi-de-Sá, M. F.(2002). Eur. J. Biochem.269: 397–412.

Gatehouse, A. M. R., and Gatehouse, J. A. (1998). Pestic. Sci.52: 165–175.

Gemmil, R. M., Fenton, I. J., Jepson, I., and Pavey, D. J. (1986). Nu-cleic Acids Res.14: 5337–5353.

Grossi-de-Sá, M. F., Mirkov, T. E., Ishimoto, M., Colucci, G.,Bateman, K. S., and Chrispeels, M. J. (1997). Planta203:295–303.

Ishimoto, M., Sato, T., Chrispeels, M. J., and Kitamura, K. (1996). En-tomol. Exp. Appl.79: 309–315.

Iulek, J., Franco, O. L., Silva, M., Slivinski, C. T., Bloch, C., Jr.,Rigden, D. J., et al.(2000). Int. J. Biochem. Cell Biol.32: 1195–1204.

Jenkins, J. N., McCarty, J. C., and Wofford, T. (1995). ProceedingsBeltwide Cotton Conference, National Cotton Council.

Jones, S., and Thornton, J. M. (1996). Proc. Natl. Acad. Sci. USA93:13–20.

Jones, T. A., Zou, J. Y., Cowan, S. W., and Kjeldgaard, M. (1991).Acta Crystallogr. A47: 110–119.

Kleywegt, G. J., Zou, J. Y., Kjeldgaard, M., and Jones, T. A. (2001).In: International Tables for Crystallography, Crystallography ofBiological Macromolecules(Rossmann, M.G. and Arnold, E.,eds), chpt. 17.1, vol. F, pp. 353–367, Dordrecht: Kluwer Acade-mic Publishers, The Netherlands.

Laemmli, U. K. (1970). Nature227(259): 680–685.Larson, S. B., Greenwood, A., Cascio, D., Day, J., and McPherson, A.

(1994). J. Mol. Biol.235: 1560–1584.Laskowski, R., MacArthur, M., Moss, D., and Thornton, J. (1993).

J. Appl. Crystallogr.26: 283–290.Le Berre-Anton, V., Nahoum, V., Payan, F., and Rouge, P. (2000).

Plant Physiol. Biochem.38: 657–665.MacGregor, E. A., Janecek, S., and Svensson, B. (2001). Biochim. Bio-

phys. Acta1546:1–20.Monnerat, R. G., Dias, S. C., Oliveira-Neto, O. B., Nobre, S. D., Silva-

Werneck, J. O., and Grossi-de-Sá, M.F. (2000). Comunicado Téc-nico Embrapa Recursos Genéticos e Biotecnologia, ISSN 0102-0099, 4pp, Brasília, Brasil.

Morton, R. L., Schroeder, H. E., Bateman, K. S., Chrispeels, M. J.,Armstrong, E., and Higgins, T. J. V. (2000). Proc. Natl. Acad.Sci. USA97(8): 3820–3825.

Nahoum, V., Farisei, F., Le Berre-Anton, V., Egloff, M. P., Rouge, P.,Poerio, E., et al. (1999). Acta Crystallogr. D55: 360–362.

Nielsen, H., Engelbrecht, J., Brunak, S., and Heijne, V. G. (1997). Pro-tein Eng.10: 1–6.

Qian, M., Haser, R., Buisson, G., Duée, E., and Payan, F. (1994). Bio-chemistry33: 6284–6294.

Sali, A., and Blundell, T. L. (1993). J. Mol. Biol.234: 779–815.Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989). Molecular

cloning: A laboratory manual. 2nd ed., Cold Spring Harbor Lab-oratory Press, NY.

Shindyalov, I. N., and Bourne, P. E. (1998). Protein Eng.11: 739–747.Silva, M. C., Grossi-de-Sá, M. F., Chrispeels, M. J., Togawa, R. C.,

and Neshich, G. (2000). Protein Eng.13: 167–177.Silva, C. P., Terra, W. R., Xavier-Filho, J., Grossi-de-Sá, M. F.,

Isejima, E. M., Damatta, R. A., et al. (2001). Insect Biochem.Mol. Biol. 31: 41–50.

Strobl, S., Maskos, K., Betz, M., Wiegand, G., Huber, R., Gomis-Ruth,F. X., et al. (1998a). J. Mol. Biol.278: 617–628.

Strobl, S., Maskos, K., Wiegand, G., Huber, R., Gomis-Ruth, F. X., andGlockshuber, R. (1998b). Structure6: 911–921.

Thomas, J. C., Adams, D. G., Keppenne, V. D., Wasmann, C. C., Brown,J. K., Kanost, M. R., et al. (1995). Plant Cell Rep.14: 758–762.

Thompson, J. D., Higgins, D. G., and Gibson, T. J. (1994). NucleicAcids Res.22: 4673–4680.

Titarenko, E., and Chrispeels, M. J. (2000). Insect Biochem. Mol. Biol.30: 979–990.