M. J. Beekwilder, Plant Research International, Postbus 16, 6700 AA Wageningen, the Netherlands. Fax: + 31 317 418094, Tel.: + 31 317 477164, E-mail: M.J.Beekwilder@plant.wag-ur.nl
Pest insects such as Helicoverpa spp. frequently feed on plants expressing protease inhibitors. Apparently, their digestive system can adapt to the presence of protease inhibitors. To study this, a trypsin enzyme was purified from the gut of insects that were raised on an inhibitor-containing diet. The amino-acid sequence of this enzyme was analysed by tandem MS, which allowed assignment of 66% of the mature protein amino acid sequence. This trypsin, called HzTrypsin-S, corresponded to a known cDNA sequence from Helicoverpa. The amino acid sequence is closely related (76% identical) to that of a trypsin, HzTrypsin-C, which was purified and identified in a similar way from insects raised on a diet without additional inhibitor. The digestive properties of HzTrypsin-S and HzTrypsin-C were compared. Both trypsins appeared to be equally efficient in degrading protein. Four typical plant inhibitors were tested in enzymatic measurements. HzTrypsin-S could not be inhibited by > 1000-fold molar excess of any of these. The same inhibitors inhibited HzTrypsin-C with apparent equilibrium dissociation constants ranging from 1 nm to 30 nm. Thus, HzTrypsin-S seems to allow the insect to overcome different defensive proteinase inhibitors in plants.
Helicoverpa zea trypsin from insects raised on control diet
Helicoverpa zea trypsin from insects raised on SKTI-containing diet
Larvae of the lepidopteran insect species Helicoverpa are a pest in Asia, Australia and the Americas. They cause yield losses on many important crops, like cotton, chickpea, corn, and tomato. For instance, of the total cotton area in China (4.7 million hectares), 30% was lost to H. armigera in the mid nineties . Chemical control of Helicoverpa insects is often not effective, as they are notorious for development of resistance to chemicals such as DDT, organophosphates and pyrethroids .
One form of natural defence of plants against insects is mediated by protease inhibitors . The inhibitors are thought to have coevolved with insect herbivory, and to function by blocking the digestive proteases in the larval gut, thereby limiting the release of amino acids from food protein. As a consequence, the larvae are arrested in development, and eventually die. Genes encoding protease inhibitors have been used to produce resistant transgenic plants as a crop-protection strategy. This has met with initial success [4–6], but disappointing results have been reported for Helicoverpa spp., and a few other pest insects. Although several groups have shown that a major part of Helicoverpa gut protease activity can be blocked by a number of inhibitors [7,8], even the most efficient inhibitor (soybean Kunitz trypsin inhibitor, SKTI), which inhibits 95% of trypsin activity in gut extracts, does not affect the larval development of this insect on artificial diet  or transgenic plants .
The lack of effect on larval development is caused by the adaptation of Helicoverpa spp. to protease inhibitors, which is mediated by their ability to alter the complement of proteolytic activity in their gut. In response to inhibitor ingestion, the arsenal of gut proteinases is switched to enzymes that are insensitive to the plant inhibitors [10,11]. The adaptation of gut proteolysis to protease inhibitors is accompanied by changes in transcription of protease genes. A number of trypsin and chymotrypsin cDNA clones have been isolated from Helicoverpa mid-guts [9,12]. Reported gene expression data provide correlations to the changes in proteolysis in the insect gut. However, due to lack of a suitable expression system, the protease genes have not yet been linked to their function in terms of sensitivity to various inhibitors, substrate specificity or relative contribution to protein digestion.
In this report, for the first time enzymes directly involved in resistance to plant defence were purified from Helicoverpa gut. Individual enzymes were sequenced and their interaction with substrates and plant protease inhibitors analysed.
Materials and methods
H. zea eggs were purchased from French Agriculture (Lamberton, MN, USA), and hatched at 28 °C on artificial diet as described . The diet contains per litre: 160 g cornmeal, 80 g wheat germs, 80 g yeast flakes, 8 g ascorbic acid, 2 g sorbic acid, 1 g p-hydroxybenzoic acid, 0.1 g streptomycin and 30 g agar. Each individual first instar larva was sealed into a chamber containing 5 mL artificial diet. In the final stages of the fourth instar, 50 larvae that were about to molt were transferred to artificial diet supplemented with 0.5% (w/v) (250 µm) SKTI soybean trypsin inhibitor (type II-S, Sigma), while another 50 larvae remained on artificial diet without inhibitor. After 48 h, insects were chilled on ice, and guts with contents were excised, aliquoted and frozen at −80 °C. Frozen guts were thawed on ice, and mixed 1 : 3 with 50 mm Tris/HCl pH 8 containing 1% polyvinylpolypyrrolidone with 0.5 m NaCl, leading to about 10 mL per 50 guts. Guts were homogenized three times using an S541 potter tube at 60 r.p.m., and centrifuged for 15 min at 10 000 g, 4 °C to remove solid particles. The supernatant was filtered through a 0.22 µm filter.
Mustard trypsin inhibitor II (MTI-2) was produced in Pichia pastoris as described . Fifteen mg MTI-2 was coupled overnight at 4 °C to 1.5 g CNBr-activated sepharose 4B (Amersham Pharmacia Biotech AB) in 20 mL 1 mm HCl according to the manufacturer's instructions. The material was used to pour a 5-mL MTI-2 column, which was equilibrated with E-buffer (50 mm Tris/HCl pH 8, 0.5 m NaCl). Four mL gut content supernatant of H. zea (4 mg total protein) was loaded on the column, after which it was washed with 35 mL E-buffer and 35 mL E-buffer without salt. MTI-2-bound protein was eluted stepwise using 5 mL of G-buffer (0.1 m HCl/glycine) pH 3.2; 5 mL G-buffer pH 2.2; 5 mL G-buffer pH 1.5 and 5 mL G-buffer pH 1.5 + 20% dimethylsulfoxide. Eluted fractions were neutralized to pH 8 using 2 m Tris pH 10.5 and stored at +4 °C.
Protein fractions were normalized for protein content. Two µg protein were mixed with 150 µL assay buffer (25 mm glycine/NaOH pH 10; 0.1 mg·mL−1 BSA; 2.5 mm CaCl2). After incubation at 22 °C for 30 min, 50 µL of substrates in assay buffer containing 10% dimethylsulfoxide were added to a final concentration of 1 mm, and substrate breakdown was monitored at 405 nm. Substrates were Na-Benzoyl-l-Arg-p-nitroanilide (BApNA), Z-Arg-Arg-p-nitroanilide (ZRRpNA), Z-Phe-Arg-p-nitroanilide (ZFRpNA), Z-Arg-p-nitroanilide (ZRpNA) and l-Arg-p-nitroanilide (RpNA) for trypsin, and N-Succinyl-Ala-Ala-Pro-Leu-p-nitroanilide (SAAPLpNA) for chymotrypsin activity. BApNA and RpNA were from Sigma, SAAPLpNA, ZRRpNA, ZFRpNA and ZRpNA were from Bachem (Bubensdorf). Assays for inhibitor specificity were carried out in the presence of N-tosyl-l-lysine-chloromethyl ketone (TLCK), SKTI, Soybean Bowman–Birk Inhibitor (SBBI; Sigma), MTI-2  or potato trypsin/chymotrypsin inhibitor PI-2 . The concentration of reactive sites of these inhibitors was determined on bovine trypsin as described previously . Inhibitors were added to the protease and buffer, and preincubated for 30 min before adding the substrate.
Dietary protein breakdown was measured as follows: 1 g artificial diet was frozen, ground into fine powder with a mortar, and extracted with 10 mL acetone and six 10-mL portions of hexane. Soluble protein was extracted from the residual pellet by vigorous stirring with 10 mL water for 2 h at 4 °C. Insoluble matter was removed by centrifugation, and small soluble peptides by precipitation with 3 vols acetone. The final acetone pellet (360 µg protein) was dissolved in 1 mL buffer (50 mm glycine/NaOH, pH 10) to a clear solution. Aliquots of 100 µL soluble protein were mixed with 100 µL insect enzyme and 100 µL buffer and incubated at 37 °C. After 30 min 100 µL 40% trichloroacetic acid was added, incubated for 5 min at room temperature and centrifuged for 5 min. The supernatant was used to measure absorption of light at 280 nm. Absorption of soluble dietary protein that had not been incubated with enzyme, and enzyme that had not been incubated with soluble dietary protein were used as controls. Control values were constant over a number of experiments. Azocasein assays were performed as described previously .
SDS/PAGE and IEF
For SDS PAGE, 2.5 µg total gut proteins and 0.5 µg (15 µL) of the fractions were diluted with 5 µL sample buffer (20% glycerol, 20 mm Tris pH 6.8, 0.4% SDS, 0.001% Bromophenol blue), and kept on ice. Protein staining was performed by using silver nitrate . Activity staining with casein as described .
For IEF, 300 µL of the fractions (10 µg protein) were precipitated with 10% trichloroacetic acid, washed with ice-cold acetone and resuspended in 125 µL rehydration solution (8 m urea, 2 m thio-urea, 2% Chaps, 2 mm dithiothreitol, 2 mm EDTA). Immobilized pH gradient (IPG) buffer (0.5%, pH 6–11, Amersham) was added, mixed and the sample was allowed to enter an Immobiline DryStrip pH 6–11 (7 cm; Amersham) overnight. Focusing was performed for 6 h from 500 to 8000 V. The strip was subsequently equilibrated in a solution containing 1% dithiothreitol, 50 mm Tris/HCl pH 8.8, 6 m Urea, 30% glycerol, 2% SDS, and stained with Coomassie brilliant blue.
Protein bands were excised from the gel, dried, and digested in gel with trypsin . Protein was extracted, and loaded onto a C18 PepMap column (15 cm × 75 µm). Peptides were eluted by a 30-min gradient from 0.5% formic acid in water to 0.5% formic acid in 50% acetonitril at a speed of 0.2 µL·min−1. The C18 column was connected to the electro-electro-spray of a Q-Tof-2 Mass spectrometer (Micromass) by a PicoTip (New Objective). The Qtof mass spectrometer was instructed to determine charge of the eluting peptides, and, if appropriate (i.e. 2 + or 3 +), the QtofMS switched to the MS/MS mode applying collision-induced dissociation (CID). The resulting CID spectrum contains the sequence information for a single peptide.
The mass-lynx package V4.3 (MicroMass) was used to process MS data. First the maxent 3 module was used to deconvolute the data. MS/MS spectra containing CID products were selected for further processing. The BioLynx PepSeq module was used to interpret MS/MS spectra and to generate peptide sequences. The MS/MS spectra (usually around 25 per peptide) were further scrutinized manually by using the ManSeq mode.
MS results were compared to 34 database accessions with the following numbers: AF045138 (H. armigera trypsin), AF233731–AF233734 (H. zea chymotrypsins), AF261980–AF261989 (H. zea trypsins) and Y12269-Y12287 (H. armigera serine proteases).
H. zea larvae adapt to the presence of SKTI in the diet
To obtain gut proteases that were resistant to protease inhibitors, larvae were adapted to SKTI. Two populations of H. zea larvae were raised in parallel. One population of larvae was reared during its entire larval development on a control diet, consisting of corn materials. When these insects were in their fifth instar, guts were isolated. Trypsin activity of gut extracts was tested, and appeared to be 95% inhibitable by 0.5 µm SKTI. Guts from the other population of larvae were isolated at the same time, but the insects had been transferred to corn diet supplemented with SKTI 48 h previously. Extracts from this population have a completely different trypsin activity (BApNAse): only 2% is inhibitable by 0.5 µm SKTI. These results confirm those published for H. armigera and for H. zea[9,12].
Insect trypsins can be efficiently purified by affinity chromatography
The H. zea crude gut extracts were used for purification of the Trypsin-like enzymes by affinity chromatography. Mustard trypsin inhibitor MTI-2, a proteinaceous trypsin inhibitor, was cross-linked to Sepharose, and used as affinity ligand. MTI-2 is known to be a very potent trypsin inhibitor, but to have a low affinity for chymotrypsin . In a pilot experiment we established that both active trypsin and chymotrypsin (bovine) can be sequestered by this material, and can be separated by eluting at different pH values (data not shown). Apparently, the relatively low affinity of MTI-2 on this column for enzymes like chymotrypsin is still sufficient to isolate them. Fig. 1 shows the activity of the eluted fractions from the H. zea samples. Trypsins (Fig. 1, bold lines) and chymotrypsins (Fig. 1, grey dotted lines) were eluted from the MTI-2 column after stepwise lowering of the pH. In this paper we focus on trypsins.
The MTI-2 column concentrates tryptic activity (as measured by BApNA degradation) from H. zea guts on both diets in a limited number of fractions. In the case of the control guts, hardly any trypsin activity is detected in the flow-through and washes of the column (Fig. 1A). All eluted tryptic activity (≈ 40% of the input) is concentrated in fractions of pH 1.5 with 15% dimethylsulfoxide, while most chymotrypsin activity is released at pH 3.2 and 2.2. For the SKTI guts, no such harsh treatment was needed to recover all protease activity (Fig. 1B). Around 33% of the tryptic activity did not bind to the column. Treatment at pH 3.2 released 14% of trypsin (mixed with chymotrypsin activity), while 43% was eluted at pH 2.2 (with hardly any chymotrypsin activity). This latter fraction was not found to be contaminated with SKTI from the diet by the IEF and MS analysis (below). The control diet trypsin eluting at pH 1.5 with 15% dimethylsulfoxide is hereafter referred to as HzTrypsin-C, and the SKTI diet trypsin eluting at pH 2.2 as HzTrypsin-S.
Trypsin fractions are functionally pure
The purity of trypsin fractions was tested using activity gels, SDS/PAGE and IEF. Fig. 2C (first lane) shows that the control gut has four caseinolytic proteins (bands C1–C4). One of the major caseinolytic bands (C1, at ≈ 23 kDa) was concentrated by the affinity chromatography, and was highly pure as judged by silverstaining (Fig. 2A). Notably, band C1 was dominant both in the chymotrypsin fraction (pH 2.2), and in the HzTrypsin-C fraction (pH 1.5 + dimethylsulfoxide). Apparently the mobility of both trypsin and chymotrypsin in this semi-denaturing gel system is similar.
The SKTI diet gut contained five caseinolytic bands (S1–S5; Fig. 2D). Caseinolytic bands S2, S3 (both in chymotrypsin fractions; pH 3.2) and S4 (in the HzTrypsin-S fraction; pH 2.2) were concentrated by the affinity chromatography. They appeared as strong bands on the silver-stained gel (Fig. 2B). The chymotrypsin fraction contains some additional proteins that did not display proteolytic activity in the casein gel. The HzTrypsin-S fraction contains a single dominant band (at ≈ 26 kDa) in the silver-stained gel, which comigrates with the bottom of a smear of activity. Therefore, we conclude that both HzTrypsin-C and HzTrypsin-S are functionally pure, and contain no significant contaminant nonprotease protein.
Trypsin fractions were separated further on IEF gels (Fig. 2E). Denaturing SDS/PAGE, which is frequently conducted following IEF in two-dimensional gel systems, did not improve the separation, as all proteins in the fractions run at about the same molecular size (23.5 kDa). The second dimension gel was therefore omitted. HzTrypsin-C (Fig. 2E, top strip) shows three dominant bands and a fourth minor band: the major bands focus around pI 9 (M1, M2 and M3), and the minor around pI 7. HzTrypsin-S does not have the pI 7 band, and has three major species (M11, M12, M13) and a minor band in the pI 9–10 area, but at a different position than in the HzTrypsin-C (Fig. 2E, lower strip). IEF strips covering a different pH range did not show additional bands, and we did not observe any protein of a pI corresponding to SKTI (which is around 5 for different isoforms of SKTI).
Identity assignment of trypsins by MS
To link the isolated digestive enzymes to a protein and gene sequence, the major IEF bands from HzTrypsin-C and HzTrypsin-S were sequenced by MS. Bands M1, M2, M3, M11, M12 and M13 were excised from the focussing gel (Fig. 2E) and digested by bovine trypsin. Tryptic peptide fragments were analysed by MS and tandem MS. Masses of peptides were matched to the full Swissprot database, and automatically sequenced. For all IEF bands between two and five peptides were identified whose masses and peptide sequences related to one of the 29 available Helicoverpa proteinase genes (Table 1, bold figures). More information than the exactly matching peptides was sought, to identify the trypsins more accurately. By manual checking of MS/MS spectra, amino acid sequences for tryptic fragments were identified that almost completely matched the identified Helicoverpa proteases (Table 1, underlined).
Table 1. Fragments, amino acid sequences and observed and predicted masses from trypsin HzT15 in M11, M12 and M13 (upper panel) and HaY12269 in M1, M2 and M3 (lower panel). Bold underlined masses have a protein sequence and mass exactly as predicted. Underlined masses have an almost completely matching sequence after manual analysis, except for nonunderlined residues. Masses in italics have been found in MS spectra, but were not analysed by MS/MS to analyse their sequence.
a Predicted masses are tryptic fragments predicted by the program peptide mass.
HzTrypsin-S corresponds to HzT15 and HzTrypsin-C to HaY12269
Although HzTrypsin-S focuses at three different isoelectric points, it appears to relate to a single trypsin gene. IEF bands M11, M12 and M13 clearly relate to trypsin cDNA HzT15 (accession AF261980; Table 1, upper panel), and not to any of the other 33 Helicoverpa protease genes in the database. Nine out of 11 predicted tryptic fragments of HzT15 are found in the spectra of M11, M12 and M13 (Table 1, last column). In Fig. 3, amino acids that were identified by the MS and MS/MS analyses are underlined. The analysed sequences cover 66% of mature HzT15.
In addition, the HzTrypsin-C peptide sequences relate it to a single trypsin. All three major IEF bands (M1, M2 and M3; Fig. 2E) represent two Helicoverpa virtually identical trypsin genes, HaY12269 (accession Y12269; Table 1, lower panel) and Hz42. As the cDNA sequence Hz42 in the database is incomplete, we refer to the HaY12269 sequence. Seven out of 11 predicted tryptic fragments of HaY12269 are found in the spectra of M1, M2 and M3 (Fig. 3). The peptides with a sequence that matches HaY12269 cover 47% of the total mature sequence of HaY12269. The covered percentage is lower than for HzT15 (66%), as fewer high-mass peptides were analysed. No peptides of M1 and M2 could be identified that relate to any other protease gene, while M3 contains some additional peptides that match a H. zea chymotrypsin, HaY12273 (not shown). HzTrypsin-C apparently contains a few minor contaminations, such as the HaY12273 chymotrypsin. In addition, the minor band that focuses at pI 7 is not a trypsin (data not shown).
Manual scrutiny of MS/MS spectra revealed some additional data. Many peptides have peptide sequences that almost match HzT15 or HaY12269, but have a mass that is different from the predicted tryptic fragments (Table 1, top panel fragments 2, 3, 7 and 9; lower panel fragments 2, 6 and 8). In M11 and M12, we identified a modified amino acid at the position of Arg62 (porcine trypsin numbering, Fig. 3) of the HzT15 cDNA sequence. In the peptides with a matching sequence here, a larger mass difference (234 and 280 Da) than accounted for by Arg (156 Da) is found (Table 1, top panel fragment 3). The molecular weight of 280 Da corresponds to that of Arg-pyrimidine, a methylglyoxal modification of Arg . In other fragments, a larger mass difference than would be predicted is found at the position of the Cys residues in the cDNA. We assume that such differences arise from incomplete reduction and modification of the Cys residues, after recovery from the IEF strip. Apart from the Cys-based artifacts, all identified peptides exactly match the sequence of the cDNAs. Only one position, the Trp in the predicted peptide 2 of HzT15 (Table 1, top panel fragment 2) is not identified in HzTrypsin-S, but instead an amino acid of the same mass as Leu and Ile is found (Fig. 4). Such an amino acid is indeed found in all available Helicoverpa protease cDNAs, except HzT15, indicating a possible cDNA sequence determination artifact or genetic variation.
It was quite unexpected to identify only a single protease in three IEF bands that focus on clearly distinct pH values, both for M1, M2 and M3 and M11, M12 and M13. We would have expected some of the other Helicoverpa trypsin genes to be represented. This was investigated more closely by comparing single-dimension MS spectra of M1, M2, M3, M11, M12 and M13. Hardly any difference between the spectra of M1, M2 and M3, or difference between M11, M12 and M13 was observed. Possibly there are minor differences in single amino acids in fragments not tested by the MS (e.g. fragments 1, 7 and 9 in Table 1, lower panel), or differences in protein modifications (either natural or artefacts of sample preparation), but the available protein and cDNA data do not provide an explanation for this phenomenon. Therefore it is concluded that both HzTrypsin-C and HzTrypsin-S have unexpected purity: all major protein bands of a fraction have the same peptide sequence.
Sequence comparison of sensitive and resistant trypsins
To obtain some insight into the differences between HzTrypsin-C and HzTrypsin-S, the amino acid sequences of HaY12269 (HzTrypsin-C) and HzT15 (HzTrypsin-S) were aligned to each other and to porcine trypsin using standard methods (Fig. 3). The mature HaY12269 and HzT15 amino acid sequences are 76% identical. Five regions can be identified in which differences between HzT15 and HaY12269 are concentrated, but where also both Helicoverpa trypsins differ most from porcine trypsin (Fig. 3, indicated by X). These regions are annotated 37, 60, 99, 145 and 175, with reference to their position in the sequence of porcine trypsin according to the chymotrypsinogen numbering. Remarkably, all five regions differing between HzT15 and HaY12269 overlap with contact residues of the enzyme with SKTI . Notably, in loops 60 and 99, one additional amino acid is present in HzT15. Four out of five regions are covered by the MS analysis of HzTrypsin-S.
Trypsin activity on protease substrates
The apparent purity of the HzTrypsin-C and HzTrypsin-S fractions provided the opportunity to compare specific enzymatic activities of the two proteins. First, TLCK was used as inhibitor to compare the number of active sites. The concentration of TLCK needed per µg protein to inhibit trypsin activity was found to be comparable for control trypsin and SKTI trypsin. To characterize proteolytic activity, breakdown of four substrates was compared. The soluble protein from the corn-based insect diet was equally well cleaved by HzTrypsin-C and HzTrypsin-S (the activity ratio S/C was 98%; Table 2). Also the activity towards azocasein, ZRRpNA, ZFRpNA, ZRpNA and RpNA was similar for both enzymes (ratios S/C were 72%, 124%, 135%, 76% and 106%; Table 2). Remarkably, the substrate BApNA differentiates clearly between the trypsins. Breakdown of this substrate by HzTrypsin-S is less efficient (11%; Table 2) than by the control trypsin. BApNA differs from the other substrates in the residue that binds the S2 substrate-binding pocket on the surface of trypsin. These observations suggest that the HzTrypsin-S can work efficiently when a natural amino -acid is in the P2 position (as in ZFRpNA and ZRRpNA, and in proteins). HzTrypsin-S is much less efficient with the N-substituted benzoyl group carried by BApNA at that position, which is clearly less flexible at the Cα position of the P2 residue than an amino acid.
Table 2. Specific activity of gut trypsin fractions towards different substrates.
Ratio S/C (%)
Activity is assayed in 200 µL using 2 µg trypsin protein of each fraction, and is expressed as pNA release (change in absorption at 405 nm) per minute.
Activity is assayed as absorbance at 280 nm per µg trypsin protein after 30 min incubation and trichloroacetic acid precipitation.
Activity is assayed as absorbance at 340 nm per µg trypsin protein after 30 min incubation and trichloroacetic acid precipitation.
Inhibition of Helicoverpa trypsins by four plant protease inhibitors
The effect of plant protease inhibitors on the isolated trypsins was tested. We anticipated that inhibitors with a different architecture would have different inhibitory properties towards the insect proteases. For that purpose, inhibitors SKTI (representative of the Kunitz family , occurring in most plant species), SBBI (of the Bowman–Birk family , primarily present in legumes), PI-2 (of the potato inhibitor II family , only found in solanacaeae) and MTI-2 (of the mustard inhibitor family, only found in cruciferae) were chosen. To quantify inhibition, low concentrations of enzyme were mixed with calibrated concentrations of inhibitors. HzTrypsin-C and HzTrypsin-S were taken to be pure, which is 80% accurate as judged by the MS analysis. Equal amounts of protein from different protease fractions were mixed with a range of concentrations of each of the protease inhibitors, and residual activities to degrade substrate ZRRpNA were measured (Fig. 4).
HzTrypsin-C was inhibited strongly by SKTI and PI-2. This allowed titration of the concentration of active sites of this enzyme to be 5 nm. This value corresponds to the measured protein concentration, and the apparent equilibrium dissociation constant Ki of SKTI and PI-2 to this enzyme at 1 nm. MTI-2 and SBBI are needed in higher concentration (around 0.1 µm; 20-fold molar excess) to achieve full inhibition of the HzTrypsin-C (Fig. 4). For medium-affinity interactions like this, the 50% inhibitory concentration (IC50) roughly corresponds with the Ki. Hence, we calculate the Ki of MTI-2 and SBBI towards HzTrypsin-C to be 30 nm. The inhibition constants for SKTI and SBBI towards HzTrypsin-C are in the same range as those reported by Johnston et al..
The HzTrypsin-S can hardly be inhibited by any of the concentrations of inhibitor tested. At 10 µm inhibitor, SKTI and PI-2 confer about 50% reduction in activity, while MTI-2 and SBBI still have almost no effect on HzTrypsin-S activity (Fig. 4). The concentration of trypsin active sites per µg protein is similar to that of HzTrypsin-C, as indicated by the TLCK inhibition (see above). Therefore the HzTrypsin-S concentration in the assay was ≈ 5 nm. This means that at least a 2000-fold molar excess of either of the inhibitors is insufficient to inhibit this enzyme. The Ki of all four inhibitors towards HzTrypsin-S is therefore > 1000 nm.
Linking enzymatic properties to H. zea trypsin genes
The aim of this paper was to characterize trypsins involved in the coevolution of plant protease inhibitors and insect digestive proteases. The plant side of this coevolution has been well characterized: inhibitor genes have been shown to be induced by wounding, insect feeding and defence-signalling hormones , and also biochemical properties of isolated or recombinantly expressed inhibitors, and their effect on proteolysis in the insect gut have been extensively studied (e.g. [14,15]). On the other hand, the insect side of the coevolution (i.e. adaptation to plant defensive inhibitors), was first noted 7 years ago [10,11], but not much progress has been made towards understanding the biochemical properties of the proteases since then. This is mainly due to lack of an appropriate recombinant expression system.
Regulation of Helicoverpa protease genes upon inhibitor ingestion has been studied [9,12]. Helicoverpa responds to plant protease inhibitors with an intricate change in the expression of protease genes. Among five H. zea trypsin genes tested, three are up-regulated, among which is HzT15, and two are slightly down-regulated, among which is Hz42 (which is 98% identical to HaY12269) in response to SKTI . Apparently the adaptation does not involve a single up-regulated protease gene. Therefore, it has been difficult to establish a conclusive link between gene expression and gene function. One may presume that at least a subset of the up-regulated genes represent the inhibitor-insensitive trypsins in the adapted gut, and, vice versa, a subset of the slightly down-regulated genes represent the inhibitor-sensitive, nonadapted protease species. However, this is obscured by a number of factors, including the presence of the inhibitor in the gut. While transcripts by which inhibitor-sensitive trypsin is encoded (HaY12269 and Hz42) have been reported to be still quite abundant under these circumstances [9,12], we could not isolate such activity from the inhibitor-adapted gut. Most likely, the HzTrypsin-C is tightly bound by SKTI from the diet, and therefore does not contribute to proteolytic activity in the gut. Hence, interpretation of gene expression data towards a physiological model is confused by the fact that not all expressed proteases are active. To fully appreciate what happens when the insect gut adapts to inhibitors, a link to enzymological data by protein identification is required.
Usually the link between gene and function is established by recombinant expression of proteases. So far, insect serine proteases could not be expressed as active enzymes in a variety of hosts tested (Escherichia coli, yeast, insect cells; unpublished data), so no definite assignment of these trypsin genes to their function has been possible. A few lepidopteran proteases have been purified, but have not been analysed and compared with respect to adaptation to plant protease inhibitors [7,23,24]. The link to corresponding trypsin genes was made by limited N-terminal sequencing, which can be quite inaccurate for a highly homologous gene family such as insect trypsins. Modern protein biochemistry provides a number of very sensitive tools (collectively referred to as proteomics) to isolate, identify and characterize proteins in a secreted body fluid, such as the content of a gut. In the given case of the Helicoverpa trypsins, these tools substituted for the use of heterologous expression and established reliable links between sequence and function.
Inhibition and physiological role of trypsins
HzTrypsin-C is one of the major trypsins (> 40%) that H. zea deploys to digest plant material without inhibitors. This trypsin could strongly be inhibited by SKTI and PI-2 (Ki = 1 nm), but less strongly by SBBI and MTI-2 (Ki = 0.03 ìM). All four inhibitors are probably effective against HzTrypsin-C at physiological concentrations, because both gut enzymes and plant inhibitors occur at approximately 10 µm concentration in the insect gut. It was calculated that 10 µm of inhibitor with Ki = 0.1 µm is able to inhibit > 90% of the activity of 10µm trypsin . To overcome the loss of protease activity due to the dietary inhibitors, a novel trypsin, HzTrypsin-S, is synthesized, which is highly insensitive to all plant inhibitors tested.
It is puzzling why these insects do not constitutively express the inhibitor-insensitive trypsin genes, but instead perform an induced, time and energy-consuming change in gene expression of protease genes. Synthesis of proteases is an important metabolic activity of the gut cells, as protease mRNAs make up ≈ 20% of gut cDNAs . One suggested answer was that there may be no such thing as a protease insensitive to all types of inhibitors encountered by a polyphagous insect, so that flexibility of regulation would be an asset allowing appropriate subsets of genes to be expressed depending on the host plant . Now it appears that a trypsin insensitive to a very wide range of plant protease inhibitors does exist (HzTrypsin-S). Remarkably, the advantageous property is not compromised by a lower efficiency in plant protein degradation compared to a sensitive enzyme like HzTrypsin-C. This is in keeping with the observation that dietary inhibitors do not affect larval growth rates of Helicoverpa[9,12]. The question why PI-insensitive proteases are not constitutively expressed remains unanswered. The protease properties may affect other fitness parameters (e.g. progeny numbers) which may only become obvious in complex ecological circumstances (e.g. direct competition) that have not been tested.
What determines insensitivity to plant protease inhibitors?
HzTrypsin-S is fully adapted to the defensive protease inhibitors of plants. The adaptation must have biophysical and protein-structural reasons. Firstly, those reasons may be revealed by analysing amino acid differences between HzTrypsin-C and HzTrypsin-S. Extensive hypotheses based on sequence comparison have been formulated . Others have concluded that the Helicoverpa protease sequences do not contain an obvious clue to the mechanism of resistance to inhibitors . There are 57 differing amino acids between the two trypsins described in this paper (Fig. 3). In Fig. 5A, these residues are shown in yellow, superimposed on the porcine trypsin crystal structure. It can be clearly seen that these amino acids preferentially map in loops of the porcine trypsin structure that border the active site groove. In Fig. 5B, the differing residues are combined with those that are in contact with inhibitors (Fig. 5B, blue). There is clearly overlap (Fig. 5B, green) between contact residues and residues that differ between the two H. zea trypsins. Differing residues seem to form a ring around the active site of the trypsins, rather than affecting the active site itself. However, as all contact-loops are affected by multiple mutations, it is difficult to estimate the importance of individual regions.
Secondly, the inhibition data may serve to assign function to regions of the trypsin. The four inhibitors tested differ in their contact residues with surface loops of porcine trypsin (or homologous enzymes). Generally the contact loops of Trypsin-like enzymes are referred to as the 37, 60, 99, 145 and 170 loops  (Figs 3 and 5B, green and blue). SKTI has very little contact with the 175 loop, and very extensive contacts with the 99 loop , while PI-2 has hardly any contacts to the 60 loop, 99 loop and 145 loop , and SBBI has very few contacts with the 37, 60, 145 and 175 loops . Because HzTrypsin-S is resistant to inhibitors MTI-2, PI-2, SKTI and SBBI, one may conclude either that there is no single feature that impairs inhibitor binding, or that such a feature is close to the active site, where all inhibitors bind.
Thirdly, clues to the mechanism of adaptation of HzTrypsin-S may be inferred from the difference in substrate specificity of both trypsins (Table 2). The chemical substrate BApNA clearly distinguishes between the two enzymes. This substrate, which carries a benzoyl group at the P2 position, is degraded by HzTrypsin-S relatively poorly, as compared with HzTrypsin-C, whereas substrates like ZRRpNA, ZFRpNA, azocasein and dietary plant protein, which carry an amino acid at the P2 position, and RpNA, without a P2 residue, do not distinguish the enzymes. Apparently, the S2 pocket of HzTrypsin-S (Fig. 5B) is functionally different from that of HzTrypsin-C, resulting in reduced accommodation of, e.g. the benzoyl group carried by BApNA. The same difference may possibly be at the root of the ability of HzTrypsin-S to avoid inhibitor binding.
Properties of the purified proteins as reported here are essential to our understanding of the way successful insects deal with plant defence. For full protein structural understanding, regions differing between HzTrypsin-S and HzTrypsin-C that may contribute to occlusion of inhibitors should be addressed by a series of targeted mutations and recombinant expression. However, to our knowledge no suitable expression system has as yet been identified for insect trypsins, despite extensive efforts. Alternatively, crystal structures of the complexes of purified enzymes with MTI-2 could help to narrow further the structural determinants of insensitivity to protease inhibitors. Such information would provide valuable insight into the molecular basis of the adaptations of generalist pests. Also, it may lead to design of novel, ‘improved’ inhibitors. It will be a challenge to use the purified enzymes now obtained to improve inhibitors through methods such as phage display and rational design [18,28–30]. Similarly, ecologists will find challenges in determining the true costs and benefits of the deployment of these enzymes for these insects.
This research was conducted as part of an EU RTD project. M. V. was supported by EMBO fellowship ASTF 9601. We are grateful to B. Oliva for helpful advise, and R. de Maagd, R. Bino, D. Reverter and J. van Loon for careful reading of the manuscript.