Background
Bacillus thuringiensis (Bt) has been
sold commericially and used as a biopesticide worldwide for
over half a century. However, growing public concern
surrounding Bt use has sparked worldwide debate over
current policies [ 1 ] . For example, in India, fear over a
potential
Bombyx mori (silkworm) epizootic, or
microbial pathogen outbreak, inspired a governmental ban on
the use of Bt, despite the nation's continuing use of
traditional chemical pesticides [ 2 ] .
While pest control with Cry toxins that possess low
B. mori activity (
i.e . Cry1Ac) is a viable solution in
affected countries, it is worthwhile to investigate the
specific molecular mechanisms that make Cry1Aa highly
active. Early work took advantage of the fact that Cry1Aa,
but not Cry1Ac, is toxic to
B. mori . For example, Ge
et al. [ 3 ] exchanged hypervariable
regions between genes encoding the two toxins and localized
the toxicity specifying region of Cry1Aa to residues
332-450 in domain II. A follow-up study demonstrated the
toxicity specifying residues were involved in binding
B. mori brush border membrane
vesicles [ 4 ] . More specifically, alanine substitution or
deletion of residues 365 to 371 removed nearly all toxicity
and binding to
B. mori BBMV [ 5 ] .
Recently, research on
B. mori has focused on purifying and
cloning the midgut epithelial receptors targeted by Cry1Aa
toxin. The first toxin-binding receptor purified from
B. mori was a 120-kDa aminopeptidase
N (APN), which appears around 110-kDa on SDS-PAGE gels when
preparative conditions are used that cleave its
glycosyl-phosphatidylinositol (GPI) anchor. This receptor
was shown to bind Cry1Aa with a 7.6 nM affinity, as
determined by Scatchard analysis with ELISA binding assays
[ 6 ] . The APN was cloned and expressed in
E. coli and demonstrated to bind
Cry1Aa toxin on ligand blots [ 7 ] . These results indicate
the Cry1Aa-APN interaction was specific and that APN
glycosylation was not required for Cry1Aa binding. This is
not altogether surprising because Cry1Aa binding to
Manduca sexta APN has not been found
to be modulated by sugar binding [ 8 ] and the
B. mori APN sequence is 73.7%
identical to
M. sexta APN-1. Sequence alignments
with
Plutella xylostella APN receptor
indicate that a highly conserved region of APN likely
functions as the toxin binding site [ 9 ] . By testing for
toxin binding to lysylendopeptidase-digested
B. mori APN fragments, the toxin
binding site was suggested to be between Ile135 and Pro198.
A later study by these authors identified 120-kDa and
115-kDa APNs coeluting from an anion-exchange column that
together yielded a Cry1Aa affinity of 53 nM [ 10 ] . These
APNs eluted just prior to a 120 kDa APN with 7.6 nM
affinity. It is unclear whether the 120- and 115-kDa
proteins represent uncleaved and cleaved GPI-anchor
isozymes. Interestingly, this study also showed that Cry1Ac
toxin binds to the 120/115 kDa APN fraction with equal
affinity as Cry1Aa, and only 4-fold reduced affinity to the
isolated 120-kDa APN. Nonetheless, Cry1Aa is 210 times more
toxic than Cry1Ac to
B. mori [ 4 ] . As a whole,
B. mori APN research indicates the
presence of at least three genetic isoforms [ 7 11 12 ] ,
with toxin affinities ranging from nanomolar to none at
all.
In addition to APN, a completely different toxin
receptor class has been affinity precipitated by toxin from
solubilized
B. mori midgut proteins. In this
manner, Nagamatsu
et al. [ 13 ] purified a 175-kDa
glycoprotein (BtR175) that bound Cry1Aa toxin.
Interestingly, these authors did not observe binding of
Cry1Aa to APN-sized bands in ligand blot studies with BBMV.
Antibodies produced to BtR175 blocked toxin binding to the
receptor in BBMV. The antibody serum also reduced Cry1Aa
activity against
B. mori when it was fed to larvae
prior to toxin addition to the diet [ 14 ] . The same group
cloned and introduced the BtR175 gene with a baculovirus
vector into
Spodptera frugiperda Sf9 cells.
Addition of Cry1Aa caused swelling and lysis of only the
Sf9 cells expressing BtR175. Based on sequence analysis,
the receptor was characterized as a cadherin-like
glycoprotein containing nine cadherin repeats, a membrane
proximal region, one transmembrane region, and a small
cytoplasmic domain [ 15 ] . Ihara
et al. [ 16 ] also purified and
partially sequenced what was presumed to be the same
cadherin-like receptor. Binding studies indicated that the
affinity of the cadherin for Cry1Aa is equivalent to that
of the brush border membrane vesicles from
B. mori [ 16 ] , an affinity that is
substantially lower than the APN affinities reported.
Recently, cDNA variants of BtR175 have been discovered,
showing at least three alleles of the cadherin-like
receptor are found in
B. mori [ 17 ] . It is likely that
glycosylation plays a major role in cadherin-like receptor
isoforms as well, as glycosylation has been observed
previously for the
M. sexta cadherin-like receptor BT-R
1 [ 18 ] .
Progress in research on silkworm receptors for Bt toxins
has provided a means for assaying mutant toxins with
potentially altered binding and activity. In this study, we
tested the hypothesis that Cry1Aa binds to both the 120-kDa
B. mori APN and the 175-kDa
B. mori cadherin-like protein. Based
on the previous work of Ge, et al. [ 3 ] and Lee, et al. [
4 ] , we also postulated that domain II of Cry1Aa is the
significant binding domain. These hypotheses were tested
for the first time in studies with purified, native
B. mori receptors (rather than BBMV)
under real-time, non-labeled toxin binding conditions.
Results
Bombyx mori aminopeptidase N and cadherin-like
receptor purification
To investigate the specificity of Cry toxins for
B. mori receptors, the two known
B. mori midgut receptors were
purified from
B. mori BBMV. Solubilized
B. mori BBMV proteins were
separated by Q Sepharose anion-exchange chromatography
and all eluted fractions were tested for APN enzymatic
activity. Additionally, Cry1Aa toxin binding capability
was assayed by "slot blotting" all fractions and probing
with biotin-Cry1Aa. The chromatogram in Fig. 1displays
the separation of cadherin and APN from BBMV proteins.
APN isozymes of 100- and 110-kDa were detected that did
not show Cry1Aa-binding in slot blot assays (Fig. 1;
fractions 24-25 and 30-31). Such isozymes have been
reported previously [ 11 12 ] . In addition, a 115-kDa
APN was detected with Cry1Aa-binding capability (Fig. 1;
fractions 33-36). As expected, fractions were also
observed that exhibited no APN enzymatic activity but
bound Cry1Aa on slot blots (Fig. 1; fractions 26-27).
Initially these fractions were predicted to contain the
cadherin-like Cry1Aa-binding protein [ 13 14 16 ] . The
candidate receptor fractions for APN and cadherin were
separately loaded on a size-exclusion column for further
purification (Fig. 2Aand 2B). A protein with APN
enzymatic activity eluted 75 minutes after injection
(Fig. 2A; fractions 15-16), approximately 4 minutes after
the 120-kDa
L. dispar APN elutes on the same
column [ 19 ] . The candidate cadherin-like receptor
fraction eluted in fractions 9-11 at around 180 kDa (Fig.
2B).
Analysis of receptor purity
The pooled and concentrated candidate receptor
fractions were examined by SDS-PAGE before and after
size-exclusion purification to assess purity (Fig. 3).
The putative cadherin-like receptor material appears at a
molecular size around 180 kDa, both before and after
secondary purification (Lanes 2 and 1, respectively).
Several BBMV proteins appear present in the
APN-containing fraction prior to size-exclusion
purification (Fig. 3; Lane 4). The molecular weight of
the final, purified APN was estimated to be 115-120 kDa
(Fig. 3; Lane 3). It is not known whether the GPI anchor
is still intact on the APN receptor; however, in the
current study, phosphatidylinositol-specific
phospholipase C (PIPLC) was not used during BBMV
preparation. It was shown previously that APN may be
purified with intact GPI-anchors if PIPLC is omitted from
the preparation buffer [ 6 ] . It is likely that our APN
has similarly retained the GPI anchor.
In view of the fact that
B. mori BtR175 possesses sequence
similarity to
M. sexta BT-R
1 , the candidate fraction was probed
on a slot blot with anti-BT-R
1 polyclonal antiserum. A weak to
moderate cross-reactivity with anti-BT-R
1 was observed for
B. mori BBMV as well as the
putative cadherin-like receptor fraction, providing
strong evidence that the material is a cadherin-like
protein (Fig. 4). Anti-Bt-R
1 antibody recognition was not
observed for fractions eluting before and after the
cadherin material, nor for the purified APN (Fig. 4).
Similar antibody assays were not performed to
substantiate the identity of the purified APN because it
clearly displayed strong, characteristic APN enzyme
activity.
The purity of both receptors was further examined by a
Cry1Aa toxin ligand blot (Fig. 5). Both APN and the
cadherin-like receptor fractions bound biotinylated
Cry1Aa. No other toxin-binding bands were apparent, and
neither purified receptor sample was visibly
cross-contaminated with the other receptor (Fig. 5).
Affinity estimation by surface plasmon
resonance
Cry toxin binding studies have been reported
previously for
B. mori that used BBMV assays or
used purified receptors in ELISA assays or blots;
however, no Cry toxin studies concerning
B. mori have been published
employing SPR analysis. The affinity of Cry1Aa binding to
B. mori APN and
B. mori cadherin receptors was
evaluated by real-time kinetic analysis on a BIAcore
2000. Simple bimolecular binding of Cry1Aa was observed
to both
B. mori APN and cadherin (Fig.
6Aand 6B). Toxin-receptor on-rates for association (
k
a ), off-rates for dissociation (
k
d ), and overall binding affinity (
k
d /
k
a , or
K
D ) were calculated for toxin binding.
The apparent rate constants for wild-type Cry1Aa and
B. mori APN were
k
a = 2.0 × 10 4M -1s -1(+/- 1.3 × 10
2),
k
d = 1.5 × 10 -3s -1(+/- 1 × 10 -5),
and
K
D = 75 nM. To
B. mori cadherin, significantly
tighter affinities were obtained:
k
a = 1.3 × 10 4M -1s -1(+/- 6.1),
k
d = 3.3 × 10 -5s -1(+/- 1 × 10 -5),
K
D = 2.6 nM. This apparent off-rate
clearly accounted for Cry1Aa's higher affinity for
cadherin than for APN. The cadherin off-rate observed in
this study could have significant consequences in vivo:
slow toxin dissociation may enable protracted lingering
near the brush border membrane surface, greatly
facilitating toxic (domain I) insertion and subsequent
pore formation. The overall affinity determined in the
present study for Cry1Aa to BtR175 (2.6 nM) agrees well
with the findings of Ihara
et al. [ 16 ] by a different assay
(0.8 nM).
We also explored the specificity of Cry1Aa for the
native
B. mori receptors by comparing the
binding response of Cry1Aa with the binding of Cry1Ab,
Cry1Ac, and domain-switched toxin 4109 (Fig. 7Aand 7B).
Hybrid toxin 4109 is particular useful in this context,
because it is comprised of domains I and II from Cry1Aa
and domain III from Cry1Ac (Aa/Aa/Ac) [ 3 ] .
Hybrid-toxin 4109 binding to both receptors was not
noticeably different from Cry1Aa: for APN binding,
k
a = 1.9 × 10 4M -1s -1(+/- 1.4 × 10
2),
k
d = 1.5 × 10 -3s -1(+/- 1.6 × 10 -5),
and
K
D = 78 nM; for cadherin binding,
k
a = 1.3 × 10 4M -1s -1(+/- 2 × 10 2),
k
d = 3.34 × 10 -5s -1(+/- 2 × 10 -5),
and
K
D = 2.6 nM. In stark contrast, Cry1Ab
and Cry1Ac showed no apparent binding to either receptor
(Fig. 7Aand 7B). These results are entirely consistent
with the hypothesis that Cry1Aa domain II (alone) is
essential for binding to both the APN and cadherin-like
receptors as purified in the present study.
Discussion
The dissociation constants presented are the first
determined for
B. mori Cry receptors by the use of
SPR technology. Additionally, the apparent affinity of
Cry1Aa for the cadherin-like receptor is the highest
observed affinity to date for Cry toxin binding to purified
receptors using SPR. This finding emphasizes the important
biological role that this receptor class plays for Bt
toxins. Recently, using phage display technology, a scFv
molecule with short sequence homology to
M. sexta and
B. mori cadherin-like receptors was
shown to bind domain II of Cry1Aa, Cry1Ab, and Cry1Ac
toxins [ 20 ] . In the present study, only Cry1Aa shows
measurable binding to the purified, native cadherin-like
receptor from
B. mori . This finding may be the
result of purification of a particular receptor variant
with Cry1Aa specificity (e.g., one of potentially several
glycosylated isoforms). Alternatively, it may reflect
greater specificity of Cry1Aa domain II for neighboring
residues on the
B. mori cadherin-like receptor beyond
the conserved Cry1A-toxin binding segment, which might be
absent in smaller peptide sequences.
The apparent Cry1Aa affinity for purified APN measured
in this study, 75 nM, is 10-fold higher than the value
reported by Yaoi
et al. [ 6 ] for purified 110-kDa APN
using a separate technique (7.6 nM). In the aforementioned
study, an ELISA assay was used to indirectly calculate
affinity by incubating receptor-bound toxin with a
peroxidase-conjugated anti-Cry1Aa antibody over 1.5 hours
at 37°C. It is possible that the difference in binding
constants reflects our condition of more direct
receptor-binding measurement in "real-time", as well as the
different binding buffers and temperature used.
Yaoi
et al. [ 21 ] estimated the
toxin-binding region of
B. mori APN to be between Ile135 and
Pro198 based on toxin blot overlays with protease-digested
APN fragments. BLAST sequence alignments [ 22 ] yielded 81%
identity and 96% similarity between this 63 residue stretch
and the homologous region of
M. sexta APN-1, which also binds
Cry1Aa toxin (Jenkins, unpublished observation).
Interestingly, in
L. dispar APN-1, which does not bind
Cry1Aa, the same stretch is only 37% identical and 56%
similar (4% unaligned gaps). These results appear to
support the findings of Yaoi
et al. [ 21 ] . However,
Heliothis virescens (tobacco budworm)
APN is only 45% identical and 53% similar to
B. mori APN, yet it binds Cry1Aa
toxin with high affinity [ 23 ] . Moreover, sequence
alignments with APN from
Lactobacillus ,
Streptococcus ,
Saccharomyces ,
Arabidopsis , rat, pig, yeast, and
human yielded more similarity than
H. virescens APN to the putative
Cry1Aa-binding region of
B. mori APN. It is likely that as the
X-ray crystal structures of APNs are solved, structural
alignments of APNs will help resolve the
specificity-determining regions more accurately.
Additionally, structural information will aid in the
rational construction of toxins with reduced binding for
beneficial insects without losing activity to target pests.
In this context a unique Cry1Aa binding epitope within
domain II has been identified that, when mutated, results
in specific reduction of toxicity to
B. mori (You, et al., unpublished
manuscript). The application of protein engineering to
B. thuringiensis insecticidal
proteins is entering a new era of tailoring pesticides with
reduced activity to beneficial insects as well as
increasing activity against pest insects [ 24 ] .
Conclusions
Domain II of Cry1Aa is both necessary and essential for
tight binding to two
B. mori midgut receptors, the
cadherin-like and aminopeptidase N receptors, a finding
that correlates with biological activity data. The Cry1Aa
binding affinity, as well as the dissociation rate for the
cadherin-like receptor, are the lowest measured using the
surface plasmon resonance technique. The SPR method
presented here may be useful for screening other Cry toxins
or Cry toxin variants specifically engineered to reduce or
eliminate specificity for receptors from this non-target
insect.
Materials and methods
Mutant toxin construction and preliminary
analysis
Hybrid toxin 4109 consisting of domains I and II of
Cry1Aa and domain III of Cry1Ac was constructed as
previously described [ 3 ] . Force-feeding bioassays on
B. mori and BBMV binding assays
were conducted as described [ 4 ] . Crystal proteins were
solubilized and trypsinized, and active toxins were
column purified as carried out previously [ 19 ] .
Receptor purification
B. mori midguts were dissected from
4 thor 5 thinstar larvae and brush border membrane
vesicles were prepared by the Wolfersberger method [ 25 ]
.
B. mori BBMV (10 mg in 10 ml) was
solubilized in 5 mg/ml CHAPS zwitterionic detergent
(Roche) overnight at 4 oC with gentle rocking.
Solubilized BBMV was centrifuged at 10,000 × g for 10 min
and supernatant was concentrated to 2 ml by Amicon YM30
ultrafiltration. The sample was then loaded on a Q
Sepharose HR 10/30 anion-exchange column. All column
chromatography was carried out on an ÄKTA Explorer
(Amersham Pharmacia Biotech). Low salt buffer (buffer A)
consisted of 20 mM Tris, 5 mM MgCl, 0.4 mg/ml CHAPS, pH
8.6, and the high salt buffer used was buffer A
containing 1 M NaCl. A step gradient of salt was used to
elute BBMV proteins. All fractions were tested for APN
enzymatic activity by the LpNA assay. Briefly, 390 μl of
sample are mixed with 10 μl of 2 mM
leucine-p-nitroanilide (containing a
leucine-phenylalanine dipeptide). A yellow chromophoric
change indicates aminopeptidase N activity, defined as
the ability to cleave a neutral amino acid from the
N-terminus of a polypeptide. Cry1Aa binding ability was
also checked by slot blotting fractions to PVDF membrane
and probing with biotinylated Cry1Aa [ 26 ] . Fractions
with Cry1Aa-binding ability and APN enzymatic activity
were concentrated to 2 ml volumes and loaded on a
Superdex 200 size-exclusion column (120 ml bed volume)
using Hepes-buffered saline (HBS; 10 mM Hepes, 150 mM
NaCl, 3.4 mM EDTA, pH 7.4) as running buffer. Absorbance
was monitored at 280 nm and 260 nm to judge protein
purity of collected peaks relative to flow through.
Fractions eluting around 115-120 kDa, the MW of APN, were
collected and protease inhibitors were added after a
final concentration. Anion-exchange fractions with
Cry1Aa-binding ability but without APN enzymatic activity
were also size-purifed, and fractions eluting around
175-250 kDa, the MW of BtR175, were collected and
concentrated. Approximately 0.10 mg (in 0.25 ml) of
cadherin-like receptor and 0.30 mg of APN (in 1 ml) were
obtained.
Analysis of receptor purity
Candidate receptor fractions were analyzed by 10%
SDS-PAGE (40 μl/lane) and stained with Coomassie
brilliant blue. For slot blot assays, 5 μg of
M. sexta or
B. mori BBMVs and 40 μl of
candidate receptor fractions were blotted onto PVDF
membrane and assays was carried out as reported
previously [ 27 ] , except for using 1:1000 anti-Bt-R
1 polyclonal antiserum. For ligand
blot assays, samples separated by SDS-PAGE (9%) were
transferred to PVDF overnight, blocked with 5% dried milk
in TTBS (50 mM Tris-HCl, 150 mM NaCl, 0.05% Tween 20, pH
7.5). Samples were probed with 50 μg biotin-Cry1Aa and
streptavidin-conjugated horseradish peroxidase for 1 hr
each, with 45 min TTBS washes, and developed in DAB/Urea
(BioRad).
Surface plasmon resonance with purified midgut
receptors
B. mori APN and cadherin were
immobilized on a CM5 sensor chip by the amine-coupling
method (Biacore AB). Receptors were diluted into ammonium
acetate, pH 4.2 prior to immobilization. An HBS (pH 7.4)
buffer flow rate of 50 μl/min was used for all
injections. Randomized toxin concentrations varying from
100 nM to 1000 nM were injected (110 μl) over the
receptor surfaces. Surfaces were regenerated with 6 μl
pulses of 10 mM NaOH, 250 μM ethylene glycol, pH 11.0 at
100 μl/min. Signal responses from a blank flowcell
containing ethanolamine as a blocking agent were
subtracted from all response curves and data were fitted
using BIAevaluation 3.0. The curves were fit to a simple
1:1 Langmuir binding model to obtain apparent rate
constants (A + B ↔ AB).
List of abbreviations
Bt,
Bacillus thuringiensis ; APN,
aminopeptidase N; GPI, glycosyl-phosphatidylinositol; BBMV,
brush border membrane vesicles; LpNA,
leucine-p-nitroanilide; PIPLC,
phosphatidylinositol-specific phospholipase C; HBS,
Hepes-buffered saline; SPR, surface plasmon resonance
