Effects of pH on comformational properties related to the toxicity of Bacillus thuringiensis δ-endotoxin

Biochimica et Biophysica Acta, 1159 (1992) 185-192 © 1992 Elsevier Science Publishers B.V. All rights reserved 0167-4838/92/$05.00

185

BBAPRO 34307

Effects of pH on conformational properties related to the toxicity of Bacillus thuringiensis 6-endotoxin Manju Grover Venugopal b, Michael G. Wolfersberger c and B.A. Wallace a,b a Department of Co'stallography, Birkbeck College, Unicersity of London, London (UK), b Department of Chemistry, Rensselaer Polytechnic Institute, Troy, NY (USA) and ¢ Department of Biology, Temple Unicersio', Philadelphia, PA (USA) (Received 14 April 1992)

Key words: Protein conformation; Toxin; Circular dichroism; 6-Endotoxin: (B. thuringiensis)

The 6-endotoxin of Bacillus thuringiensis subspecies kurstaki is an intracellular crystalline proteinaceous inclusion which, upon ingestion, is toxic to lepidopteran insects. Upon dissolution at pH > 9 it yields a protein subunit called protoxin. Under appropriate conditions, protoxin is hydrolyzed to a toxin molecule, which is responsible for killing the insect. It is known that this toxic activity decreases considerably above pH 10. In this study, circular dichroism spectroscopy has been used to examine the secondary structures of the protoxin and toxin molecules at different pH values to determine if there are detectable conformational changes associated with their pH-dependent functional properties. At pH 10. where toxic activity is approximately maximal, both the protoxin and toxin molecules were found to assume a conformation that is on an average approx. 26% a-helix and approx. 45%/3-structure. As the pH was increased above 10, where the insecticidal activity decreases, the magnitude of the CD spectrum at 222 nm decreased for protoxin and the calculated o~-helix contents of both protoxin and toxin molecules decreased. The net secondary structure did not change significantly at pH values below 10. Significant conformational differences are observed between the secondary structure of the protoxin and toxin molecules at different pH values. The pH-dependent changes in secondary structure of the protoxin and toxin can be correlated with the effects of pH on the insecticidal activity of these proteins.

Introduction During sporulation, the gram-positive aerobe Bacillus thuringiensis produces parasporal inclusions which are toxic to susceptible insect larvae upon ingestion [1]. Products containing these inclusions have been available commercially for several decades and continue to dominate the bioinsecticide market [2]. The parasporal inclusions, also called 6-endotoxins, produced by most subspecies of B. thuringiensis:are active only against the larvae of a few lepidopteran insects. Lepidopteran-active parasporal inclusions are usually bipyramidal crystals consisting of one or more 130-140 kDa polypeptides. These insecticidal polypeptides have been designated 'protoxins'. The complete insecticidal activity of each protoxin resides in a 5 5 - 7 0 kDa proteinase-resistant fragment, called the 'toxin'. Toxicity results from solubilization and partial digestion of the crystals in the larval midgut [3]. In vitro, bipyrami-

Correspondence to: B.A. Wallace, Department of Crystallography, Birkbeck College, University of London. Malet Street, London WG 1E 7HX, UK.

dal crystals can be dissolved at pH 9 or higher in the presence of reducing agents to yield protoxin. Protoxin can be converted into toxin by treatment with proteolytic enzymes such as trypsin [4]. The most widely used bioinsecticides for agriculture and forestry in North America are based on the HD-1 strain of B. thuringiensis subsp, kurstaki [2]. This strain is active against more than 55 species of lepidopteran pests. The relatively broad spectrum, for a bioinsecticide, of this strain is due to its expression of at least five genes which encode different 6-endotoxin polypeptides [5]. Three of these polypeptides are encoded by homologous genes designated co'IA(a), crylA(b) and crylA(c) [6]. The cry/A gene products form the bipyramidal crystals. The other insecticidal polypeptides of HD-1 are encoded by cryH genes. The co,H gene products form cuboidal inclusions which are found associated with the bipyramidal parasporal crystals of strain HD-1 [7]. The bipyramidal crystals are responsible for most of the activity of HD-1 6-endotoxin against lepidopteran larvae. The cuboidal inclusions are insecticidal to mosquito larvae but also show some activity against lepidopterans [6]. In the absence of denaturing agents, the cuboidal inclusions are insoluble below p H 10.

186 The

site of action of lepidopteran-specific

B.

thutingiensis toxins is the brush-border membrane of larval lepidopteran midgut cells [8]. The presence of high affinity receptors for B. thuringiensis toxins on this membrane has been shown to be a critical determinant of insect susceptibility [9-10]. However, the pH and proteinase content of larval midguts are also important determinants of larval susceptibility to B. thuringiensis 6-endotoxins [11]. The pH in the midgut of phytophagous lepidopteran larvae is generally alkaline. However, midgut pH varies from less than 8 to greater than 12 within and among larvae [12]. B. thuringiensis parasporal crystals retain full activity toward Bombyx mori larvae when stored in buffers ranging in pH between 2.2 and 10.8. However, the crystals dissolve only in alkaline buffers [13]. In order to better understand the structural basis for the pH-dependent solubility, proteolytic enzyme susceptibility and toxic properties of lepidopteran-specific B. thuringiensis 6-endotoxins, we have used circular dichroism (CD) spectroscopy to investigate the effects of pH on the secondary structure of both the protoxin and toxin from strain HD-I. Materials and Methods

Protoxin preparations Purified parasporal inclusions from the HD-1 strain of B. thuringiensis subspecies kurstaki were obtained from Syntro (San Diego, CA). In order to extract co,IA protoxin free from ctyll gene products and maintain a constant protein composition in all samples, parasporal crystals were suspended in a 0.05 M sodium carbonate/bicarbonate buffer (pH 9.5) containing 10 mM dithiothreitol (DTT). The suspensions were incubated for 1 h at 37°C. Undissolved material was removed by centrifugation at 12 800 × g for 3 rain. The pH of each aliquot (150 p_l) of the supernatant was adjusted to either 7.0, 8.0, 9.0, 10.0, 11.0 or 12.0 by dialysis overnight against two changes of 500 ml of 0.05 M sodium carbonate/bicarbonate buffer of the desired pH. 1 M NaOH was mLxed with a 0.05 M sodium carbonate solution to make a buffer of pH 12.0.

Toxin preparations, Protoxin solutions (1-2 m g / m l ) at each pH (as prepared above), were incubated separately with trypsin (0.5 m g / m l ) (Miles Laboratories) for 1 h at 37°C. The protoxin:trypsin molar ratio was 1:1.5. Trypsin was removed by washing the solutions in an Ultrafree-MC centrifugal filter unit with a 30000 MW cutoff (Millipore) at 12800 × g for 3 min. During centrifugation, the sample volume was reduced to 10/~1. 30 p.l of fresh buffer was added and the concentration step was repeated three times. The filtrate from the fourth wash was saved for spectroscopic baseline measurements.

Protein determination Protein concentrations were initially estimated using the method of Lowry et al. [14] for insoluble proteins, with bovine serum albumin as the standard. A scale factor (1.14) for protoxin samples, relating these concentrations to the absolute concentrations, was determined by quantitative amino-acid analysis. This factor was then applied to the values determined by the Lowry assay to produce the absolute concentrations of protoxin in each sample.

Polyacr3'lamide gel electrophoresis (SDS-PAGE) The polypeptide compositions of the various preparations were analyzed by polyacrylamide gel electrophoresis in the presence of 0.1% sodium dodecyl sulfate (SDS) [15], using a 4.5% acrylamide stacking and a 7.5% acrylamide separation gel. Intact parasporal crystals, solubilized protoxin samples or trypsin-digested protoxin samples were mixed with sample buffer, containing 20% glycerol, 10% 2-mercaptoethanol, 4% SDS, 0.002% Bromophenol blue and 125 mM Tris-HCl (pH 6.8) and incubated at 100°C for 5 rain. The gels were stained with Coomassie brilliant blue (Bio-Rad) (0.25~, w/v). The molecular weights were estimated from a plot of relative mobility versus the logarithm of the molecular weight, established with standard proteins: myosin,/3-galactosidase, phosphorylase b, bovine serum albumin and hen egg-white ovalbumin (Bio-Rad).

Circular dichroism spectroscopy Circular dichroism (CD) spectra were recorded on an Aviv 60DS spectropolarimeter. The instrument was calibrated with d-10-camphorsulfonic acid at both 290 nm and 192.5 rim. The calibration was checked for optical rotation using horse myoglobin and for wavelength using benzene vapor. CD measurements were routinely made at room temperature in 0.l-ram or 0.2-mm pathlength demountable quartz cells. Data from 300 nm to 190 nm were collected at 0.2 nm intervals. A minimum of five scans was collected for every sample. 4 or 5 independent preparations were examined for each sample type or pH condition. Absorption spectra (from 400 nm to 190 nm) were collected in the same demountab[e cell using a Cary 2200 spectrophotometer.

Data processing and analysis Multiple scans of each sample were averaged [16] and the averaged baseline spectrum of the dialysis buffer without protein was subtracted from the sample spectrum to yield the net spectrum of the protein. Based on the amino-acid sequence of this protein, a mean residue weight of 114.8 was calculated. Spectra of different samples prepared in identical manners were then averaged. The number of samples, n, listed in Tables I and II is the number of independent

187 preparations of each protoxin or toxin sample. Averaged scans were smoothed using a Savitzky-Golay filter [17]. Statistical analyses were performed to calculate the variation not only within a data set of five scans, but also between averaged data sets from different preparations. The former gives an estimate of the reproducibility of the instrument, while the latter shows variation between samples [16]. For each sample type, the standard deviation was computed at all wavelengths to determine whether significant differences existed between the averaged spectra of these types of sample preparations. Secondary structural analyses were performed using an unconstrained, normalized, linear least-squares fitting procedure [18-19]. The reference data set used in the analyses was derived from 15 water-soluble proteins [20]. An average helix length of 8 residues was used in the analyses. For each least-squares solution, a calculated spectrum, which represented the best fit of the reference data set to the experimental spectrum was generated. A fit parameter, the normalized root mean square deviation (NRMSD), which indicates how closely the calculated spectrum corresponds to the experimental spectrum, was determined [18]. The

abcdef

g h

i J kl

'm

t

NRMSD values reflect the extent of agreement between the calculated structure and the measured CD spectra. NRMSD of < 0.1 for soluble proteins usually indicates the calculated structure and X-ray structure are in excellent agreement. If 0.1 < NRMSD < 0.2, the calculated structure is characterized as having a secondary structure similar to the actual structure. If NRMSD > 0.2, the calculated structure generally does not resemble the actual structure [21]. Since most of the data (Tables I and II) yielded NRMSD values < 0.1, the calculated CD data may provide a reasonably accurate estimate of the secondary structure of these proteins. For soluble proteins, helical content calculated by this type of analysis has been shown to correlate strongly with the helical component as determined by X-ray crystallographic studies. The total /3structure (the sum of /3-sheet plus turn), while not as reliably determined, may be considered to be reasonably estimated. For these reasons, we focus our discussion of the experimentally determined secondary structure primarily on the a-helix content. Results

pH-dependent conformational changes of protoxin Intact &endotoxin crystals, when electrophoresed, produce a major band of apparent molecular weight approx. 135000 and a second minor band of MW approx. 55000 on SDS-PAGE (Fig. la). The minor band has been identified as cuboidal inclusion protein. Insoluble material peletted after the dissolution of the 6-endotoxin crystals at pH 9.5 in the presence of DTT, produced a single band of apparent molecular weight of approx. 55 000 (data not shown). Extracted protoxin samples consist of a single protein band with an apparent molecular weight of approx. 135000 (Fig. 1b-f). Toxin samples (prepared by proteolysis of protoxin at various pH values) showed a single band with an apparent molecular weight of approx. 68000 (Fig. 1g-l). 3-Endotoxin crystals were dissolved at pH 9.5 in the presence of 10 mM D T T and then changed to other pH values by dialysis against different pH buffers.

TABLE I

Calculated secondao, structures of protoxin at different pH ealues Abbreviations: NRMSD, Normalized root mean square deviation and n, number of independent preparations. Fig. 1. SDS-PAGE of 6-endotoxin samples. (a), Whole 6-endotoxm crystals; (b), Protoxin, pH 8; (c), Protoxin, pH 9; (d), Protoxin, pH 10; (e), Protoxin, pH 11; (f), Protoxin, pH 12; (g), Protoxin treated with trypsin at pH 7; (h), Protoxin treated with trypsin at pH 8; (i), Protoxin treated with trypsin at pH 9; (j), Protoxin treated with trypsin at pH 10; (k), Protoxin treated with trypsin at pH 11; (I), Protoxin treated with trypsin at pH 12; (m), Molecular weight standards: myosin, /3-galactosidase, phosphorylase b, bovine serum albumin and ovalbumin.

pH

% Helix

% 13-Structure

% Coil

NRMSD

n

7 8 9 10 11 12

31 +__4 28 +__5 33 + 2 26 +- 2 22 +- 3 15+__3

44 48 42 45 46 50

26 25 24 29 32 36

0.042 0.030 0.047 0.058 0.099 0.155

5 5 5 5 4 4

188 r

I

I

I

I

F--

L

2

LJ

12

1'

T,,=

~

-

0 (~

.

-22

I ]g5

/ ~.~J

]

I

~

I

I

215

I

235

I 255

WAVELENGTH

]

I 275

295

(rim)

), pH 10 ( . . . . . . ) and pH 12 ( - - - - - - ) . Significant differences above the error levels (indicated by Fig. 2. CD spectra of protoxin at pH 7 ( cross-bars) are observed. Analysis of these spectra yields a different secondary structure, indicating that pH affects the net conformation of the molecule.

SDS-PAGE showed all protoxin preparations in the pH range 7-12 to consist of a single band of apparent molecular weight approx. 135000 (Fig. lb-f). Therefore, any differences between samples of different pH must reflect protein conformational differences in the samples, not differences in polypeptide composition of the samples. No evidence of proteolysis was seen in any of these samples. The CD spectra of protoxin in the peptide (far UV) region at the two extremes of pH, 7 and 12, (Fig. 2) show profound differences in both shape and magnitude (mean residue ellipticity). These differences are considerably larger than the error levels and hence must represent a conformational change in the molecule. The spectrum of protoxin at pH 7, shows double minima at 210 and 223 nm, instead of the single minimum at 207 nm exhibited in the spectrum of protoxin at pH 12. The ellipticity at 198 nm in the spectra of protoxin at pH 7 and of protoxin at pH 12 differs in magnitude. Analysis of the CD spectrum at pH 7 indicates that the protoxin is composed of apTABLE II

Calculated secondary structures of toxin at different pH t'alues Abbreviations: NRMSD, Normalized root mean square deviation and n, number of independent preparations. pH

% Helix

%/3-Structure

% Coil

NRMSD

n

7 8 9 10 11 12

35 + 4 36 _+5 31-+5 27 _+5 32 -+ 2 22-+5

39 40 41 45 43 48

26 24 29 28 25 31

0.056 0.039 0.061 0.065 0.065 0.096

4 5 4 4 5 5

prox. 31% a-helix and approx. 44%/3-structure. At pH 12, the helix content is decreased to 15% and the sheet content increased to approx. 50% These differences are well above the error levels. The spectra at intermediate pH values can be compared to determine where this conformational change takes place. The spectra of protoxin at pH values 7, 8 and 9 are identical within error limits (not shown). All these samples display a maximum at 198 nm, a crossover at 200 nm and minima at 212 nm and 218 nm. They are also quite similar in shape and magnitude. The calculated secondary structures at these pH values are also the same, to within the error limits (Table I). Thus, at pH 7-9, the secondary structure of the protoxin essentially does not change. At pH 10, however, the spectrum begins to differ. There is a shift in the position of both minima to 208 nm and 221 nm, respectively (Fig. 2). Also, the magnitude of the spectrum decreases compared to that at pH 7, 8, or 9 (not shown). The shift in the position of both minima increases further for pH 11 and even more for pH 12. The differences between the spectra of protoxin at pH 10, 11 and 12 are larger than the associated reproducibility limits between preparations of the same type of sample. These changes in the spectra shape reflect a decrease in the calculated a-helix content by approx. 11% at pH 12 compared to pH 10 and a concomitant increase in the random coil structure to 36% at pH 12 compared to 29% at pH 10 (Table I). The differences between the spectra of protoxin at pH 11 and 12 are smaller than the differences between protoxin at pH 10 and 12. The decrease in the c~-helix content between pH 10 and 11 is approximately half of that between pH 11 and 12. This correlates with the observation that the

189 q

? ~5

I

I

l

I

I

T

×

H

d

t~

-5 D [E

w I

-15

-25

I

195

[

I

~

I

215

I

235

I

255

WAVELENGTH

Fig. 3. CD spectra of toxin at pH 7 ( . . . . . . ), and pH 12 (

I

l

275

I

I

295

(nm)

). Significantdifferencesare observed.

larvicidal activity of parasporal crystals decreases by 50% after storage at pH 11 and very little toxic activity remains after storage at pH 12 [13].

amount of a-helix than protoxin. At pH 12, the toxin molecule assumes a conformation that is approx. 22% a-helix and 48%/3-structure.

pH-depe~dent conformational changes of toxin

Discussion

In vivo, protoxin molecules are digested by enzymes in the insect gut to produce toxin molecules of approx. molecular weight 68 000. This process can be mimicked in vitro by treatment of protoxin with proteinases. Protoxin solutions of various pH values, prepared as described above were treated with trypsin to produce toxin. On SDS-PAGE (Fig. lg-1), a single band at approx. 68 000 was seen for all the toxin samples in the pH range 7-12. The efficiency of cleavage with the extreme enzyme-to-substrate ratio used was nearly 100%. These results suggest that the protoxin is folded in a manner such that its core region is relatively resistant to proteinase treatment at all the pH values tested. As all proteinase products appear identical, it is expected that all samples examined by CD spectroscopy will have the same polypeptide composition and any differences will reflect differences in conformation. The CD spectra of the toxin at the extremes, pH 7 and pH 12, differ by more than the error levels (Fig. 3). The differences, while significant, are substantially smaller than the differences manifested by the protoxin at these pH values. At pH 7, the secondary structure of the toxin is approx. 35% u-helical and approx. 39% /3-structure. Again, as was seen with protoxin, there is little effect of pH changes between pH 7 and 9 on either the spectra (not shown) or the calculated secondary structure (Table II). At higher pH values, the spectrum of the toxin changes in a manner similar to that of protoxin, indicating a decrease in the a-helix content of toxin. However, toxin generally has a greater

The goal of this study was to examine whether pH-dependent conformational changes were inherent to any steps in the conversion of /~-endotoxin crystals to active toxin. To accomplish this the pH-dependence of the conformation of protoxin, the pH-dependence of the site and extent of cleavage of protoxin and the pH-dependence of the conformation of toxin were tested using a combination of SDS-PAGE and CD spectroscopy. Although the net secondary structure of protoxin was nearly constant across the pH range 7-10, where the stability of protoxin is relatively invariant [13], a striking pH-dependence was seen for the conformation of protoxin at pH values where its activity decreases, suggesting the secondary structure is directly correlated with activity. A possible consequence of this result is that the conformational change could lead to an altered susceptibility to proteolytic cleavage. However, even under the exhaustive proteolysis conditions used in these studies, no apparent difference in the efficiency or site of cleavage could be detected. These results are in agreement with those of Choma and Kaplan [22] who found that the C-terminal portion of B. thuringiensis protoxin was more susceptible to denaturing agents than the N-terminal (toxin) portion of the protoxin. Finally, small but significant changes were also seen in the conformation of the toxin with pH. They were similar in nature (decrease in helix content) to that seen for the protoxin and could be responsible for the increased susceptibility of toxin to denaturing agents at high pH [23]. The observed pH-

190 dependent changes in secondary structure of both the protoxin and toxin conformations provide a structural basis for in vivo studies that suggest efficient production of toxic molecules will occur in organisms which have alkaline guts [3,11]. Previous circular dichroism studies [24] had detected effects of temperature and pH on the secondary and tertiary structure of the protoxin from B. thuringiensis subspecies alesti. In those studies, differences were reported in secondary and tertiary structures within the pH range 9-12. However, with no estimate of experimental error levels, it was not possible to determine if they were significant. The trend of decrease in the amount of a-helix seen by Gasparov et al. [24] is similar to what we have seen, but the same is not true for the amount of fi-structure. Gasparov et al. [24] found an overall decrease in the amount of/3-structure with increasing pH, whereas in our studies, the amount of fi-structure increased with increasing pH. Gasparov et al. [24] did not study toxin. Convents et al. [23] used CD spectra to study the unfolding of lepidopteran-specific toxin from the berliner 1715 strain of B. thmqngiensis. They recorded CD spectra, between 200 and 250 nm, of native toxin at pH 8 and pH 11. Although there are noticeable differences, particularly between 205 and 220 nm, in their published spectra, they say that 'within experimental error the CD spectra can be considered as identical'. These authors did not calculate the secondary structures from the spectra. Rather, they monitored the toxin's CD signal at 220 nm as a function of increasing guanidine hydrochloride concentration. They found that this signal decreased more rapidly at pH 11 than at pH 8. From these observations they concluded that the secondary structure of the protein responsible for this signal was less stable at pH 11 than at pH 8. This conclusion is consistent with our findings of a decrease in the a-helical secondary structure in HD-1 toxin above pH 9. Pozsgay et al. [25] used Raman spectroscopy to study the secondary structure of crystal proteins produced by both the HD-1 and HD-73 strains of B. thuringiensis subspecies kurstaki. Spectra were obtained from intact crystals pelleted in water, presumably at pH 7. Analyzing their spectra by the method of Lippert et al. [26] they estimated that the proteins of HD-I crystals contained approx. 25% a-helical and 21% /3-structure secondary structure. The spectra obtained from HD-73 crystals were very similar to those from HD-1 crystals. They estimated that the HD-73 crystal protein contained approx. 25% a-helical and 25%/3-structure secondary structure, Their estimate of a-helical and /3-structure secondary structure in HD-1 crystal proteins are slightly lower than those calculated in this study from CD spectra of HD-1 protoxin recorded at pH 7. Since the methods they used to

prepare HD-1 crystals for spectroscopy were unlikely to have removed the cuboidal inclusions, one might speculate that these differences in estimated secondary structure could be due to the presence of cuboidal inclusion proteins in their samples. However, this is unlikely to be a major reason for differences between studies. Analysis of Raman spectra from HD-73 crystals yielded estimates of secondary structure content similar to those for HD-1 crystals. Since strain HD-73 does not produce cuboidal inclusion proteins, differences in estimated secondary structure are more likely due to differences in spectroscopic methods. This conclusion is strengthened by the recent work of Chorea et al. [27], which estimated the secondary structure of HD-73 toxin, as derived from Raman, infrared and CD spectra. Their CD and IR spectra were recorded at pH 10.5. Raman spectra were recorded at pH 7, From Raman spectra they estimated only 20% a-helical secondary structure. Whereas, we calculated from CD spectra that HD-1 toxin contains 35% a-helical secondary structure at pH 7. Analysis of their IR and CD spectra yielded estimates of 40 and 34% a-helical secondary structure, respectively. Our estimates of a-helical secondary structure content from CD spectra of HD-1 toxin at pH 10 and 11 (Table II) are much closer to their estimates of a-helical secondary structure content from CD spectra of HD-73 toxin at pH 10.5 than to their estimates of a-helical secondary structure content of HD-73 toxin derived from IR spectra at pH 10.5 or Raman spectra at pH 7. In order to assess the accuracy of our analysis method for calculating secondary structure from CD data, the results from two different algorithms were compared. Similar calculated secondary structures were obtained using these two different algorithms: for example, for protoxin at pH 11, the linear least-squares method [18] produced 22% a-helix, 46% /3-structure and 32% random coil, while the singular value deconvolution method [28] produced 20.ok a-helix, 47% /3structure and 37% random coil. The correspondence of the results obtained with these two different analysis procedures increases our confidence in the final calculated structure. Choma et al. [27] estimated from the IR absorption band at 1672 cm l that HD-73 toxin contains 22-32% fl-structure at pH 10.5, while from CD spectra they estimated that HD-73 toxin contains 32% fl-structure at the same pH. Their CD-based estimate of fl-structure at pH 10.5 is 11% less than ours for HD-1 toxin at pH 11 (Table II). From Raman spectra Choma et al. [27] estimated that HD-73 toxin contains 35%/3-structure at pH 7. This value is close to the percentage of /3-structure that we calculated (39%) from CD spectra for HD-1 toxin at pH 7 (Table II). The small difference observed between these two values may be more indicative of the variations due to methodology than

191 Acknowledgements

TABLE III Predicted secondary structures of HD-1 6-endotoxin Using the method of Garnier et al. [30] with neutral decision constants.

crylA(a) crylA(b) crylA(c) co'lA(a) crylA(b) crylA(c) a b c d

protoxin a protoxin b protoxin c toxin d toxin toxin

% Helix

%/3-Structure

% Coil

24 24 24 10 8 8

52 51 53 58 58 61

24 25 23 32 34 31

crylA(a), protoxin sequence crylA(b), protoxin sequence crylA(c), protoxin sequence Toxin sequences = protoxin

from Schnepf et al. [31]. from Geiser et al. [32]. from Adang et al. [33]. sequences from 29 to 609.

variations in structure with pH. This conclusion is supported by the excellent agreement between the secondary structure of HD-73 toxin calculated from CD spectra by Choma et al. [27] and the secondary structure of HD-1 toxin calculated from CD spectra in this study. Recent studies indicate that the bipyramidal crystals of B. thuringiensis subspecies kurstaki strain HD-1 are formed by co-crystallization of the products of crylA(a), crylA(b) and crylA(c) 6-endotoxin genes [29]. The amino-acid sequences of the protoxin proteins encoded by these three genes are more than 80% identical [6]. The bipyramidal parasporal crystals of B. thuringiensis subsp, kurstaki strain HD-73 are composed of the product of only the crylA(c) a-endotoxin gene. As discussed above, the secondary structure contents of HD-1 toxin and HD-73 toxin calculated from CD spectra are essentially identical. Likewise, predictions of the secondary structure of the three protoxin proteins produced by strain HD-1 or the toxin portions of these proteins using the method of Garnier et al. [30] yield nearly identical results (Table III). However, the correlation between the predicted secondary structure content of these proteins and their secondary structure composition calculated from the CD spectra is rather weak. We find it especially interesting that whereas toxins are predicted to contain a lower percentage of a-helical secondary structure than protoxins, the CD spectra indicate that toxin molecules contain a slightly higher percentage of a-helical secondary structure than protoxin molecules. Perhaps these insecticidal bacterial proteins contain structural domains of unusual aminoacid sequence, the elucidation of which could contribute to our understanding of protein folding or because of their relatively hydrophobic nature, predictions based on a database comprised of soluble proteins may not be entirely accurate.

We thank M. Gawinowitz and S. Helpine of Columbia University for performing the quantitative amino-acid analyses. This work was supported in part by NIH Grants GM27292 (to BAW) and GM41766 (to MGW). References 1 Angus, T.A. (1954) Nature 173, 545-546. 2 Wilcox, D.R., Shivakumar, A.G., Malin, B.E.. Miller, M.F., Benson, T.A., Schoop, C.W., Casuto, D., Gundling, G.J.. Boiling, T.J., Spear, B.B. and Fox, J.L. (1986) Protein Engineering (Inouye, M. and Sama, R., eds.), pp. 395-413, Academic Press, New York. 3 Aronson, A.I., Beckman, W. and Dunn, P. (1987) Microbiol. Rev. 50, 1-24. 4 Huber, H.E. and Luethy, P. (1981) Pathogenesis of Invertebrate Microbial Diseases (Davidson. E.W., ed.), pp. 209-234, Allanheld, Totowa. 5 Widner, W.R. and Whiteley, H.R. (1989) J. Bacteriol. 171, 965974. 6 Hoefte, H. and Whiteley, H.R. (1989) Microbiol. Rev. 53, 242255. 7 Yamamoto, T. and Maclaughlin, R. (1981) Biochem. Biophys. Res. Commun. 103, 414-421. 8 Luethy, P., Jaquet, F., Hofmann, C., Huber-Lukac, M., and Wolfersberger, M.G. (1986)Zentralbl. Bakteriol. Mikrobiol. Hyg. I (Suppl. 15), 161-166. 9 Hofmann, C., Vanderbruggen, H., Hoefte. J., Van Rie, J., Jansens, S. and Van Mellaert, H. (1988) Proc. Natl. Acad. Sci. USA 85, 7844-7848. 10 Van Rie, J., Mcgauhey, W.H., Johnson, D.E., Barnett, B.D., and Van Mellaert, H. (1990) Science 247, 72-74. 11 Jacquet, F., Huetter, R. and Luethy, P. (1987) Appl. Environ. Microbiol. 53, 500-504. 12 Dow, J.A.T. (1986) Adv. Insect Physiol. 19, 187-328. 13 Nishiitsutsuji-Uwo, J., Ohsawa, A. and Nishimura, M.S. (1977) J. Invert. Pathol. 29, 162-169. 14 Lowry, O.H., Rosebrough, N.J., Farr, A.L. and Randall, R.J. (1951) J. Biol? Chem. 193, 265-275. 15 Laemmli. U.K. (1980) Nature 227. 680-685. 16 Mielke, D.L. and Wallace, B.A. (1988) J. Biol. Chem. 263, 31773182. 17 Savitzky, A. and Golay. M.J.E. (1964) Anal. Chem. (1964) 36, 1627-1639. 18 Mao, D. and Wallace, B.A. (1984) Biochemistry 23, 2667-2673. 19 Wallace, B.A. and Teeters, C.L. (1987) Biochemistry 26, 65-70. 20 Chang, C.T., Wu, C.S.C. and Yang, J.T. (1978) Anal. Biochem. 91, 13-31. 21 Brahms, S. and Brahms, J. (1980) J. Mol. Biol. 138, 149-178. 22 Choma, C.T. and Kaplan, H. (1990) Biochemistry 29, 1097110977. 23 Convents, D., Houssier, C., Lasters, I. and Lauwereys, M. (1990) J. Biol. Chem. 265, 1369-1375. 24 Gasparov, V.S.. Chestukhina, G.G., Bolotina, I.A.. Stepanov, V.M. and Permogorov, V.I. (1985) Mol. Biol. (Mosk) 19, 14221428. 25 Pozsgay, M., Fast, P., Kaplan, H. and Carey, P.R. (1987) J. Invert. Pathol. 50, 246-253. 26 Lippert, J.L., Tyminski, D. and Desmeules, P.J. (1976) J. Amer. Chem. Soc. 98, 7075-7080. 27 Choma, C.T., Surewicz, W.K., Carey, P.R., Pozsgay, M., and Kaplan, H. (1990) J. Protein Chem. 9. 87-94.

192 28 Hennessey, J.P., Jr. and Johnson, W.C., Jr. (1981) Biochemistry 20, 1085-1094. 29 Masson, L., Marcotte, P., Prefontaine, G. and Brousseau, R. (1989) Nucleic Acids Res. 17, 446. 30 Gamier, J., Osguthorpe, D.J. and Robson, B. (1978) J. Mol. Biol. 120, 97-120.

31 Schnepf, H.E., Wong, H.C. and Whiteley, H.R. (1985) J. Biol. Chem. 260, 6264-6272. 32 Geiser, M., Schweitzer, S., Grimm, C. (1986)Gene 48, 109-118. 33 Adang, M.J., Staver, M.J., Rocheleau, T.A., Leighton, J., Barker, R.F. and Thompson, D.V. (1985) Gene 36, 289-300.