Available via license: CC BY 4.0
Content may be subject to copyright.
THE JOURNAL OF BIOLOGICAL CHEMISTRY
Vol. 265, No. 34, Issue of December 5, pp. 20923-20930, 1990
cc) 1990 by The American Society for Biochemistry and Molecular Biology, Inc.
Printed in U. S. A.
Specificity-determining Regions of a Lepidopteran-specific
Insecticidal Protein Produced by Bacillus thuringiensis*
(Received for publication, June 15, 1990)
H. Ernest Schnepf, Kathleen Tomczak, Jose Paz Ortega, and H. R. WhiteleyS
From the Department
of
Microbiology, University
of
Washington, Seattle, Washington 98195
The lepidopteran-specific, insecticidal crystal pro-
teins of Bacillus thuringiensis vary in toxicity to dif-
ferent species of lepidopteran larvae. We report studies
of CryIA(a) and CryIA(c), two related proteins that
have different degrees of toxicity to Heliothis vires-
tens yet very similar degrees of toxicity to Manduca
sextu. The amino acid differences between these pro-
teins are located primarily between residues 280 and
722. We have constructed a series of chimeric proteins
and determined their toxicities to both insects. The
most significant findings arise from the replacement
of three segments of the cryIA(c) gene with homologous
portions of the cryIA(a) gene: codons 332-428, 429-
447, and 448-722. Each of these segments contributed
substantially and largely additively toward efficacy
for H. virescens. However, replacement of the 429-
447 segment of cryIA(c) gene with the cryIA(a) se-
quence resulted in a 27-50-fold reduction in toxicity
toward M. sexta whereas the reduction in toxicity to
H, virescens was only 3-4-fold. Subdivision of the
429-447 segment and replacements involving residues
within this segment reduced toxicity to M. sexta by 5-
to more than 2000-fold whereas toxicity to H. vires-
tens was only reduced 3-lo-fold. These observations
indicate that: 1) different but overlapping regions of
the cryIA(c) gene determine specificity to each of the
two test insects; 2) some of the examined gene segments
interact in determining specificity; and 3) different
sequences in the cryIA(a) and cryIA(c) genes are re-
quired for maximal toxicity to M. sexta.
Bacillus thuringiensis produces proteins that cause a lethal
intoxication of specific types of insects. Each B. thuringiensis
strain usually carries several toxin genes that determine the
range of insects affected. Of the toxin sequences published to
date,
all but one show some significant sequence similarity to
other toxin genes, indicating that these proteins are encoded
by a gene family (1). Although most of the isolated B. thurin-
giensis strains affect only lepidoptera, some strains are spe-
cific only to diptera or to coleoptera, and a few are lethal to
both lepidoptera and diptera. Many of the
B. thuringiensis
strains that affect lepidoptera have quite different efficacies
on different insects (2). Analysis of the insecticidal specificity
of the
products of several cloned genes or of proteins from
* This research was supported in part by Public Health Service
Grant GM-20784 from the National Institute of General Medical
Sciences and by grants from the Washington Technology Center.
The costs of publication of this article were defrayed in part by the
payment of page charges. This article must therefore be hereby
marked “aduertisement” in accordance with 18 U.S.C. Section 1734
solely to indicate this fact.
$ Recipient of Research Career Award K6-GM-442 from the Na-
tional Institute of General Medical Sciences.
strains of B. thuringiensis thought to carry a single toxin gene
has shown that the individual gene products have differential
activities toward specific insects (1, 3).
It has been known for some time that the genes encode
protoxins that are cleaved in the insect gut to release a toxic
fragment. Few details are known about the mechanisms in-
volved in the intoxication of insect gut cells although it has
been shown recently that the toxic proteins recognize specific,
high affinity receptors on the midgut brush border (4,5). The
specifically bound toxin then directly or indirectly induces a
breakdown of the permeability barrier of the cell, leading to
cell death (6). Although specific receptor binding is one de-
terminant of species targeting, it is not known whether addi-
tional steps leading to intoxication are also species specific or
whether different toxins may vary with regard to the region
of the protein that determines specificity toward a given
insect.
There are three well characterized genes in the CryIA class
(l), i.e. cryIA(a), (b), and (c), which were previously called the
“4.5-, 5.3-, and 6.6-kilobase genes” (7). These genes code for
protoxins of 1156-1178 residues and are lethal to lepidopteran
larvae. The minimum toxin-encoding fragment, produced by
proteolysis in vitro and presumably in the insect gut, resides
between residues 29 and lysine 623 for C&A(c) (8) and
presumably the homologous residues, 29-lysine 621, of
CryIA(a). Deletion analyses of the three cvIA genes to deter-
mine the gene segment encoding toxicity agree with the pro-
teolysis data (9-11). As noted previously (12, 13), the differ-
ences among the CryIA(a), CryIA(b), and CryIA(c) proteins
are located mostly between codons 280 and 722, and Ge et al.
(14) have identified a segment within this variable region
which is required for the toxicity of CryIA(a) to Bombyx mori
(silk worm). Our current studies concern the analysis of the
lepidopteran specificities of CryIA(a) and CryIA(c) toward
Manduca sexta (tobacco hornworm) and Heliothis virescens
(tobacco budworm). HGfte et al. (3) have shown that these
two proteins have roughly equivalent activity on M. sexta
whereas CryIA(c) is about 50 times more toxic to H. virescens
than is CryIA(a).
One approach to determining the different activities of
related gene products is the construction of hybrid, or chi-
merit, genes followed by assays of the chimeric gene products.
This type of analysis relies upon the observation that related
genes have the same overall three-dimensional structure (e.g.
Ref. 15) so that linear combinations of related gene products
often lead to relatively functional, stable proteins. This paper
reports the toxicity to M. senta and H. virescens of proteins
synthesized by Escherichia coli after transformation with plas-
mids bearing the cryZA(a), cryIA(b), and cryZA(c) genes or
chimeric recombinants between them. Our experiments show
that each of three segments between codons 332 and 722
contributes to the specificity of CryIA(c) toward H. virescens
whereas toxicity to M. sexta depends on a small region located
20923
by guest, on July 10, 2011www.jbc.orgDownloaded from
20924 Crystal Protein Specificity-determining Regions
between codons 429 and 447. Overall, the data show that
different sequences in a given
B. thuringiensis
insecticidal
protein determine toxicity to different insects and also that
toxicity
to a given insect may involve different regions in
closely related proteins.
MATERIALS AND METHODS
Bacterial Strains and Plasmids-E. coli strains MC1000 (16),
RZ1032 (17), and B. thuringiensis HD-1-Dipel (18) have been de-
scribed previously. Plasmids pUC118 and pUC119 (19) were used to
manipulate portions of the toxin genes. A derivative of pUC119,
pUCllSA, was created by digestion with EcoRI and SmaI, filling in
of the EcoRI site. and re-ligation. M13K07 (19) was used to produce
sequencing templates frompUC118, pUC119, and derivatives. Plas-
mids pMAR4, pJWK29, and pJWK20 (7) as well as pHES41 (9) have
been described.
Insect Sources-The tobacco hornworm M. sexta was obtained
from Drs. L. Riddiford and J. Truman, Dept. of Zoology, University
of Washington. The tobacco budworm H. virescens was obtained from
the USDA-ARS, Stoneville, MS.
Plasmid Constructions-Commonly used procedures in molecular
biology were performed essentially as described (20). The plasmid
pCAa contains the NdeI-NdeI fragment having the cyZA(a) gene
from pHES41 in pUC119A (see Fig. 1, A and B). Plasmid pCAb,
containing the crylA(b) gene from pJWK29, has the filled-in NdeI
fragment containing the toxin gene cloned into the filled-in XbaI site
of pUC119A (Fig. 1, A and B). Plasmid pCAc, containing the cryZA(c)
gene from pJWK20, has the filled-in I?deI fragment containing the
toxin gene cloned into the filled-in XbaI site of nUC119A (Fig. 1. A
and B). In all three cases the crystal protein genes are oriented in
the same direction as the luc promoter. Plasmids pCS3 and pCS4
contain the indicated EcoRI-Sac1 fragment of the cryZA(a) and
cryZA(c) genes, respectively, in the same sites of pUC118 (Fig. 1C);
pCS1 and pCS2 were formed by exchange of the respective ClaI-Sac1
subfragments (Fig. IC). Plasmids pCC1, pCC2, and pCC3 were
formed by introducing the EcoRI-Sac1 fragments from pCS1, pCS2,
and pCS3, respectively, into pCAc that had been partially methylated
with EcoRI methylase, digested with EcoRI, digested with SacI, and
the appropriate fragment-purified. Construct& of plasmids pCC5,
pCC6. pCC7. and pCC4 proceeded bv replacement of the SacI-KrmI
fragments oi pCCi, p&2, pCC3, and &A3, respectively, with the
SacI-Kpnl fragment of cryZA(a) (Fig. 1D).
Oligonucleotide-directed mutagenesis (17) was used to introduce
PvuII sites (see Fig. 6A) into the &I-Sac1 fragment of cryZA(c)
contained in pCS4 (Fig. 1C). The sites were introduced at codons
439-441 of cryZA(c) corresponding to the same codons in cryZA(a)
(pCS5) or at codons 442-443 of cryZA(c) corresponding to a gap
introduced into the cryZA(a) sequence to optimize the alignment
between the two genes (pCSG; see Fig. 6A). The changes were verified
by sequencing the entire EcoRI-Sac1 fragments. The CZaI-PvuII frag-
ments of pCS3 (from cryZA(a)), pCS5, and pCS6 were purified and
ligated into pCS4 in the following combinations: pCS6 &I-PvuII
plus pCS3 PvuII-Sac1 to form pCS7; pCS5 &I-PvuII plus pCS3
PvuII-Sac1 to form pCS8; pCS3 &I-PvuII plus pCS5 P&I-Sac1 to
form pCS9; and pCS3 ClaI-PuuII plus pCS6 PvuII-Sac1 to form
pCS16. The EcoRi-Sac1 fragments from pCS5, pCS6, pCS7, pCS8,
uCS9. and pCSl0 were then inserted into pCA3 to form pCC11,
&Clb, pCc12, pCC14, pCC13, and pCCl.$ respectively, by the
methods shown in Fig. 1, B-D, to make pCC1, pCC2, and pCC3.
Protein Purification and Concentration Estimation-Parasporal
crystals of B. thuringiensis HD-1-Dipel were purified as described
previously (21). Inclusions containing crystal protein were purified
from E. coli by repeated sonication and washing in 0.01
M
Tris, 0.01
M
EDTA, pH 7.9, 1 mg/ml lysozyme. A 0.5
M
NaCl wash removed
the rest of the adsorbed lysozyme. Samples were stored in H20 at
4 “C after two to three additional washes in HZO.
Soluble protoxin was prepared from inclusions or crystals by ex-
traction with 0.1 M Na&OII, 0.025 M P-mercaptoethanol, pH 9.5. The
resulting supernatant was generally 90% or-more of approximately
1.73.kDa protoxin, with most of the residue consisting of antigenically
related fragments of greater than 70 kDa. Protein concentrations
were determined both by using the method of Bradford (22) and by
scanning Coomassie Blue-stained sodium dodecyl sulfate-polyacryl-
amide gels.
Bioassay and Data Reduction-Bioassays were performed in 24-
well culture plates with 2-cm2 wells, using the diet of Bell et al. (23).
Dilutions of purified protoxin were spread on the surface of the diet
and allowed to dry before placing one larva (less than 1 day old) in
each well. Eight to twelve larvae were tested per dilution, and after 7
days, immobile, unresponsive larvae (regardless of size) were scored
as dead.
Mean lethal concentrations were estimated for each test prepara-
tion using the probit model (24), which estimates a parameter equal
to the log LCsO and confidence limits that are linearly spaced with
the log of the dose. Each group of bioassays included a control dose-
response assay using purified solubilized crystals from B. thuringien-
sis strain HD-l-Dipel. All constructions in the set were assayed at
the same time, using the same batch of diet and were normalized to
the control by subtraction of the control log LC& from the sample
log L&o according to Finney (24). Each recombinant protoxin was
tested three to eight times. Our results showed that the confidence
limits determined by probit analysis (24) for the LCsO values of
individual experiments clearly underestimated the variation observed
between repeated experiments; therefore, the LCso values reported
are the averages of the log LCW values for several determinations
relative to the HD-I-Dipel protoxin standard. On average, the one
standard deviation confidence limits determined in this way equaled
approximately the two standard deviation confidence limits deter-
mined as recommended by Finney (24) and represent more conserv-
ative estimates of the accuracy of our results. Since it is the log of
the dose, rather than the dose, which is used to estimate the L&o
and the associated confidence limits, the one standard devia-
tion confidence interval for the L& (rig/cm’) is presented as
lo”“9 ~‘%a - s D. 1~ %d to lo”“9 ~CSLI + s D. 1% %). In the case of CryI,‘,(
which was only assayed once, the one standard deviation confidence
limits were derived directly from the LCfiO calculation (24).
When the differences between the mean values were tested for
statistical significance, several instances were observed in which L&o
values differing by less than P-fold were significant at the p < 0.05
level. Since we do not believe that observed differences of this
magnitude can be considered reliable without numerous replications,
such differences are referred to as “of marginal significance” whereas
LC,, values differing by more than 2-fold and having significance at
p < 0.05 by
t
test are referred to as “significant.”
RESULTS
Constructions, Protein Purification, and Assays-To
inves-
tigate the contribution of individual portions of the
crylA(a)
and
cryIA(c)
genes to insecticidal specificity, each was sub-
cloned so that it could be expressed efficiently in
E. coli
(Fig.
IA). The
cryIA(b)
gene was included in these studies for
specific comparisons as described below. Chimeric genes were
then constructed as described in Fig. 1 and under “Materials
and Methods.” The positions of some of the amino acid
differences in CryIA(a), CryIA(b), and the chimeric proteins
relative to CryIA(c) as well as the sources of DNA for the
chimeric genes are diagrammed in Fig. 2. The chimeric con-
structions shown in Fig. 2 test the effects of differences
between the
cryZA(a)
and
cryIA(c)
genes in the following
segments: codons 1-331 plus 723-1156 and in codons 332-
428, 429-447, and 448-722. Several constructions, including
CC3 and CC8, allowed a test of the effect of sequences in
cryIA(b)
which differ from
cryIA(a)
and
cryZA(c).
For each chimeric gene, inclusions synthesized in
E. coli
were purified, and the protoxin was solubilized and tested for
toxicity to M. sexta and H. virescens.
No significant differ-
ences were found in the amounts of inclusions produced from
the
chimeric genes or in the solubility of the inclusions. In all
assays, the toxicity results were normalized to the results
obtained with similarly treated crystals from strain
HD-l-
Dipel. The latter strain is a derivative of strain HD-1, which
is used frequently as a standard in assays of
B. thuringiensis
insecticidal activity. Crystals from strain HD-1 contain the
CryIA(a), (b), and (c) proteins; crystals from strain HD-l-
Dipel contain
only the CryIA(a) and CryIA(c) proteins (7).
Strains HD-1-Dipel and
HD-1 also express the cryZZA
gene;
however, its product has a much higher LCk of about 150-
by guest, on July 10, 2011www.jbc.orgDownloaded from
Crystal Protein Specificity-determining Regions
20925
- cryIn
FIG. 1. Restriction maps and diagrams showing the con-
struction of chimeric plasmids pCCl-pCC7. The crylA(a) se-
quences are shown with a slightly thicker line than the vector; the
cryIA(b) sequences, with a dotted line; and the cryIA(c) sequences,
with a
thick line.
N, NdeI; RI, R2, EcoRI sites 1 and 2; P, PuuII; K,
KpnI; S, SacI. A, restriction maps of the crylA(a), (b), and (c) genes
from B. thuringiensis. The arrow indicates the direction of transcrip-
tion. B, representation of subclone of the NdeI fragments of each of
these genes into pUC119A. C, subclones and EcoRI-&I-Sac1 frag-
ment exchanges between c&A(a) and crylA(c). D and E, construc-
tions from plasmids indicated in B and C above. See “Materials and
Methods” for details.
I II
x Y
2
Source of DNA: cryI/\ - crylA(b) m cry,A(c) i
FIG. 2. The positions of amino acid differences in the
CryIA(a), CryIA(b) proteins and in chimeric proteins CCl-
CC9 relative to CryIA(c). Differences in amino acids are indicated
with a vertical line. The positions of the CryIA(a) segments X, Y, and
Z are shown below CC7; these segments were introduced into CryIA(c)
to make chimeric proteins CClLCC7.
250 pg/cm* on
the two test
insects’
(26) and should not
contribute appreciably to toxicity. The results are presented
as LC& values in rig/cm* and are plotted on a log scale with
1 H. E. Schnepf, unpublished data.
the associated one standard deviation confidence interval,
since toxicity results are linear with the log of the dose.
Toxicity of
CryZA(a),
(b), and (c) to H. virescens and M.
sexta-Table I shows that all three protoxins have similar
toxicity to M. sexta whereas the toxicities to
H. virescens
are
quite different, with CryIA(a) being much less toxic than
CryIA(b), which is less toxic than CryIA(c). Because of vari-
ations in the susceptibilities of different lots of insects, these
average values are 2-4-fold higher than values reported pre-
viously (3) for the individual proteins tested with
M. sexta
and 44&fold higher than results observed previously (3) for
H. virescens.
However, the relative potencies of the three
proteins to the two insects agree with the prior values. Assays
performed with different batches of insects (e.g. Table II in a
later section) yielded values that agreed with published values
(3).
Effect of Differences
in the cryIA(b) Gene Versus the
cryIA(a) and crylA(c)
Genes-Comparison of the differences
in the amino acid sequences diagrammed in Fig. 2 indicates
that the
cryZA(b)
gene can be viewed essentially as a recom-
binant between the
cvIA(a)
and
cryIA(c)
genes at approxi-
mately codon 460 with a deletion of 26 codons at codon 793
and a divergent sequence beginning at codon 1065 which
includes a 4-codon insertion (25). One of the chimeras be-
tween CryIA(a) and CryIA(c), CC3 (Fig. 2), has a sequence
very similar to CryIA(b) prior to codon 722. A comparison of
the toxicity of CC3 and CryIA(b) on both
H. virescens
and
M.
sexta
shows no significant differences, demonstrating that the
sequence differences distal to codon 722 have no detectable
effect on toxicity toward these insects.
The sequences of CryIA(a) and CryIA(b) between codons
448 and 722 differ by 3-4 residues, depending on the source
of the
cryZA(b)
gene. Chimeric protein CC8 is CryIA(a) with
the CryIA(b) sequence for residues 448-722. A comparison of
toxicities of CryIA(a) and CC8 indicates little difference to-
ward
M. serta
and a difference in toxicity to
H. virescens
TABLE
I
Toxicities
of
chimeric proteins constructed by exchanging segments
of
C&A(a) and C&A(c) between codons 332 and 772
H.
&escens
M. sexta
Protein”
L&o* 1 S.D. C.I.’ LCro
1 S.D. C.I.
&cm’
B.t. HD-1-Dipel 18 11-28 17 9-29
‘WA(c)
16 12-21 20 12-31
CwWW 43 27-68 20 15-26
CrylA(a)
472 226-988 23 20-27
CC8
240 118488 31 18-53
cc9 131 87-196 37 26-52
Protein
Segmentd
x
Y
z
cc1
cc2
cc3
cc4
cc5
CC6
cc7
+
+
+
+ +
+ +
+ +
+ + +
LGO
1 S.D. C.I.
M. sexta
Go
1 S.D. (2.1.
ngfcm’
49 43-55
59 42-82
34 16-72
147 115-187
842 704-1007
106 56-202
605
573-638
56 31-101
537 146-1968
23 12-44
23
10-53
36 33-39
83
63-109
19 16-23
’ Solubilized crystals from B. thuringiensis (B.
t.)
subspecies kur-
stuki HD-1-Dipel or solubilized inclusions from cloned genes ex-
pressed in E. coli. Compositions of chimeric proteins Ccl-CC9 are
shown in Fig. 2.
b rig/cm* of diet.
’ One standard deviation confidence interval; see “Materials and
Methods ” for details.
dCrylA(a) segment exchanged into CrylA(c): X = residues 332-
428; Y = residues 429-447; Z = residues 448-772.
by guest, on July 10, 2011www.jbc.orgDownloaded from
20926 Crystal Protein Specificity-determining Regions
which is not significant (1.9-fold, p 5 0.12; Table I). Three
additional constructions (not shown in Fig. 2) were made
which were identical to CC5, CC6, and CC7, respectively,
except that the codon 448-722 segment was from the
cryZA(b)
gene rather than
cryZA(a).
In each case, the potency of the
CryIA(b)-containing protein was very similar to that of the
CryIA(a)-containing protein for both insects (data not
shown). Therefore, the few differences between CryIA(a) and
CryIA(b) between residues 448 and 722 do not account for
their different toxicities toward
H. virescens.
Comparison
of CryZA(a) and CryZA(c) Prior to Codon 331
and Distal to Codon
722-There are 10 differences in sequence
between CryIA(a) and CryIA(c) prior to residue 331 and 2
differences distal to codon 722 (Fig. 2). Two pairs of proteins,
CC4
uersus
CC9 and CC7
versus
CryIA(a), allow comparison
of the effects on toxicity of these regions. There were
no
significant differences in toxicity to either insect for either
pair of proteins (Table I), indicating that in the contexts used,
these two segments do not contribute to insecticidal specific-
ity.
Combinatorial Analysis
of
CryZA(a) Substitutions into
CryZA(c) from Codons
332-722-The majority of the differ-
ences between CryIA(a) and CryIA(c) are located between
codons 332 and 722. As stated above, chimeric protein CC7,
in which amino acids 332-722 of CryIA(c) are replaced by
those in CryIA(a), had very similar toxocity to CryIA(a) for
both
H. virescens
and
M. sexta.
To analyze which portions of
this region determine the greater toxicity to
H. virescens,
the
region was subdivided into three parts: codons 332-428 (seg-
ment X), 429-447 (segment Y), and 448-722 (segment Z).
Each of the six combinations (Ccl-CC6, Fig. 2) of these
segments from CryIA(a) was used to replace the correspond-
ing segment of CryIA(c). Figs. 3 and 4 show a graphic analysis
of the effect of the substitutions on the L&o to each insect.
Each of the three panels in Figs. 3 and 4 shows the LCSO
values of two of the six different pathways by which CryIA(c)
can be converted to CC7 by substitution of one of the three
segments followed by a second and third segment.
With
H. uirescens,
substitutions of one segment of CryIA(a)
into CryIA(c) resulted in a drop of 2.1-3.7-fold in toxicity,
regardless of the segment (CCl2, CC2, CC3 in Fig. 3,
A-C).
However, CC3 (substituting segment C; p = 0.06) did not
meet our criteria for a significant difference in toxicity from
CryIA(c) (Fig. 3C). Double substitutions involving segments
X and Y (CC4) or Y and Z (CC6) resulted in a further 1.8-
4.4-fold reduction in toxicity (Fig. 3,
A-C),
with CC6 not
differing significantly from CC2
(p = 0.14, Fig. 3B). The
substitution involving segments X and Z (CC5) was the least
toxic protein to
H. uirescens,
being marginally less toxic than
CC7 (Fig. 3,
A
and C, Table I). With the exception of conver-
sion pathways that initiated with segments X and Z (CC5,
Fig. 3,
A
and C), there was a general linear trend to the data
indicating that each of the segments tested contributed to
specificity toward
H. virescens.
Cooperativity between the CryZA(c) 332-428 (x) and 429-
447 (Y) Segments-When the same set of chimeric proteins
was tested with
M. sexta,
quite different results were obtained.
Whereas segment Z caused an insignificant change in toxicity
(Fig. 4C, CC3), segment X caused a 3-fold drop in toxicity
(Fig. 4A, Ccl), and segment Y caused a 27-fold drop in
toxicity (Fig. 4B, CC2) when compared with CryIA(c). For
double substitutions, segments X and Y restored toxicity fully
relative to the single substitutions (Fig. 4,
A
and B, CC4),
segments X and Z caused a statistically significant
(p = 0.002)
but marginal (2-fold) reduction from parent protein toxicity
(Fig. 4,
A
and C, CC5), and segments Y and Z retained a 4.3-
A.
40
(rig/cm*)
0 1 2
3
No. crylA(a) Segments
B.
LC50
(rig/cm*)
0 1 2
3
C.
No. crylA(a) Segments
LC50
(rig/cm*)
0
1 2 3
No. crylA(a) Segments
FIG.
3. Effect of substitution order on toxicity to H. vires-
tens. The figures show the LC&, values of chimeric proteins as a
function of the number of CryIA(a) segments introduced into
CryIA(c). The first point of each panel is a different segment substi-
tution. Vertical bars show one standard deviation confidence inter-
vals.
fold reduction in toxicity (Fig. 4, B and C, CC6) compared
with the parent proteins.
Thus, with the exception of CC5, which was 2-fold less
toxic than the parent proteins, separation of the r&IA(c)
332-428 and 429-447 segments (segments X and Y) led to a
dramatic loss of toxicity on
M.
sexta (Fig. 4,
A-C).
Notably,
CC2, containing only codons 429-447 from CryIA(a), was 27-
fold less toxic than the parent proteins, suggesting a key role
for these relatively few residues of the
cryZA(c)
gene product
in the intoxication of
M. sexta.
Addition of codons 332-428
(segment X) from CryIA(a) to form CC4 resulted in complete
restoration of parental toxicity whereas the alternative addi-
tion of codons 448-722 to form CC6 (Fig.
4B)
yielded a protein
that was 3.6-4.3-fold less toxic than the parent proteins. The
addition of the CryIA(a) 448-722 segment (segment X) had
little effect by itself either before (CC3) or after (CC7) the
addition of the whole 332-447 CryIA(a) segment (Fig. 4,
A-
C). We conclude therefore that the decreased toxicity created
by replacement of the C&A(c) 429-447 segment with that
from CryIA(a) was suppressed most effectively by the addition
of the 332-428 segment from the CryIA(a).
When tested on
M. sexta,
the CryIA(a) segments X and Z
together had little dependence on the CryIA(a) Y segment
since the latter could be replaced with the corresponding
CryIA(c) region (to form CC5) with little effect on toxicity.
The strong dependence of the CryIA(c) but not CryIA(a)
variable region on its own segment Y suggests that different
portions of these two toxins are critical for their interaction
by guest, on July 10, 2011www.jbc.orgDownloaded from
Crystal Protein Specificity-determining Regions
20927
A.
+I0
0Wcm*)
1000
t
‘:py+fc7
0 1 2 3
No. crylA(a) Segments
B.
L%O
Wcm*)
1000
f
cc2
I
No. &A(a) Segments
FIG. 4. Effect of substitution order on toxicity to M. sexta.
The figure shows the LCm values of chimeric proteins as a function
of the number of CryIA(a) segments introduced into CryIA(c). The
first point
of each
panel is a different segment substitution. Vertical
bars show one standard deviation confidence intervals.
0.01
0.1 1 10
100 LC,,Hv
FIG. 5. A plot of the ratios of LCao values of control and
chimeric proteins Ccl-CC7 to two insects as a function of
combined L&O values.
Filled circles denote the parent protein
values. Horizontal lines indicate one standard deviation confidence
intervals. ms, M.
senta;
Hu, H. uirescens.
with the M. sexta midgut. Interestingly, CC5 was marginally
the least toxic of the chimeric toxins to H. uirescens (Fig. 3A),
suggesting that the combined action of the CryIA(a) X and Z
segments is somewhat impeded by the Y segment of CryIA(c).
Discrimination Versus Activity
of the
Toxins-An alterna-
tive way of viewing the activity of each of the insecticidal
proteins to two different insects is shown in Fig. 5. This figure
presents a plot of the ability of each toxin to discriminate
between the two insects as indicated by the ratio of the LCSo
values uersu.s combined LCSo, i.e. 10(sverage log Lcm) of each pro-
tein toward both insects. The differential efficacy of B. thu-
ringiensis strains toward insects has been characterized pre-
viously by the ratio of LCsO values to two insects (e.g. Ref.
27). Although this is an adequate measure of the ability of B.
thuringiensis insecticidal proteins to discriminate between
two insect targets, it does not provide the absolute level of
toxicity toward the two insects (i.e. two toxins could have the
same ratio of toxicities but be more or less toxic to both
insects). The comparison method introduced in Fig. 5 uses
the ratio of LCSo values as the horizontal axis, preserving this
information, and enhances the comparison by including a
measure of overall toxicity, the combined L&o. This facili-
tates comparison of the toxins and shows clearly, for example,
that CryIA(c) and CC1 have similar discrimination although
CC1 is clearly less toxic. Similarly, CC2 and CC5 have com-
parable combined LCSo values but a 250-fold difference in
discrimination
between
the two insects.
In terms of the current analysis, the points near the line
connecting the value for CryIA(c) with that for CryIA(a) in
Fig. 5 indicate those proteins (i.e. CC3, CC4, CC7, CC8, CC9,
CryIA(b), and perhaps CC5) in which the increase in com-
bined LCSo and decrease in discrimination are attributable to
the loss of toxicity to H. uirescens. For CC1 and CC6, the
combined LCbo value is increased, but the discrimination value
is little changed from CryIA(c), indicating that the changes
in these proteins have caused primarily a loss of toxicity to
both insects and not an altered specificity. The unique loca-
tion of CC2 in Fig. 5 demonstrates the large shift in discrim-
ination caused by the much larger loss of toxicity to M. se&a
than H. uirescens, a result unexpected from the activities of
either parent protein.
Fine Structure Analysis
of
the 429-447 Region (Segment
Y)-A further subdivision of this segment of the cryIA(c)
gene was made by using site-directed mutagensis to create
PuuII restriction sites at two positions of potential homology
to the PuuII site in cryIA(a) (Fig. 6A). These sites, introducing
substitution mutations, were used to form a set of cryIA(a)/
cryIA(c) recombinants between codons 429 and 447 which
either incorporate or omit a two-amino acid insertion relative
to the CryIA(a)-derived sequence of CC2, as shown in Fig.
6B. The purified, soluble protoxins were bioassayed three to
four times on M. sexta and H. uirescens in parallel with the
control crystal proteins from B. thuringiensis HD-1-Dipel,
CryIA(c), and CC2, as shown in Table II. In this series of
bioassays, the M. sexta larvae were generally more susceptible
to the toxins than those used for the bioassays in Table I, and
the LCSo of CC2 was 50-fold higher than that of CryIA(c).
The first construct, CClO, with the PuuII site at codons
A CrylA(a)
WA(c)
pcc10
pcc11
B
crylA(c)
cc10
cc1 1
cc12
cc13
cc14
cc15
cc2
FIG. 6. Construction of chimeric proteins CClO-CCl5. A,
nucleotide
and deduced amino acid sequences of
cryZA(a), cryZA(c),
and base changes in two constructs introduced by site-directed mu-
tagenesis. PuuII sites are
ouerlined.
B, amino acid sequences of
CryIA(c) and chimeric proteins CC2 and CClO-CClB. Highlighted
sequences are from CryIA(c).
by guest, on July 10, 2011www.jbc.orgDownloaded from
20928 Crystal Protein Specificity-determining Regions
TABLE II
Toxicities
of chimeric proteins constructed by exchanging segments of
CrylA(a) and CrylA(c) between codons 429 and 447
H.
uirescens M. sexta
Protein”
LCmb
1
S.D. C.I.’ LC, 1 S.D. C.I.
wh’
rig/cm’
B.t. HD-l-Dipel 13 8-22 4
3-5
CrylA(c) 9
6-14 5 4-6
cc2 34 21-56 243 178-330
cc10
35 20-62
19
13-27
cc11
87 62-122 543 372-794
cc12 34 20-59 604 507-720
cc13 122 64-233 >8000 -
cc14 45 25-81 590 411-847
cc15 52 35-77 1277 808-2017
a Solubilized crystals from
B. thuringiensis (I3.t.)
subspecies
kur-
staki
HD-l-Dipel or solubilized inclusions from cloned genes ex-
pressed in
E. coli.
Composition of chimeric proteins CC2, CClO-CC16
are shown in Fig. 6.
’ rig/cm’ of diet.
’ One standard deviation confidence interval; see “Materials and
Methods” for details.
442 and 443, caused two amino acid substitutions (Asn + Ala
at residue 442 and Ser + Gly at residue 443; Fig. 6) and
resulted in an approximately 4-fold reduction in toxicity to
both M. se&a and H. virescens when compared with CryIA(c)
(Table II). The second construct, CCll, with the PvuII site
at codons 439-441, yielded three amino acid substitutions
(Gly w Ala at residue 439, Phe + Ala at residue 440, and Ser
-+ Gly at 441; Fig. 6), and this protein had an approximately
IlO-fold lower toxicity to M.
sexta
although the toxicity to
H.
uirescens
showed a lo-fold decrease from that of the parent
protein.
Chimeric proteins CC12 and CC14 were constructed by
combining the
cryZA(c)
gene from codon 429 to each of the
two PLJUII sites with the remainder of the Y segment coming
from cryZA(a). Reversing the sources of the genes used to
prepare the CC12 and CC14 constructs yielded chimeric pro-
teins CC13 and CC15 (Fig.
6B).
As shown in Table II, proteins
CC12 and CC14 were 3.5~&fold less toxic to
H. uirescens
than
CryIA(c) whereas the toxicity to
M. sexta
dropped approxi-
mately 120-fold (the near identity of the toxicities of these
two proteins seems remarkable since they differ by a two-
amino acid deletion and one substitution (Fig.
6B)).
An even
greater loss in toxicity to
M. sexta
was observed with chimeric
protein CC15 (about 260-fold), and a >1600-fold loss of tox-
icity to this insect was found with CC13. Large amounts of
CC13 inclusions were capable of killing
M. sexta,
so this
protein cannot be regarded as completely nontoxic. The tox-
icity of chimeric proteins CC15 and CC13 to
H. uirescem
was
reduced only about 4-fold and about IO-fold, respectively. The
improved toxicity to M. sexta of CC15 compared with the
nearly nontoxic CC13 results from the deletion of amino acids
442 and 443, the same ones substituted in CC10 (Fig.
6B).
Fig. 7 shows a plot similar to that of Fig. 5 displaying the
combined LC& values as a function of the ratio of the LCbO
values for the two insects using the constructions in Table II.
As noted in Fig. 5, CC2 has a higher combined L& than
CryIA(c) and a dramatically altered discrimination between
M. sexta
and
H. virescens.
The relationship between these
two proteins remains similar in Fig. 7. The two-amino acid
substitution in CryIA(c), CClO, shows an increased combined
LCsO with little change in discrimination compared with
CryIA(c). By contrast, CCll, differing by only three amino
acid substitutions from CryIA(c), has the same discrimination
for the two insects as CC2 with a higher combined LCsO than
CC2. Three
cryZA(a)/cryZA(c)
recombinants, CC12, CC14,
Combined
LC50
p cc13
/ 1 ... 0 . . . ..a . . . . . ..a
0.01
0.1
1
10
. ..--J
e- LC50Ms
100 LC<,, Hv
-_
>hk
Eq”tll >H”
Toxlclly
Twdclly TOXiClty
FIG. 7.
A plot of the ratios of L&O values of control and
chimeric proteins CClO-CC15 to two insects as a function of
combined LCeo values. Filled
circles indicate values for the parent
proteins.
Horizontal lines
denote one-standard deviation confidence
intervals.
ms, M. se&a; Hv, H. virescens.
and CC15 have values that cluster relatively close to CC2
and CCll. Chimeric protein CC13, with an undeterminably
high LCso on
M. senta,
has a higher combined LCbo and higher
M. sexta
to
H. virescens
LC,o ratio than any of the other
proteins tested.
DISCUSSION
Although significant advances have been made in charac-
terizing the genes coding for
B. thuringiensis
insecticidal
proteins (l), relatively little is known about the mechanisms
leading to the death of susceptible insects. It has been dem-
onstrated that specific receptors on the epithelial gut cells
bind different toxins (4,5), and it would be expected that the
number of such binding sites and the binding affinities would
be important in determining host range and potency. The
interaction of the toxin with the epithelial gut cells is followed
by changes in permeability, as documented by electron mi-
croscopy and the release of small molecules (28). Two studies
(14, 29) have identified partially overlapping specificity-de-
termining regions, with different
B. thuringiensis
proteins,
but the precise functions of these regions have not been
determined, and it remains unknown whether the same region
in each is required for toxicity to insects other than the ones
that were tested.
In the present investigation, we have analyzed two related
lepidopteran-specific proteins, CryIA(a) and C&A(c), to de-
termine the regions controlling the specificity of these pro-
teins toward
M. sexta
and
H. uirescens.
CryIA(c) is equally
toxic to both insects whereas CryIA(a) is significantly less
toxic to
H. virescens
than to
M. sexta.
To locate the determi-
nants of toxicity to
H. virescens,
chimeric genes were con-
structed, and the gene products were assayed for insecticidal
activity. Pairwise comparisons of data obtained with the
parent proteins and several chimeric constructs indicated that
differences in the amino acid sequences prior to codon 331
and following codon 722 had no effect on toxicity to either
insect. However, since we observed some unpredicted changes
in
toxicity depending on the precise combination of sequences
in experiments combining segments of these genes between
codons 332 and 722, we cannot rule out the possibility that
exchanging the 1-331 and 722-1156/78 segments in some of
the untested combinations would result in a difference.
The main variability in sequence between CryIA(a) and
CryIA(c) lies between residues 332 and 722. Deletion analysis
and protein sequencing data suggest that the carboxyl termi-
nus of the toxicity and possibly also the specificity determi-
nants lies between codons 607 and 623 (9, 13).
The codon
by guest, on July 10, 2011www.jbc.orgDownloaded from
Crystal Protein Specificity-determining Regions
20929
332-722 region was divided into three segments (332-428,
429-447, and 448-722), and a series of constructs was pre-
pared in which segments of cryIA(c) were replaced by the
corresponding sequences from cryIA(u). Assays of the result-
ing chimeric proteins showed that the dependence of toxicity
on particular segments was different with the two test insects.
With M. sextu, the toxicity of the chimeric proteins did not
differ dramatically from that of the parent proteins except for
constructs CC2 and CC6. In the latter two proteins, the
presence of the CryIA(a) codon 429-447 segment without the
CryIA(a) 332-428 segment resulted in a significant reduction
in toxicity which could be restored completely by addition of
the CryIA(a) 332-428 segment (constructs CC4 and CC7).
The replacement of the corresponding CryIA(c) segment with
the CryIA(a) 332-428 segment alone (Ccl) also resulted in a
modest decrease in toxicity.
The 27-50-fold increase in LCsO on M. sexta observed with
the chimeric protein CC2 relative to the C&A(c) parent
protein prompted a further analysis of the codon 429-447
sequence. Additional manipulation to substitute and subdi-
vide in the codon 429-447 segment resulted in three proteins
with significant differences in toxicity. First, CClO, with
substitution mutations at residues 442 and 443, displayed a
modest loss of toxicity with no change in discrimination
between the two insects from the parent C&A(c). Second,
CCll, having three amino acid changes (codons 439-441),
showed toxicity and discrimination closely approximating
those of CC2, which has 11 substitutions and two insertion/
deletions relative to CryIA(c). Third, CC13, having seven
substitutions in the 434-441 segment relative to CryIA(c),
was sufficiently nontoxic to M. sexta to prevent determination
of an LCsO value with soluble protoxin, yet this protein re-
tained moderate toxicity to H. uirescens. The broad range of
toxicity and specificity demonstrated by the few amino acid
changes in this second set of chimeric proteins suggests that
a limited number of single amino acid substitutions will
allow the identification of residues controlling specificity of
CryIA(c) toward M. sextu. Furthermore, preliminary experi-
ments indicate that CC2 is as stable to proteolysis as CryIA(c)
in the presence of either M. sexta or H. uirescens brush border
membrane vesicles so that it should be possible to determine
whether the effect on CC2 toxicity is caused by reduction in
binding to a receptor or to some later stage in intoxication
(the possibility of enhanced susceptibility to proteolysis has
not yet been ruled out for the other mutations in the 429-447
segment).
For H. virescens, replacement of the cryIA(c) segments with
counterparts from cryIA(u) resulted in a progressive reduction
in toxicity depending on the number of segments replaced.
An exception was construct CC5, which was marginally less
toxic than CryIA(a) but contained only two of the three
segments (332-428 and 448-722). However, these two seg-
ments represent the great majority of the variable sequence
and may exert an overall dominant effect on toxicity.
Thus, the response of the two insects to the same set of
chimeric gene products falls into two classes: context-depend-
ent toxicity (M. se&u) or progressive conversion of one pa-
rental type to another with increasing amounts of sequence
substitution (H. uirescens). In contrast, Ge et al. (Id), who
studied the same two genes, showed that the determinants of
toxicity of CryIA(a) to B. mori were located solely
in the
codon 332-447 region. Our preliminary results with Tricho-
plusia ni* suggest that specificity of CryIA(c) to the latter
insect can also be accounted for by the codon 332-447 seg-
ment. In all, these results indicate the existence of three
2 H. E. Schnepf and K. Tomczak, unpublished data.
different patterns of toxicity determinants for specific lepi-
dopteran larvae involving different segments of the codon
330-620 region of the toxin-encoding portion of the proteins:
1) incremental and involving the entire variable region (as in
the toxicity of CryIA(c) to H. uirescens); 2) context dependent
(as in the toxicity of CryIA(c) to M. sextu); and 3) a single,
independent specificity domain (as in the toxicity of CryIA(a)
to B. mori).
A modification of the third pattern of toxicity was found in
studies of chimeric proteins containing segments of CryIIA
and CryIIB (29). CryIIA (“P2” toxin; Ref. 30), a major com-
ponent of cuboidal crystals, is toxic to mosquito larvae (Aedes
uegypti) and lepidopteran larvae; the closely related CryIIB
protein is toxin only to lepidopteran larvae. Analyses of the
insecticidal activities of chimeric CryIIA/CryIIB proteins
showed that residues 307-382 were required for dipteran
toxicity. However, the data obtained with CryIIA/CryIIB
chimeras indicated that distally located segments also influ-
enced toxicity, possibly by affecting the overall protein con-
formation. Structural incompatibilities may also explain our
current results with CryIA(c)/CryIA(a) chimeric proteins. M.
sextu toxicity of the codon 332-431 segment of CryIA(a) had
moderate dependence on the source of the 432-447 segments;
however, M. sexta toxicity of the codon 332-431 segment of
CryIA(c) was critically dependent on the 432-447 segment of
CryIA(c), implying close structural interactions between the
latter two segments. By contrast, exchanges of the codon 332-
447 and 448-722 segments yielded proteins with relatively
predictable activities and discriminations, as illustrated in
Fig. 5, suggesting that these segments are relatively independ-
ent structural elements. This notion is reinforced by the
existence of the naturally occurring CryIA(b) containing an
apparent crossover point between CryIA(c) and CryIA(a) at
codon 460.
Only one of the chimeric insecticidal proteins we have
constructed was essentially nontoxic, and this protein, which
had seven amino acid substitutions, displayed the tremendous
loss in toxicity only to one of the two test insects. In contrast,
Ge et al. (14), working with the same genes but assaying only
for toxicity to B. mori, noted that exchanges at a point in the
448-722 segment resulted in total inactivation of the chimeric
proteins. Kobilka et al. (31) observed that a subset of chimeric
adrenergic receptors had lost activity, and they suggested that
specific combinations of transmembrane helices made essen-
tial interactions that could not be supplied by the related
protein in the chimera. Analogously, we take the data of Ge
et al. (14) to mean that the 448-722 segment contains within-
segment structural interactions that can be perturbed easily
by internal crossovers between these related genes. Interest-
ingly, three of the highly conserved regions found in most B.
thuringiensis insecticidal proteins (Ref. 1 and Fig. 8) are in
this segment of the protein. Taken together with our current
results regarding the M. sextu specificity of CryIA(c), these
data support the proposal that the codon 332-447 and 448-
607 segments of the crystal protein gene are independently
acting protein structures, if not separate domains of the
protein; each variant of each of these two segments can be
combined with the other to yield a protein with a combined
activity that is predictable from its discrimination between
the two insects, and internal exchanges in these segments are
disruptive, resulting in less than expected toxicity.
Fig. 8 presents a diagram that shows the location of the
specificity determining regions in CryIA(c) to H. virescens,
M. sexta, and T. ni, in CryIA(a) to B. mori and in CryIIA to
A. uegypti as discussed above (see Ref. 1 for alignments of the
proteins). Fig. 8 also includes the region of toxicity of the
by guest, on July 10, 2011www.jbc.orgDownloaded from
20930
N , c
Crystal Protein Specificity-determining Regions
Variable ReQlOn
Toxin
CrylA(c) Heliothis virescens
CrylA(c) Manduca sexta
CrylA(c) Trichoplusia .#j
WA(a) E!QJ&U mpri
CrylA(a). CrylA(b) J&2&&&- (interred)
variant CrylA(b) lepidopteran vs. dipteran
CfyllA &t&s aepypti
amino acid number
FIG. 8. Diagram showing the location of specificity deter-
mining regions in
B. thuringiensis
crystal proteins. The me-
dium thick Eine depicts insecticidal proteins, the b&k rectangles mark
the locations of toxicity-determining regions for the indicated pro-
teins and insects, the empty rectangle shows the position of a deletion
in a variant CryIA(b) protein, the shaded vertical areas indicate the
positions of the five conserved regions present in most crystal protein
genes (l), the thick lines above the proteins show the locations of the
variable region and the toxin as indicated.
CryIA(a) and CryIA(b) proteins to Mumestrcz
brussicae
as
inferred from results reported by Hiifte et al. (3); it has not
been determined whether toxicity depends on the entire re-
gion or one or more segments thereof. Also shown are results
obtained by Haider and Ellar (32) for a variant CryIA(b)
protein toxic to both lepidoptera and diptera. Their data
indicate that three amino acid differences in the codon 537-
568 segment of CryIA(b) of
B.
thuringiensis
subspecies
uizu-
wui
ICl specify novel dipteran and lepidopteran protease
cleavage sites that allow the production of the toxin for the
respective insect targets. They have also shown that deletion
of codons 242-523 of this protein
(open box
in Fig. 8) allows
synthesis of a dipteran but not a lepidopteran toxin.
It is interesting that although all of the specificity-deter-
mining domains or regions fall within the variable region of
the
cryIA
genes (13) the specificity regions differ in size and
position. Also, the amino-terminal half of the variable region
(corresponding to residues 280-447 of CryIA(a)) is located
between the second and third of the five conserved regions
revealed by comparison of the sequences of 13 crystal proteins
(Ref. 1 and Fig. 8). The latter half of the variable region,
however, contains conserved regions 3-5 (Fig. 8). If, like
immunoglobulins, the conserved sequences represent struc-
tural scaffolding whereas the variable or hypervariable regions
confer differential specificity, the conserved regions of the
toxins may be involved more with protein conformation or
some common determinant of toxin activity rather than with
specificity-determining interactions with target cells. This
could explain, in part, the loss of toxicity observed with
chimeric proteins having exchange points in the carboxyl-
terminal half of the variable region (14).
The finding that the same set of chimeric proteins elicits
different responses by two different insects suggests that the
most active of the toxins to these two insects, CryIA(c),
interacts differently with the two insects. It follows then that
if alterations are made in the specificity-determining region,
not all of the species that are susceptible to CryIA(c) may be
affected to the same extent. Conversely, if insect resistance
developed to an insecticidal protein, such resistant insects
would not necessarily be resistant to a related protein. In fact,
it has been demonstrated recently that the Indian meal moth
(Plodiu
interpunctellu),
in which resistance was developed to
CryIA(b), is more sensitive to the CryIC protein than the
1.
2.
3.
4.
5.
6.
I.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23,
REFERENCES
Hiifte, H., and Whiteley, H. R. (1989) Microbial. Reu. 53, 242-
255.
Jaquet, F., Hiitter, R., and Liithy, P. (1987) Appl. Environ.
Microbial. 53,500-504
HBfte, H., Van Rie, J., Jansens, S., Van Houtven, A., Vander-
bruggen, H., and Vaeck, M. (1988) Appl. Environ. Microbial.
54,2010-2017
Hofmann, C., Vanderbruggen, H., HBfte, H., Van Rie, J., Jansens,
S., and Van Mellaert, H. (1988) Proc. Nutl. Acad. Sci. U. S. A.
85,7844-7848
Hofmann, C., Liithy, P., Hiitter, R., and Pliska, V. (1988) Eur.
J. Biochem. 173,85-91
Knowles, B. H., and Ellar, D. J. (1987) B&him. Biophys. Actu
924,509-518
Kronstad, J. W., and Whiteley, H. R. (1986) Gene (Amst.) 43,
29-40
Bietlot, H., Carey, P. R., Choma, C., Kaplan, H., Lessard, T., and
Pozsgay, M. (1989) Biochem. J. 260,87-91
Schnepf, H. E., and Whiteley, H. R. (1985) J. Biol. Chem. 260,
6273-6280
Wabiko, H., Raymond, K. C., and Bulla, L. A., Jr. (1986) DNA
(N. Y.) 5,305-314
Hafte, H., de Greve, H., Seurinck, J., Jansens, S., Mahillon, J.,
Ampe, C., Vandekerckhove, C., Vanderbruggen, H., van Mon-
tagu, M., Zabeau, M., and Vaeck, M. (1986) Eur. J. Biochem.
161,273-280
Whiteley, H. R., and Schnepf, H. E. (1986) Annu. Reu. Microbial.
40,549-576
Geiser, M., Schweitzer, S., and Grimm, C. (1986) Gene (Am&.)
48,109-118
Ge, A. Z., Shivarova, N. I., and Dean, D. H. (1989) Proc. N&l.
Acad. Sci. U. S. A. 86.4037-4041
Chothia, C., and Lesk, A. M. (1986) EMBO J. 5, 823-826
Casadaban, M., and Cohen, S. N. (1980) J. Mol. Biol. 138, 179-
201
Kunkel, T., Roberts, J., and Zakour, R. (1987) Methods Enzymol.
(Wu, R., and Grossman, L., eds) Vol. 154, pp. 367-382, Aca-
demic Press, New York
Kronstad, J. W., Schnepf, H. E., and Whiteley, H. R. (1983) J.
Bacterial. 154, 419-428
Vieira, J., and Messing, J. (1987) Methods Enzymol. (Wu, R., and
Grossman, L., eds) Vol. 153, pp. 3-11, Academic Press, New
York
Maniatis, T., Fritsch, E. F., and Sambrook, J. (1982) Molecular
Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory,
Cold Spring Harbor, NY
Schnepf, H. E., and Whiteley, H. R. (1981) Proc. Nutl. Acud. Sci.
U. S. A. 78,2893-2897
Bradford, M. M. (1976) Anal. B&hem. 72, 248-254
Bell, R. A., Owens, C. D., Shapiro, M., and Tardif, J. R. (1981)
in The Gypsy Moth: Research toward Integrated Pest Mannge-
ment (Doane, C. C., and McManus, M. L., eds) pp. 599-655,
Forest Service Technical Bulletin 1584
24. Finney, D. J. (1973) Probit An&y&s. Cambridge University Press,
Cambridge, U. K.
25. Adang, M. J., Idler, K. F., and Rocheleau, T. A. (1987) in Bio-
technology in Invertebrate Pathology and Cell Culture (Mara-
morosch;-K., ed) pp. 85-99, Academic Press, San Diego, CA
26. Widner, W. R., and Whiteley, H. R. (1989) J. Bacterial. 171,
965-974
27. Dulmage, H. T., and Cooperators (1981) in Microbial Control of
Pests and Plant Diseases 1970-1980 (Burges, H. D., ed) pp.
193-222, Academic Press, London
28. Liithy, P., and Ebersold, H. R. (1981) in Puthogenesis of Inuer-
tebrate Microbial Diseases (Davidson, E. W., ed) pp. 235-267,
Allenheld, Osmun and Co., Totowa, NJ
29. Widner, W. R., and Whiteley, H. R. (1990) J. Bacterial. 172,
2826-2832
30. Yamamoto, T., and McLaughlin, R. E. (1981) Biochem. Biophys.
Res. Commun. 103,414-421
31. Kobilka, B. K., Kobilka, T. S., Daniel, K., Regan, J. W., Caron,
M. G.. and Lefkowitz, R. J. (1988) Science 240.1310-1316
32. Haider, M. Z., and Ella;, D. J.‘(1984) J. Mol. BioZ: 208, 183-194
33. Van Rie, J., McGaughey, W. H., Johnson, D. E., Barnett, B. D.,
nonresistant parental insects (33).
and Van Mellaert, H. (1990) Science 247,72-74
by guest, on July 10, 2011www.jbc.orgDownloaded from
