Eur. J. Biochem. 261, 275±284 (1999) q FEBS 1999

Characterization of low-molecular-mass trypsin isoinhibitors from oil-rape(Brassica napus var. oleifera) seed

Paolo Ascenzi1, Margherita Ruoppolo2, Angela Amoresano3,*, Piero Pucci3,², Roberto Consonni4, Lucia Zetta4,

Stefano Pascarella5, Fabrizio Bortolotti6 and Enea Menegatti6

1Dipartimento di Biologia, UniversitaÁ di Roma Tre, Italy; 2Dipartimento di Chimica, UniversitaÁ di Salerno, Italy; 3Centro Internazionale di Servizi di

Spettrometria di Massa, CNR-UniversitaÁ di Napoli Federico II, Italy; 4Istituto di Chimica delle Macromolecole, Laboratorio NMR, Milano, Italy;5Dipartimento di Scienze Biochimiche Alessandro Rossi Fanelli, UniversitaÁ di Roma La Sapienza, Italy; 6Dipartimento di Scienze Farmaceutiche,

UniversitaÁ di Ferrara, Italy

A new low-molecular-mass (6767.8 Da) serine proteinase isoinhibitor has been isolated from oil-rape (Brassica

napus var. oleifera) seed, designated 5-oxoPro1-Gly62-RTI-III. The 5-oxoPro1-Gly62-RTI-III isoinhibitor is longer

than the Asp2-Pro61-RTI-III and the Ser3-Pro61-RTI-III forms, all the other amino acid residues being identical. In

RTI-III isoinhibitors, the P1-P10 reactive site bond (where residues forming the reactive site have been identified as

Pn¼P1 and P10¼Pn

0, where P1-P10 is the inhibitor scissile bond) has been identified at position Arg21-Ile22. The

inhibitor disulphide bridges pattern has been determined as Cys5-Cys27, Cys18-Cys31, Cys42-Cys52 and Cys54-

Cys57. The disulphide bridge arrangement observed in the RTI-III isoinhibitors is reminiscent of that found in a

number of toxins (e.g. erabutoxin b). Moreover, the organization of the three disulphide bridges subset Cys5-Cys27,

Cys18-Cys31 and Cys42-Cys52 is reminiscent of that found in epidermal growth factor domains. Preliminary1H-NMR data indicates the presence of aaNOEs and 3JaNH coupling constants, typical of the b-structure(s). These

data suggest that the three-dimensional structure of the RTI-III isoinhibitors may be reminiscent of that of toxins and

epidermal growth factor domains, consisting of three-finger shaped loops extending from the crossover region.

Values of the apparent association equilibrium constant for RTI-III isoinhibitors binding to bovine b-trypsin and

bovine a-chymotrypsin are 3.3 £ 109 m21 and 2.4 £ 106 m21, respectively, at pH 8.0 and 21.0 8C. The serine

proteinase : inhibitor complex formation is a pH-dependent entropy-driven process. RTI-III isoinhibitors do not

show any similarity to other serine proteinase inhibitors except the low molecular mass white mustard trypsin

isoinhibitor, isolated from Sinapis alba L. seed (MTI-2). Therefore, RTI-III and MTI-2 isoinhibitors could be

members of a new class of plant serine proteinase inhibitors.

Keywords: trypsin inhibitors; oil-rape seed; Brassica napus var. oleifera; amino acid sequence determination;

disulphide bridge location; 1H-NMR investigation; inhibitory properties.

Serine proteinase inhibitors are widespread in the plant kingdombeing found particularly in seeds of Cruciferae, Graminaceaeand Leguminosae, as well as in tubers of Solanaceae. Theirphysiological roles include the regulation of endogenous protein-ases during seed dormancy, the reserve protein mobilization, and

the protection against the proteolytic enzymes of parasites andinsects. Moreover, they may also act as storage or reserveproteins. Plant serine proteinase inhibitors are grouped intoSoybean (Kunitz), Bowman-Birk, potato I and II, and squashfamilies. Several other inhibitor families, such as barley, ragi 1and 2, and thaumatin, were also suggested [1±10].

Different types of serine proteinase inhibitors have beenidentified in the same plant, suggesting that these proteins haveevolved separately to perform distinct physiological roles. In thisrespect, oil-rape (Brassica napus var. oleifera) seed provides aninteresting model for investigating these different functions[2,4]. In fact, in the oil-rape seed, three distinct classes ofinhibitors (RTI-I, RTI-II and RTI-III, representing 4±10%,13±33% and 60±85% of the total serine proteinase inhibitoryactivity, respectively) have been identified [4]. RTI-I and RTI-IIbelong to the soybean family, are thermolabile, show a mol-ecular mass of about 19 kDa, and selectively inactivate trypsin[4]. On the other hand, RTI-III is the prototype of a new familyof serine proteinase inhibitors, is thermostable, shows a mol-ecular mass of 6.5 kDa, and inactivates both trypsin andchymotrypsin [2]. The complete amino acid sequence of theAsp2-Pro61-RTI-III and Ser3-Pro61-RTI-III isoinhibitors hasbeen determined. Both RTI-III isoinhibitors contain eight cysteineresidues at positions 5, 18, 27, 31, 42, 52, 54 and 57, forming

Correspondence to P. Ascenzi, Dipartimento di Biologia, UniversitaÁ di Roma

Tre, Viale G. Marconi 446, I-00146 Roma, Italy. Fax: +39-6-5517-6321,

Tel.: +39-6-5517-6329, E-mail: ascenzi@bio.uniroma3.it

Abbreviations: trypsin, bovine b-trypsin; chymotrypsin, bovine

a-chymotrypsin; ESMS, electrospray mass spectrometry; MALDIMS,

matrix assisted laser desorption ionisation mass spectrometry;

RTI-I±RTI-III, serine proteinase isoinhibitors I±III from oil-rape seed

Brassica napus var. oleifera; MTI-1 and MTI-2, serine proteinase

isoinhibitors 1 and 2 from white mustard Sinapis alba L.

Enzymes: bovine a-chymotrypsin (EC3.4.21.1); bovine b-trypsin

(EC3.4.21.4); endoproteinase Asp-N (EC3.4.24.33); endoproteinase Lys-C

(EC3.4.21.50); thermolysin (EC3.4.24.27).

*Present address: Centro Ingegneria Genetica, Biotecnologie Avanzate, scrl,

Via Pansini 5, I-80131, Napoli, Italy.²Present address: Dipartimento di Chimica Organica e Biologica, UniversitaÁ

di Napoli Federico II, Via Mezzocannone 16, I-80134, Napoli, Italy.

(Received 2 November 1998, revised 12 January 1999, accepted 26 January

1999)

276 P. Ascenzi et al. (Eur. J. Biochem. 261) q FEBS 1999

four disulphide bridges. The reactive site loop of the RTI-IIIisoinhibitors encompasses the Cys18-Cys27 amino acid sequence.The Arg21 and Ile22 side chains correspond to the P1 and P1

0

inhibitor reactive site residues, interacting with the S1 and S10

subsites of trypsin and chymotrypsin (i.e. form the potentiallyscissile P1-P1

0 inhibitor peptide bond) [2].Serine proteinase isoinhibitors almost identical to RTI-III,

from both the functional and structural viewpoints, have beenpurified from white mustard (Sinapis alba L.) seed (MTI-2) [1].The gene coding for MTI-2 isoinhibitors is under developmentaland environmental control, being expressed in immature seedsas well as in wounded leaves [3]. From the biotechnologicalviewpoint, MTI-2 isoinhibitors appear to be very interesting. Infact, the high expression level of MTI-2 isoinhibitors in trans-genic tobacco and Arabidopsis plants has deleterious effects oninstar larvae of the plant pathogen Spodoptera littoralis Bois-duval (Lepidoptera: Noctuidae), causing mortality and decreas-ing mean larval weight. Moreover, the high expression level ofMTI-2 isoinhibitor expression correlates with the decreased leafsurface eaten, thus providing effective protection against thispest [3,11].

The present study reports the detailed characterisation of anew low molecular mass serine proteinase inhibitor from oil-rape seed (5-oxoPro1-Gly62-RTI-III isoinhibitor). In particular,the amino acid sequence and the location of the disulphidebridges have been determined, preliminary 1H-NMR analysishas been performed, and the inhibitory properties have beeninvestigated. Present results have been analysed in parallel withthose of related systems. The common features of MTI-2 andRTI-III isoinhibitors strongly suggest that Cruciferae containsproteins which could be members of a new class of plant serineproteinase inhibitors [1,2,4].

MATERIALS

The 5-oxoPro1-Gly62-RTI-III isoinhibitor was prepurified froma commercial variety (cv Anima) of oil-rape (Brassica napusvar. oleifera) seed (from SV Semundo, Busseto, PN, Italy),according to Ceciliani et al. [2]. The 5-oxoPro1-Gly62-RTI-IIIisoinhibitor was further purified by HPLC, using a Vydac218TP54 reversed-phase C18 column (0.46 £ 25 cm) from TheSeparations Group (Hesperia, CA, USA). The elution systemconsisted of 0.1% trifluoroacetic acid (solvent A) and 0.07%trifluoroacetic acid in 95% CH3CN (solvent B). A lineargradient of solvent B from 5% to 95% in 30 min at flow rate of1 mL´min21 was employed. The UV absorbance of the eluentwas monitored at 220 nm. The purity of the protein preparationwas tested by electrospray mass spectrometry (ESMS) analysis,matrix-assisted laser desorption ionization mass spectrometry(MALDIMS) analysis, and N-terminus determination [12±16].The concentration of the 5-oxoPro1-Gly62-RTI-III isoinhibitorwas determined from the amino acid composition, and account-ing for the serine proteinase : inhibitor 1 : 1 stoichiometry,under conditions where trypsin concentration exceeds the valueof the dissociation equilibrium constant for the binary complexformation (Ka

21; see Eqn 1) [2].Bovine b-trypsin (hereafter referred to as trypsin), treated

with N-a-tosyl-l-phenylalanine chloromethyl ketone in order toabolish chymotryptic activity, was purified from commercialenzyme preparations (Sigma), according to Luthy et al. [17].Bovine a-chymotrypsin (hereafter referred to as chymotrypsin),treated with N-a-tosyl-l-lysine chloromethyl ketone in order toabolish tryptic activity, endoproteinase Asp-N, endoproteinaseLys-C, thermolysin, bovine insulin, horse heart myoglobin, N-a-benzoyl-l-arginine p-nitroanilide, N-a-carbobenzoxy-l-tyrosine

p-nitrophenyl ester, trifluoroacetic acid, Bistris, Mes, and Triswere purchased from Sigma. All the other products were fromMerck AG. All chemicals were of analytical grade, and wereused without further purification.

METHODS

Amino acid sequence determination of the5-oxoPro1-Gly62-RTI-III isoinhibitor

The 5-oxoPro1-Gly62-RTI-III isoinhibitor was reduced, denatu-rated and alkylated, as previously reported [2], and analysed byESMS [13]. The alkylated 5-oxoPro1-Gly62-RTI-III isoinhibitorwas digested with endoproteinase Lys-C and endoproteinaseAsp-N in 5.0 £ 1023 m ammonium bicarbonate, at pH 8.5 and37.0 8C, for 18 h. Endoproteinase Asp-N was activated using10% CH3CN. The enzyme : substrate ratio was 1 : 100 (w/w).The alkylated 5-oxoPro1-Gly62-RTI-III isoinhibitor peptidemixtures were directly analysed by MALDIMS following themass mapping procedures [15,16]. The assignment of masssignals was confirmed by a single step of manual Edmandegradation [14].

Cysteine bridge location in the 5-oxoPro1-Gly62-RTI-IIIisoinhibitor

The native 5-oxoPro1-Gly62-RTI-III isoinhibitor was digestedwith thermolysin in 0.4% ammonium bicarbonate, at pH 8.5 and37.0 8C, for 18 h. The enzyme : inhibitor ratio was 1 : 100 (w/w).Then, the thermolytic peptide mixture was digested by trypsin in0.4% ammonium bicarbonate, at pH 8.5 and 37.0 8C, for 4 h.The enzyme : substrate ratio was 1 : 50 (w/w). The resultingpeptide mixtures were directly analysed by MALDIMS. Thedisulphide bridges pattern was assigned as previously reported[14]. As the 5-oxoPro1-Gly62-RTI-III isoinhibitor sample had tobe exposed to slightly alkaline conditions (pH 8.5), the experi-mental data were carefully examined for disulphide bridgeinterchange. No evidence for disulphide bridge rearrangement(s)was observed.

Mass spectrometric analyses

ESMS analyses were carried out using a BIO-Q triple quadru-pole mass spectrometer equipped with an Electrospray ionsource (Micromass, Manchester, UK). In a typical experiment:10 mL of the HPLC peaks were directly injected into the ionsource via loop injection, at a flow rate of 10 mL´min21. Spectrawere recorded by scanning the first quadrupole at 10 s´scan21.Data were acquired and elaborated by the masslynx software(Micromass). Mass scale calibration was performed by means ofmultiply charged ions from a separate injection of horse-heartmyoglobin (average molecular mass 16 951.5 Da).

MALDIMS analyses were carried out using a Voyager DEmass spectrometer (PerSeptive Biosystems, Boston, MA, USA).The mass range was calibrated using bovine insulin (averagemolecular mass 5734.6 Da) and a matrix peak (379.1 Da), asinternal standards. Samples were dissolved in 0.1% trifluoro-acetic acid, the concentration being 10 pmol´mL21. In a typicalexperiment: 1.0 mL was applied to a sample slide and allowedto air-dry, before applying 1.0 mL of a solution of a-cyano-4-hydroxycinnamic acid (10 mg´mL21) in 1 : 1 : 1 (v/v/v)ethanol/CH3CN/0.1% trifluoroacetic acid. The matrix wasallowed to air-dry before collecting spectra. Mass spectra weregenerated from the sum of 50 laser shots.

q FEBS 1999 Oil-rape seed low molecular mass trypsin isoinhibitors (Eur. J. Biochem. 261) 277

1H-NMR spectroscopy of the 5-oxoPro1-Gly62-RTI-IIIisoinhibitor

All 1H-NMR spectra of 5-oxoPro1-Gly62-RTI-III were acquiredon a Bruker DMX-500 spectrometer equipped with a SiliconGraphics INDY workstation. Inhibitor samples (2.0 £ 1023 m)were dissolved in a mixture of 90% H2O and 10% 2H2O. The pHwas adjusted to 4.5 using 2HCl. Standard pulse sequences wereapplied: 2D DQF COSY [18], Clean TOCSY [19], and NOESY[20]. In all the spectra, the solvent suppression was achieved bypresaturation of the water resonance during both relaxation andevolution times. The acquisition of spectra was carried out with4096 data points in t2 and 1024 tl increments, and a spectralwidth of 8064 Hz. Clean TOCSY spectra were collected usingMLev17 sequence for spin-locking with a mixing time of60±100 ms. NOESY spectra were acquired with mixing timesof 100 and 150 ms. Prior to Fourier transformation, the datawere zero-filled to 2048 points in the tl dimension and massedwith p/3 squared shifted sine-bell in both dimensions. Abaseline correction was performed in both dimensions using apolynomial function. Data were processed using the xwinnmrprogram from Brucker. All spectra were collected at 27.0 8C,and chemical shifts were referenced to sodium-trimethylsilyl[2,2,3,3,-2H4] sulfonate.

Determination of values of the apparentthermodynamic parameters for the serine proteinase :5-oxoPro1-Gly62-RTI-III isoinhibitor complex formation

Values of the apparent association equilibrium constant (Ka) for5-oxoPro1-Gly62-RTI-III isoinhibitor binding to trypsin andchymotrypsin were determined by measuring the inhibition ofthe serine proteinase activity (using N-a-benzoyl-l-argininep-nitroanilide and N-a-carbobenzoxy-l-tyrosine p-nitrophenylester as substrates, respectively) as a function of the inhibitorconcentration, [I], according to Eqn 1 [2,21±23]:

Y � 1/�1 � �K21a /�I��� �1�

where Y is the molar fraction of the inhibited serine proteinase.Values of Ka were obtained between pH 5.0 and 9.0 (I = 0.1 m),and between 5.0 8C and 45.0 8C,

Values of apparent DG8 for 5-oxoPro1-Gly62-RTI-III iso-inhibitor binding to trypsin and chymotrypsin were calculated,at pH 8.0 and 21.0 8C, from Ka values, according to Eqn 2[21,23]:

DG8 � 2 2:3´logKa´R´T �2�

Values of apparent DH8 for 5-oxoPro1-Gly62-RTI-III isoinhi-bitor binding to trypsin and chymotrypsin were determined, atpH 8.0, from the linear dependence of logKa on T 21-values,between 5.0 8C and 45.0 8C, according to Eqn 3 [21,23]:

DH8 � 22:3´R´�dlogKa/dT 21� �3�

Values of apparent DS8 for 5-oxoPro1-Gly62-RTI-III isoinhi-bitor binding to trypsin and chymotrypsin were calculated, atpH 8.0 and 21.0 8C, from values of DG8 and DH8, according toEqn 4 [21,23]:

DS8 � �DH8 2 DG8�´T21 �4�

An error value of ^ 7% was evaluated for Ka values, of ^ 10%for DG8 and DH8 values, and of ^ 13% for DS8 values [2,22].

The pH profile was explored using the following buffers: Mes(between pH 5.0 and 6.0), Bistris (between pH 5.527.5) andTris (between pH 7.029.0); all at I = 0.1 m (sodium salts). Nospecific effects were found using different buffers with over-lapping pH values.

RESULTS AND DISCUSSION

Amino acid sequence determination of the5-oxoPro1-Gly62-RTI-III isoinhibitor

The ESMS spectrum of the native 5-oxoPro1-Gly62-RTI-IIIisoinhibitor (Fig. 1, panel A) shows a single component exhibit-ing a molecular mass of 6767.8 ^ 0.9 Da. This value is167.3 Da higher than that calculated on the basis of the aminoacid sequence of the Asp2-Pro61-RTI-III isoinhibitor (6600.5 Da)[2]. The ESMS spectrum of the reduced, denatured and alkylated5-oxoPro1-Gly62-RTI-III isoinhibitor (Fig. 1, panel B) shows asingle component exhibiting a mass value of 7240.4 ^ 0.4 Da,indicating the presence of eight carboxymethyl groups. Thesedata confirm the presence of eight Cys residues in the RTI-IIIisoinhibitor, all involved in disulphide bridges [2].

The alkylated 5-oxoPro1-Gly62-RTI-III isoinhibitor was thendigested by endoproteinase Lys-C, and the peptide mixtureanalysed by MALDIMS (Fig. 1, panel C). The mass signal atm/z 863.4 is 110.6 Da higher than that of the Asp2-Lys7 peptideof the Asp2-Pro61-RTI-III isoinhibitor [2]. Analogously, themass signal at m/z 4161.8, generated by the noncompletecleavage of the Lys7-Glu8 bond by endoproteinase Lys-C, is111.1 Da higher than that of the Asp2-Lys35 peptide of theAsp2-Pro61-RTI-III isoinhibitor [2]. The reported differences inthe molecular mass suggest the presence of a 5-oxo-prolylresidue (pyroglutamic acid) at the N-terminus of the 5-oxoPro1-Gly62-RTI-III isoinhibitor. Moreover, the mass signal at m/z1453.5 is 57.1 Da higher than that of the Cys52-Pro61 peptideof the Asp2-Pro61-RTI-III isoinhibitor [2]. The reported differ-ence in the molecular mass suggests the presence of a Glyresidue at the C-terminus of the 5-oxoPro1-Gly62-RTI-IIIisoinhibitor. All the other signals reported in Fig. 1 panel Ccorrespond to theoretical values predicted on the basis of theamino acid sequence [2].

All signal assignments given in Fig. 1 panel C were con-firmed by the analysis of the peptide mixture of the 5-oxoPro1-Gly62-RTI-III isoinhibitor by MALDIMS, after a single step ofmanual Edman degradation [14]. The mass signals correspond-ing to the peptides 5-oxoPro1-Lys7 and 5-oxoPro1-Lys35 didnot shift back, indicating the presence of a 5-oxoPro residue atthe N-terminus of the isoinhibitor.

An alternative proteolytic digest of the alkylated 5-oxoPro1-Gly62-RTI-III isoinhibitor was performed by endoproteinaseAsp-N, and the peptide mixture analysed by MALDIMS. Datagiven in Table 1 indicate that the RTI-III isoinhibitor shows apyroglutamic acid and a glycyl residue at the N- and theC-terminus, respectively.

Taking into account the presence of these two extra residuesat the N- and C-terminus, respectively, of the 5-oxoPro1-Gly62-RTI-III isoinhibitor, the molecular mass predicted for the nativeprotein is 6768.1 Da, in excellent agreement with the experi-mental value obtained by ESMS (6767.8 ^ 0.9 Da; Fig. 1,

Table 1. Mass signals and corresponding peptides detected in the

MALDIMS analysis of the carboxymethylated 5-oxoPro1-Gly62-RTI-

III isoinhibitor digested with endoproteinase Asp-N. CMCys indicates

carboxymethylcysteine.

MH+ Peptide Residues

3919�.2 5-oxoPro1-Ala33 CMCys5, CMCys18, CMCys27, CMCys31

3341�.8 Asp34-Gly62 CMCys42, CMCys52, CMCys54, CMCys57

2667�.7 Asp12-Ala33 CMCys18, CMCys27, CMCys31

1269�.9 5-oxoPro1-Gly11 CMCys5

278 P. Ascenzi et al. (Eur. J. Biochem. 261) q FEBS 1999

panel A). The presence of a 5-oxoPro residue at the N-terminalposition has also been observed in the RTI-III nonhomologoustrypsin and lysyl endopeptidase isoinhibitor III, belonging tothe squash family, isolated from oriental pickling melon(Cucumis melo L. var. Conomon Makino) seeds [24].

The amino acid sequences of the 5-oxoPro1-Gly62-RTI-IIIisoinhibitor and of related proteins are reported in Fig. 2. Theamino acid sequences of the 5-oxoPro1-Gly62-RTI-III andAsp1-Gln63-MTI-2 isoinhibitors consist of 62 and 63 aminoacid residues, respectively. A total of 43 out of 62 residues of the5-oxoPro1-Gly62-RTI-III isoinhibitor are identical to the Asp1-Gln63-MTI-2 isoinhibitor, the amino acid sequence identitybeing 69% [1,2]. Moreover, the Ser3-Pro61 segment of the5-oxoPro1-Gly62-RTI-III isoinhibitor shows a high sequencesimilarity with the Asn10-Pro68 (70%), Asp28-Cys80 (71%),Ser30-Cys81 (71%), and Cys33-Val71 (71%) regions of the puta-tive trypsin inhibitors which have been identified in theArabidopsis thaliana transcribed genome [25]. Furthermore,the Ser3-Leu53 segment of the 5-oxoPro1-Gly62-RTI-III iso-inhibitor is homologous with the Ala52-Leu102 amino acidsequence of chelonianin, a basic serine proteinase inhibitorobtained from egg white of red sea turtle. The amino acididentity is 27%, including the six Cys residues present in the

protein region considered [8,26,27]. Finally, the Asp2-Val28segment of the 5-oxoPro1-Gly62-RTI-III isoinhibitor showssome sequence similarity (39% identity) with the Lys51-Asp77segment of the rice bran trypsin Bowman-Birk inhibitor.However, this protein region, representing the connectingpeptide between domains II and III, does not include theinhibitor reactive site [1,28].

As expected for macromolecular inhibitors of trypsin-likeserine proteinases [1,2,8,29], the P1 and P1

0 reactive site residues,forming the scissile peptide bond, are Arg and Ile, respectively,in the 5-oxoPro1-Gly62-RTI-III isoinhibitor, in the Asp1-Gln63-MTI-2 isoinhibitor [1], as well as in the Asn10-Pro68, Asp28-Cys80 and Ser30-Cys81 regions of the putative trypsin inhibitorswhich have been identified in the A. thaliana transcribed genome[25] (Fig. 2). On the other hand, the P1-P1

0 scissile peptide bond isLys-Phe, in the Cys33-Val71 region of the putative trypsininhibitor identified in the A. thaliana transcribed genome [25], andLys-Thr, in the Ala52-Leu102 domain of chelonianin [8].

Althought toxins show a low sequence similarity with serineproteinase inhibitors (, 20% identity), some structurallyrelevant residues (i.e. Cys5, Gly10, Gly11, Cys18, Pro20,Cys27, Cys31, Gly40, Cys42, Gly45, Gly47 Cys52, Cys54 andCys57 present in the 5-oxoPro1-Gly62-RTI-III isoinhibitor)

Fig. 1. ESMS analysis. (A), ESMS analysis of

the native 5-oxoPro1-Gly62-RTI-III isoinhibitor.

(B), ESMS analysis of the reduced, denatured and

carboxymethylated 5-oxoPro1-Gly62-RTI-III

isoinhibitor. (C), MALDIMS analysis of the

peptide mixture of the reduced, denatured and

carboxymethylated 5-oxoPro1-Gly62-RTI-III

isoinhibitor digested with endoproteinase Lys-C.

CMCys indicates carboxymethylcysteine.

q FEBS 1999 Oil-rape seed low molecular mass trypsin isoinhibitors (Eur. J. Biochem. 261) 279

appear to be conserved (Fig. 2, and discussion thereafter).Moreover, the hydrophobic residues Phe26 and Ile25 in RTI-IIIand MTI-2 isoinhibitors, respectively [1,2], appear to beconserved and mostly aromatic in the putative trypsin inhibitorswhich have been identified in the A. thaliana transcribedgenome [25], in the Ala52-Leu102 domain of chelonianin [8],as well as in toxins, i.e. erabutoxin b [30], fasciculin 1 [31],toxin FS2 [32], cardiotoxin g [33], and toxin MTX2 [34](Fig. 2).

Assignment of disulphide bridges in the5-oxoPro1-Gly62-RTI-III isoinhibitor

The native 5-oxoPro1-Gly62-RTI-III isoinhibitor was digestedwith thermolysin, and the peptide mixture analysed by MALDIMS(see Fig. 3, and Table 2). The signals at m/z 3886.2 and 3758.7have been tentatively assigned to the three-peptide clusters,including fragments 5-oxoPro1-Gly14, Phe17-Arg21 and

Ile22-Ala37 or Ile22-Lys35, respectively, linked by twodisulphide bonds involving Cys5, Cys18, Cys27 and Cys31.These assignments have been confirmed by the presence ofsignals at m/z 3316.2, 3209.6, 3026.3, 2767.2, 2583.7,2455.1, 2398.1 and 2032.1, originated by the same clustersof peptides, but containing progressively shortened forms ofthe individual components. The formation of such a hetero-geneous mixture of cluster species involving the samedisulphide bridges pair reflects the broad specificity ofthermolysin.

Further peptide clusters were produced by thermolytic digestfrom the C-terminal portion of the inhibitor. The signal at m/z2840.8, in fact, was assigned to the sequence Ala37-Gly62cleaved at two internal sites (possibly Asn49 and Cys54, seebelow) and held together by the two remaining disulphidebridges involving Cys42, Cys52, Cys54 and Cys57. Moreover,two three-peptide clusters containing fragments Leu38-Asn49,Val50-Cys54, Phe56-Gly62 and Ala37-Asn49, Lys51-Cys54,

Fig. 2. Sequence comparison between the

5-oxoPro1-Gly62-RTI-III isoinhibitor and

related proteins. A, 5-oxoPro1-Gly62-RTI-III

isoinhibitor; B, Asp1-Gln63-MTI-2 isoinhibitor

[1]; C, Asn10-Pro68 region of the putative trypsin

inhibitor identified in the A. thaliana transcribed

genome [25]; D, erabutoxin b [29]; E, fasciculin 1

[30]; F, toxin FS2 [31]; G, cardiotoxin g [34];

H, toxin MTX2 [33]. Arrows indicate the serine

proteinase inhibitor P1-P1

0

scissile reactive site

bond. The disulphide bridge pattern of the

5-oxoPro1-Gly62-RTI-III isoinhibitor (A) and

toxin MTX2 (H) are shown. Secondary structural

elements of toxin MTX2 (H) are reported

(S, b-strand; T, b-turn). The amino acid sequences

of the 5-oxoPro1-Gly62-RTI-III, of the

Asp2-Pro61-RTI-III and of the Ser3-Pro61-RTI-III

isoinhibitors, differing only at their N- and

C-termini, have been deposited in the EMBL

Database, the common accession number being

P80301. Amino acid sequences have been

recovered from the SwissProt Data Bank

(accession numbers: A, P80301; B, P26780;

C, Z46816; D, P01435; E, P01403; F, P01414;

G, P01468; H, P18328), and aligned according to

the position of cysteines. For further details, see

text.

Fig. 3. MALDIMS analysis of the native

5-oxoPro1-Gly62-RTI-III isoinhibitor digested

with thermolysin. Assignment of the mass signals

to the corresponding peptides is reported in

Table 2.

280 P. Ascenzi et al. (Eur. J. Biochem. 261) q FEBS 1999

Cys57-Glu60, respectively, linked by the same two disulphideswere identified through mass signals at m/z 2656.1 and 2326.5,respectively. Finally, in the low mass region of the spectrum twosignals were detected at m/z 1097.2 and 1007.2 and assigned tothe peptide pairs Cys42-Asn49 + Cys52-Leu53 and Lys41-Trp44 + Val50-Leu53, respectively, joined by the Cys42-Cys52bridge. This assignment led to the identification of the seconddisulphide bridge. Because the Cys42-Cys52 bridge is involvedin the formation of the three-peptide clusters within theC-terminal portion of 5-oxoPro1-Gly62-RTI-III isoinhibitor, infact, the second disulphide bridge can only occur betweenCys54 and Cys57.

These results indicate that the 5-oxoPro1-Gly62-RTI-IIIisoinhibitor consists of two lobes, encompassing the N- andC-terminal halves of the protein, respectively, each containingtwo disulphide bridges. The two disulphide bridges occurring inthe inhibitor C-terminal region had been identified, meanwhilethe correct assignment of the remaining two disulphides wasdependent on the ability to cleave the polypeptide chain betweenCys27 and Cys31.

The thermolytic digest was then submitted to tryptic hydro-lysis and analysed by MALDIMS. The spectrum showed thedisappearance of the mass signals relative to the three-peptideclusters containing Cys5, Cys18, Cys27 and Cys31 reported inTable 2. A new signal was detected at m/z 1182.4 correspondingto the peptide Phe17-Arg21 linked to the fragment Cys31-Lys35via the Cys18-Cys31 bridge, indicating that trypsin had cleavedthe polypeptide chain at Arg30. The fourth disulphide bridgewas therefore inferred by exclusion as Cys5-Cys27.

The disulphide bridge pattern occurring in the 5-oxoPro1-Gly62-RTI-III isoinhibitor, and in related proteins, i.e. Asp2-Pro61-RTI-III and Ser3-Pro61-RTI-III isoinhibitors, wasestablished to be Cys5-Cys27, Cys18-Cys31, Cys42-Cys52and Cys54-Cys57. The disulphide bridge pattern occurringin the RTI-III isoinhibitors is reminiscent of that found inerabutoxin b [30], in fasciculin 1 [31], in toxin FS2 [32], in

cardiotoxin g [35], and in toxin MTX2 [34] (Fig. 2). Moreover,the organization of the Cys5-Cys27, Cys18-Cys31 and Cys42-Cys52 subset is reminiscent of that found in epidermal growthfactor domains present in several serine proteinases andextracellular proteins [36,37]. Next, the three disulphide bridgepattern observed in the epidermal growth factor domains isreminiscent (although in the reverse order) of that observed inhirudin, the potent thrombin inhibitor from Hirudo medicinalis

Table 2. Mass signals and corresponding peptides detected in the

MALDIMS analysis of the native 5-oxoPro1-Gly62-RTI-III isoinhibitor

digested with termolysin. S-S indicates a disulphide bridge, SH indicates a

free cysteine.

MH+ Peptide Notes

3886�.2 5-oxoPro1-Gly14 + Phe17-Arg21 + Ile22-Ala37 2S-S

3758�.7 5-oxoPro1-Gly14 + Phe17-Arg21 + Ile22-Lys35 2S-S

3316�.2 Cys5-Gly14 + Phe17-Arg21 + Ile22-Lys35 2S-S

3209�.6 5-oxoPro1-Gly11 + Phe17-Arg21 + Pro24-Lys35 2S-S

3026�.3 5-oxoPro1-Gly11 + Phe17-Arg21 + Phe26-Lys35 2S-S

2840�.8 Ala37-Gly62a 2S-S

2767�.2 Cys5-Gly11 + Phe17-Arg21 + Pro24-Lys35 2S-S

2656�.1 Leu38-Asn49 + Val50-Cys54 + Phe56-Gly62 2S-S

2583�.7 Cys5-Gly11 + Phe17-Arg21 + Phe26-Lys35 2S-S

2455�.1 Cys5-Gly11 + Phe17-Arg21 + Phe26-Asp34 2S-S

2398�.1 Cys5-Gly10 + Phe17-Arg21 + Phe26-Asp34 2S-S

2326�.5 Ala37-Asn49 + Lys51-Cys54 + Cys57-Glu60 2S-S

2275�.7 Phe17-Arg21 + Ile22-Lys35 1S-S, 1SH

2032�.1 Cys5-Glu8 + Phe17-Arg21 + Cys27-Asp34 2S-S

1503�.6 Val50-Gly62 1S-S, 1SH

1482�.8 5-oxoPro1-Gly14 SHCys5

1405�.7 Val50-Cys54 + Phe56-Gly62 1S-S, 1SH

1250�.3 Leu38-Asn49 SHCys42

1097�.2 Cys42-Asn49 + Cys52-Leu53 Cys42-Cys52

1007�.2 Lys41-Trp44 + Val50-Leu53 Cys42-Cys52

760�.8 Ile43-Asn49

592�.7 Phe17-Arg21 SHCys18

a The Ala37-Gly62 peptide shows two cleavages.

Fig. 4. NMR spectra. (A), 1H-NMR spectrum of the native 5-oxoPro1-

Gly62-RTI-III isoinhibitor in aqueous solution (90% H2O/10% 2H2O):

downfield shifted aCH region. The signal at 5.18 p.p.m. corresponds to two

a protons; (B), 2D TOCSY 1H-NMR spectrum of the native 5-oxoPro1-

Gly62-RTI-III isoinhibitor in aqueous solution (90% H2O/10% 2H2O):

aromatic region; (C), 1H-NMR spectrum of the native 5-oxoPro1-Gly62-

RTI-III isoinhibitor in aqueous solution (90% H2O/10% 2H2O): upfield

shifted methyl region. Asterisks indicate the two methyl groups of Val28.

The two signals at 0.48 and 0.58 p.p.m. correspond to two methyl groups

each. 1H-NMR spectra of the native 5-oxoPro1-Gly62-RTI-III isoinhibitor

have been obtained at pH 4.5 and 27.0 8C.

q FEBS 1999 Oil-rape seed low molecular mass trypsin isoinhibitors (Eur. J. Biochem. 261) 281

[38]. Nevertheless, RTI-III-type serine proteinase inhibitors,toxins and epidermal growth factor domains do not share anysequence similarity with hirudin and related proteins [1,2,30±38].

1H-NMR Spectroscopy of the 5-oxoPro1-Gly62-RTI-IIIisoinhibitor

High-quality 1H-NMR spectra were obtained with exceptionallywide chemical shift dispersion and resolution for the 5-oxoPro1-Gly62-RTI-III isoinhibitor (Fig. 4, panel A). Six aCH protonresonances, shifted downfield of the water line at 4.7 p.p.m. inthe range 5±6 p.p.m., indicate an extensive b-structure orstructures [39]. In the NOESY spectrum, these resonances showaa dipolar correlations, as expected for b-structure(s) [40].Exchange experiments allowed the separation of the amideprotons into two classes: fast exchanging NHs, not observablebeyond 2 h, and slow exchanging NHs, still detectable after fewdays. These amides appear to be involved in hydrogen bondsand are characterized by large values of 3JaNH couplingconstants (in the range 8±9.5 Hz), confirming the presence ofb-structure(s) [40].

In addition to the secondary chemical shifts, large shiftsresulting from the interaction with the aromatic rings (eight)were observed to affect many residues. In particular, in the5-oxoPro1-Gly62-RTI-III isoinhibitor, the chemical shifts ofthree aromatic residues, namely Tyr23, Phe26 and Trp44,experienced significant upfield ring current shift, compared tothose expected for a random conformation, namely Dd of 0.5and 1.19 p.p.m., respectively, for protons 2,6 and 3,5 of Phe26,Dd of 0.36 p.p.m. for protons 3,5 of Tyr23, and Dd of 0.5 p.p.m.for proton 7 of Trp44 (Fig. 4, panel B). Phe26 is most likelylocated in a hydrophobic cluster including Tyr23 and Trp44,surrounded by a number of apolar residues. In fact, probably dueto a close proximity with the aromatic rings, the protons of somemethyl groups, in the 5-oxoPro1-Gly62-RTI-III isoinhibitor,experienced strong diamagnetic shift in the range 20.4±0.6 p.p.m. In particular, Val28 shows the most upfield shiftedmethyl resonances (Fig. 4, panel C). Even without structural data,the presence of a hydrophobic core is also suggested by theobservation of a large number of NOEs exhibited in the NOESYspectrum between aromatic and upfield shifted methyl resonances.

The presence of a centre hydrophobic cluster has beenobserved in toxins, presenting a three-finger structure. Thiscluster, organized around a conserved aromatic residue Tyr25 inerabutoxin b [30], Tyr23 in fasciculin 1 [31] and in toxin FS2[32], Tyr22 in cardiotoxin g [33], and Phe25 in toxin MTX2[34], has been suggested to play a role in the stability of the

b-sheet structure, allowing more flexible regions to be solventexposed. In RTI-III and MTI-2 isoinhibitors, such a cluster,possibly organized around Val28 and Tyr27, respectively, mightcontribute to the right orientation of the P1-P1

0 scissile pep-tide bond. Interestingly, serine proteinase inhibitors (i.e. the5-oxoPro1-Gly62-RTI-III isoinhibitor) posses some prolyl andglycyl residues (i.e. Gly10, Gly11, Pro20, Gly40, Gly45 andGly47) which, in the toxins examined, are conserved at thebeginning of reverse or right turns, where they possibly facilitatethe reversal of the peptide chain (Fig. 2).

As a whole, structural data reported here suggests that thethree-dimensional structure of the RTI-III isoinhibitors may bereminiscent of that of toxins [30±35] and epidermal growthfactor domains [36,37], consisting of three finger-shaped loopsextending from the crossover region, and including a two- and athree-stranded anti parallel b-sheets. Thus, Cys5-Cys27 andCys18-Cys31 disulphides link finger I to II, the Cys42-Cys52disulphide connects finger II to III, and the Cys54-Cys57disulphide links the C-terminus to finger III (Fig. 2). Moreover,the protein crossover region stabilised by two disulphide bridges(i.e. Cys5-Cys27 and Cys18-Cys31 in 5-oxoPro1-Gly62-RTI-IIIisoinhibitors) may represent the stable T-knot scaffold [41,42],around which different three-dimensional structures could beorganized.

In such a three-dimensional organization, the Cys18-Cys27polypeptide loop of RTI-III isoinhibitors shows some particularstructural properties, which are found in serine proteinasereactive sites [29]. In fact, this region is strongly connected tothe protein core by two disulphide bridges (i.e. Cys5-Cys27 andCys18-Cys31). Moreover, Pro20 and Pro24, expected as the P2

and P30 residues, respectively, of the 5-oxoPro1-Gly62-RTI-III

isoinhibitor reactive site, may play a central role in the achieve-ment of proper conformation for the P1-P1

0 scissile peptide bond.Thus, the RTI-III isoinhibitor Arg21(P1)-Ile22(P1

0) scissile peptidebond is in a solvent-exposed region, which corresponds to turn 2in toxins. However, in toxins, amino acid residues involved inprotein-receptor recognition (turns 3 and 4) are located on theopposite site with respect to turn 2 [30±35]. As already reportedfor Kazal, Kunitz and Bowman-Birk inhibitors [22,29,43,44],the RTI-III isoinhibitor Arg21(P1)-Ile22(P1

0) scissile peptidebond is consistent with the specificity of trypsin-like serineproteinases, possibly representing the binding site also forchymotrypsin-like enzymes [2]. Accordingly, the affinity of theRTI-III isoinhibitors for trypsin is higher than that for chymo-trypsin (Fig. 5 and Table 3). However, we cannot exclude thatthe RTI-III isoinhibitors may display distinct reactive sites fortrypsin- and chymotrypsin-like serine proteinases.

Table 3. Values of thermodynamic parameters for the binding of the RTI-III, MTI-1 and MTI-2 inhibitors to trypsin and chymotrypsin, at pH 8.0.

Values of Ka, DG8 and DS8 were obtained at 21.0 8C. Values of DS8 were obtained from the linear dependence of logKa on T 21, between 5.0 8C and 45.0 8C.

Values of thermodynamic parameters for the binding of the native and of the P1-P10 reactive site bond cleaved RTI-III isoinhibitors to trypsin and chymotrypsin

are indistinguishable [2]. Again, values of thermodynamic parameters for the binding of the different RTI-III isoinhibitors to trypsin and chymotrypsin are

superimposable [2]. Data given in Table 3 refer to the 5-oxoPro1-Gly62-RTI-III isoinhibitor (present study). Values of thermodynamic parameters for the binding

of the native and of the P1-P10 reactive site bond cleaved MTI-2 isoinhibitors to trypsin and chymotrypsin are indistinguishable [1]. Again, values of

thermodynamic parameters for the binding of the different MTI-2 isoinhibitors to trypsin and chymotrypsin are superimposable [1]. Data given in Table 3 refer to

the Asp1-Gln63-MTI-2 isoinhibitor [1].

Proteinase Inhibitor Ka (m21)

DG8

(kJ´mol21)

DH8

(kJ´mol21)

DS8

(kJ´mol21´K21)

Trypsin RTI-III 3�.3 £ 109 253�.1 +10�.5 +0�.22

Chymotrypsin RTI-III 2�.4 £ 106 235�.6 +15�.5 +0�.17

Trypsina MTI-1 4�.5 £ 108 248�.5 +17�.2 +0�.22

Trypsin MTI-2 6�.3 £ 109 254�.8 ND ND

Chymotrypsin MTI-2 2�.0 £ 106 235�.1 ND ND

a Data from [6]; ND, not determined.

282 P. Ascenzi et al. (Eur. J. Biochem. 261) q FEBS 1999

Trypsin and chymotrypsin inhibition by the5-oxoPro1-Gly62-RTI-III isoinhibitor

Under all the experimental conditions, the binding of the5-oxoPro1-Gly62-RTI-III isoinhibitor to trypsin and chymo-trypsin conforms to a simple equilibrium, as indicated by theHill coefficient (n) always being equal to 1. Values of Ka arealways independent of the serine proteinase, inhibitor andsubstrate concentration. Moreover, values of Ka, obtained atpH 8.0, between 5.0 8C and 45.0 8C, are linearly dependent(within experimental error) with respect to T21, indicating DCp8values equal to zero. RTI-III isoinhibitors inactivate trypsin andchymotrypsin with a 1 : 1 stoichiometry [2].

Table 3 shows thermodynamic parameters for RTI-III, MTI-1and MTI-2 inhibitor binding to trypsin and chymotrypsin atpH 8.0, and between 5.0 and 45.0 8C. These data allow thefollowing considerations. The inhibitory properties of MTI-2and RTI-III isoinhibitors are superimposable, as supported bytheir high sequence identity (69%; Fig. 2) [1,2]. Moreover, thenative and the P1-P1

0

reactive site bond cleaved isoinhibitorsshow the same inhibitory properties [1,2]. The affinity of MTI-2and RTI-III isoinhibitors for trypsin is higher than that forchymotrypsin by about three orders of magnitude, in agreementwith the presence of an arginyl residue at the inhibitor reactivesite P1 position [1,2]. Values of Ka for MTI-2 and RTI-IIIisoinhibitors binding to trypsin are higher than those observedfor MTI-1 association by about one order of magnitude.However, MTI-1 does not affect significantly chymotrypsinactivity, representing the most selective trypsin inhibitor knownup to now [1,2,4±6]. DH8 and DS8 values for MTI-1 and RTI-IIIisoinhibitors binding to trypsin and chymotrypsin [6] are com-parable with those reported for the association of macromol-ecular inhibitors to serine (pro)enzymes, and indicate that thecomplex formation is an entropy-driven process. In agreementwith crystallographic analyses, the positive DS8 values shouldreflect the expulsion of the serine (pro)enzyme- and/or the

inhibitor-bound solvent water molecules during the binaryadduct formation [22,45].

The pH-profile of Ka for the 5-oxoPro1-Gly62-RTI-III bindingto trypsin and chymotrypsin (Fig. 5) is similar in magnitude tothat reported for Kazal, Kunitz and Bowman-Birk inhibitors aswell as eglin c and hirudin association to serine (pro)enzymes[22,38,43,44,46], and thus may be described with the samemodel. Therefore, on lowering the pH from 9.0 to 5.0, thedecrease of Ka values (i.e. of the affinity) for 5-oxoPro1-Gly62-RTI-III binding to trypsin and chymotrypsin reflects the acid-perturbation, upon inhibitor association, of a single ionisinggroup. According to linkage relations [47], this model leads tothe following expression (Eqn 5):

logKa � C 2 log���H�� � KUNL�/��H�l � KLIG��

2 log�KLIG/KUNL� �5�

where C is a constant that corresponds to the alkaline asymptoteof logKa, and pKUNL and pKLIG are the pK values of protondissociation constants for the inhibitor-free and the inhibitor-bound trypsin and chymotrypsin, respectively. Equation 5 hasbeen used to generate the continuous lines shown in Fig. 5, theagreement with the experimental data being fully satisfactory. Itis generally accepted [6,22,37,42±44,46] that the ionization ofthe enzyme His57 catalytic residue may affect serine protein-ase : inhibitor complex formation over the pH range explored.

CONCLUSIONS

The presence of different serine proteinase inhibitors in thesame plant, i.e. white mustard and oil-rape seed, provides aninteresting model for investigating different, but related,functions. In this respect, MTI-1 is the most selective trypsininhibitor known [5,6]. In view of its peculiar specificity, MTI-1may be interesting for pharmaceutical applications; in fact, thegreater the capability of MTI-1 to modulate selectively theactivity of trypsin in the presence of related proteinases, the greateris in principle, its possible therapeutic value as a drug. Moreover,MTI-2 and RTI-III isoinhibitors may be relevant from thebiotechnological viewpoint. In fact, the high expression level ofMTI-2 isoinhibitors has deleterious effects on specific plantpathogens, increasing the resistance of specific crop plants[3,11]. Moreover, as the three-dimensional structure of theRTI-III and MTI-2 isoinhibitors may be reminiscent of that oftoxins [30±35] and epidermal growth factor domains [36,37],protein engineering methodologies might be exploited in orderto produce a bifunctional protein of biotechnological interestendowed of both inhibitory and toxic activities. Finally, itmay be observed that proteinase inhibitors and toxins areinvolved in host±parasite and/or prey±predator interaction(s)[1±4,8±10,29,30±38]. Therefore, the reported observations mayshed light on the molecular evolution and ecology of inter-actions among species.

ACKNOWLEDGEMENTS

Authors thank Martino Bolognesi, Raffaele Gallerani, Gennaro Marino,

Sandro Palmieri and Severino Ronchi for helpful discussions. This study was

partially supported by grants from the Ministry of University, Scientific

Research and Technology of Italy (target oriented projects `Biocatalisi e

Bioconversioni' and `Biologia Strutturale'), from the Ministry of Agricul-

ture, Food and Forestry Resources of Italy (target oriented project

`Resistenze Genetiche delle Piante Agrarie agli Stress Biotici ed Abiotici'),

from the National Research Council of Italy (target oriented project

`Biotecnologie', and strategic project `Biologia Strutturale'), and from the

Fondazione `Antonio De Marco' (Milano, Italy).

Fig. 5. pH Dependence of Ka values (mm21) for 5-oxoPro1-Gly62-RTI-III

isoinhibitor binding to trypsin (W) and chymotrypsin (A), at 21.0 8C. The

continuous lines, obtained with an iterative nonlinear least squares curve-

fitting procedure, were calculated according to Eqn 5 with the following

sets of parameters: trypsin 2C = 9.5 ^ 0.1, pKUNL = 7.0 ^ 0.1, and

pKLIG = 5.1 ^ 0.2; and chymotrypsin 2C = 6.4 ^ 0.1, pKUNL = 7.0 ^ 0.1,

and pKLIG = 5.2 ^ 0.2.

q FEBS 1999 Oil-rape seed low molecular mass trypsin isoinhibitors (Eur. J. Biochem. 261) 283

REFERENCES

1. Menegatti, E., Tedeschi, G., Ronchi, S., Bortolotti, F., Ascenzi, P.,

Thomas, R., Bolognesi, M. & Palmieri, S. (1992) Purification,

inhibitory properties and amino acid sequence of a new serine

proteinase inhibitor from white mustard (Sinapis alba L.) seed. FEBS

Lett. 301, 10±14.

2. Ceciliani, F., Bortolotti, F., Menegatti, E., Ronchi, S., Ascenzi, P. &

Palmieri, S. (1994) Purification, inhibitory properties, amino acid

sequence and identification of the reactive site of a new serine

proteinase inhibitor from oil-rape (Brassica napus) seed. FEBS Lett.

342, 221±224.

3. Ceci, L.R., Spoto, N., de Virgilio, M. & Gallerani, R. (1995) The gene

coding for the mustard trypsin inhibitor-2 is discontinuous and wound-

inducible. FEBS Lett. 364, 179±181.

4. Visintin, M., lori, R., Valdicelli, L. & Palmieri, S. (1992) Trypsin

inhibitory activity in some rapeseed genotypes. Phytochem. 31,

3677±3680.

5. Menegatti, E., Palmieri, S., Walde, P. & Luisi, P.L. (1985) Isolation and

characterization of a trypsin inhibitor from white mustard (Sinapis

alba L.). J. Agric. Food Chem. 33, 784±789.

6. Menegatti, E., Boggian, M., Ascenzi, P. & Luisi, P.L. (1987) Binding of

the trypsin inhibitor from white mustard (Sinapis alba L.) seeds to

bovine b-trypsin: thermodynamic study. J. Enzyme Inhibition 2, 67±71.

7. Schechter, I. & Berger, A. (1967) On the size of the active site in

proteases: I, papain. Biochem. Biophys. Res. Commun. 27, 157±162.

8. Laskowski, M. Jr & Kato, I. (1980) Protein inhibitors of proteinases.

Annu. Rev. Biochem. 49, 593±626.

9. Ryan, C.A. (1990) Protease inhibitors in plants: genes for improving

defences against insects and pathogens. Annu. Rev. Phytopathol. 28,

425±449.

10. Broadway, R.M. (1993) Purification and partial characterization of

trypsin/chymo-trypsin inhibitors from cabbage foliage. Phytochem.

33, 21±25.

11. De Leo, F., BonadeÂ-Bottino, M.A., Ceci, L.R., Gallerani, R. & Jouanin,

L. (1998) Opposite effects on Spodoptera littoralis larvae of high

expression level of a trypsin proteinase inhibitor in transgenic plants.

Plant Physiol. 118, 997±1004.

12. Zappacosta, F., Pessi, A., Bianchi, E., Venturini, S., Sollazzo, M.,

Tramontano, A., Marino, G. & Pucci, P. (1996) Probing the tertiary

structure of proteins by limited proteolysis and mass spectrometry: the

case of minibody. Protein Sci. 5, 802±813.

13. Nitti, G., OrruÁ, S., Bloch, C. Jr, Morhy, L., Marino, G. & Pucci, P.

(1995) Amino acid sequence and disulphide-bridge pattern of three

g-thionins from Sorghum bicolor. Eur. J. Biochem. 228, 250±256.

14. Morris, H.R. & Pucci, P. (1985) A new method for rapid assignment

of S-S bridges in proteins. Biochem. Biophys. Res. Commun. 126,

1122±1128.

15. Morris, H.R., Panico, M. & Taylor, G.W. (1983) FAB-mapping of

recombinant-DNA protein products. Biochem. Biophys. Res. Commun.

140, 28±37.

16. Beavis, R.C. & Chait, B.T. (1990) Rapid, sensitive analysis of protein

mixtures by mass spectrometry. Proc. Natl Acad. Sci. USA 87,

6873±6877.

17. Luthy, J.A., Praissman, M., Finkenstadt, W.R. & Laskowski, M. Jr

(1973) Detailed mechanism of interaction of bovine b-trypsin with

soybean trypsin inhibitor (Kunitz). J. Biol. Chem. 248, 1760±1771.

18. Piantini, U., Sùrensen, O.W. & Ernst, R.R. (1982) Multiple quantum

filters for elucidating NMR coupling networks. J. Am. Chem. Soc. 104,

6800±6801.

19. Griesinger, C., Otting, G., WuÈthrich, K. & Ernst, R.R. (1988) Clean

TOCSY for 1H spin identification in macromolecules. J. Am. Chem.

Soc. 110, 7870±7872.

20. Jeener, J., Meier, B.H., Bachmann, P. & Ernst, R.R. (1979) Investigation

of exchange processes by two-dimensional NMR spectroscopy. J. Chem.

Phys. 71, 4546±4553.

21. Keleti, T. (1983) Errors in the evaluation of Arrhenius and van't Hoff

plots. Biochem. J. 209, 277±280.

22. Onesti, S., Matthews, D.J., Aducci, P., Amiconi, G., Bolognesi, M.,

Menegatti, E. & Ascenzi, P. (1992) Binding of the Kunitz-type trypsin

inhibitor DE-3 from Erythrina caffra seeds to serine proteinases: a

comparative study. J. Mol. Recogn. 5, 105±114.

23. Weber, G. (1995) van't Hoff revisited: enthalpy of association of protein

subunits. J. Phys. Chem. 99, 1052±1059.

24. Nishino, J., Takano, R., Kamei-Hayashi, K., Minakata, H., Nomoto, K.

& Hara, S. (1992) Amino acid sequences of trypsin inhibitors from

oriental pickling melon (Cucumis melo L. var. Conomon Makino)

seeds. Biosci. Biotech. Biochem. 56, 1241±1246.

25. HerveÂ, C., Bardet, C., Dabos, P., Tremousaygue, D. & Lescure, B.

(1995) Putative trypsin inhibitor, SwissProt Data Bank, accession

number: Z46816.

26. GruÈtter, M.G., Fendrich, G., Huber, R. & Bode, W. (1988) The 2.5 AÊ

x-ray crystal structure of the acid-stable proteinase inhibitor from

human mucous secretions analysed in its complex with bovine

a-chymotrypsin. EMBO J. 7, 345±351.

27. Perona, J.J., Tsu, C.A., Craik, C.S. & Fletterick, R.J. (1993) Crystal

structures of rat anionic trypsin complexed with the protein inhibitors

APPI and BPTI. J. Mol. Biol. 230, 919±933.

28. Tashiro, M., Hashino, K., Shiozaki, M., Ibuki, F. & Maki, Z. (1987)

The complete amino acid sequence of rice bran trypsin inhibitor.

J. Biochem. 102, 297±306.

29. Bode, W. & Huber, R. (1992) Natural protein proteinase inhibitors and

their interaction with proteinases. Eur. J. Biochem. 204, 433±451.

30. Hatanaka, H., Oka, M., Kohda, D., Tate, S.-i., Suda, A., Tamiya, N. &

Inagaki, F. (1994) Tertiary structure of erabutoxin b in aqueous

solution as elucidated by two-dimensional nuclear magnetic reso-

nance. J. Mol. Biol. 240, 155±166.

31. le Du, M.H., Marchot, P., Bougis, P.E. & Fontecilla-Camps, J.C. (1992)

1.9 A resolution structure of fasciculin 1, an anti-acetylcholinesterase

toxin from green mamba snake venom. J. Biol. Chem. 267,

22122±22130.

32. Albrand, J.-P., Blackledge, M.J., Pascaud, F., Hollecker, M. & Marion,

D. (1995) NMR and restrained molecular dynamics study of the

three-dimensional solution structure of toxin FS2, a specific blocker of

the l-type calcium channel, isolated from the black mamba venom.

Biochemistry 34, 5923±5937.

33. Gilquin, B., Roumestand, C., Zinn-Justin, S., MeÂnez, A. & Toma, F.

(1993) Refined three-dimensional solution structure of a snake

cardiotoxin: analysis of the side-chain organization suggests the

existence of a possible phospholipid binding site. Biopolymers 33,

1659±1675.

34. SeÂgalas, I., Roumestand, C., Zinn-Justin, S., Gilquin, B., MeÂnez, R.,

MeÂnez, A. & Toma, F. (1995) Solution structure of a green mamba

toxin that activates muscarinic acetylcholine receptors, as studied by

nuclear magnetic resonance and molecular modeling. Biochemistry

34, 1248±1260.

35. Dauplais, M., Neumann, J.M., Pinkasfeld, S., MeÂnez, A. & Roumestand,

C. (1995) A NMR study of the interaction of cardiotoxin g from

Naja nigricollis with perdeuterated dodecylphosphocholine micelles.

Eur. J. Biochem. 230, 213±220.

36. Appella, E., Weber, I.T. & Blasi, F. (1988) Structure and function of

epidermal growth factor like regions in proteins. FEBS Lett. 231, 1±4.

37. Kohda, D. & Inagaki, F. (1992) Three-dimensional magnetic resonance

structure of mouse epidermal growth factor in acidic and physiological

pH solutions. Biochemistry 31, 11928±11939.

38. Ascenzi, P., Amiconi, G., Bode, W., Bolognesi, M., Coletta, M. &

Menegatti, E. (1995) Proteinase inhibitors from the European

medicinal leech Hirudo medicinalis: structural, functional and bio-

medical aspects. Mol. Aspects Med. 16, 215±313.

39. Wishart, D.S., Sykes, B.D. & Richards, F.M. (1991) Relationship

between nuclear magnetic resonance chemical shift and protein

secondary structure. J. Mol. Biol. 222, 311±333.

40. WuÈthrich, K. (1986) NMR of Proteins and Nucleic Acids. John Wiley,

New York, USA.

41. Lin, S.L. & Nussinov, R. (1995) A disulphide-reinforced structural

scaffold shared by small proteins with diverse functions. Nat. Struct.

Biol. 2, 835±837.

42. Ascenzi, P., Bolognesi, M., Catalucci, D., Pascarella, S., Ruoppolo, M.

& Rizzi, M. (1998) Leech antihemostatic proteins share the T-knot

284 P. Ascenzi et al. (Eur. J. Biochem. 261) q FEBS 1999

scaffold, a disulphide- reinforced structural motif. Biol. Chem. Hoppe

Seyler 379, 1387±1389.

43. Menegatti, E., Guarneri, M., Bolognesi, M., Ascenzi, P. & Amiconi, G.

(1987) Binding of the bovine pancreatic secretory trypsin inhibitor

(Kazal) to bovine serine (pro) enzymes. J. Mol. Biol. 198, 129±132.

44. Ascenzi, P., Amiconi, G., Bolognesi, M., Guarneri, M. & Menegatti, E.

(1990) Binding of the soybean Bowman-Birk proteinase inhibitor

and of its chymotrypsin and trypsin inhibiting fragments to bovine

a-chymotrypsin and bovine b-trypsin: a thermodynamic study. J. Mol.

Recogn. 3, 192±196.

45. Amiconi, G., Ascenzi, P., Bolognesi, M., Guarneri, M. & Menegatti, E.

(1987) Enthalpy-entropy compensation phenomena in serine (pro)

enzyme: inhibitor complex formation. Adv. Biosci. 65, 177±180.

46. Ascenzi, P., Amiconi, G., Coletta, M., Lupidi, G., Menegatti, E., Onesti,

S. & Bolognesi, M. (1992) Binding of hirudin to human a, b and

g-thrombin: a comparative kinetic and thermodynamic study. J. Mol.

Biol. 225, 177±184.

47. Wyman, J. (1964) Linked functions and reciprocal effects in

hemoglobin: a second look. Advan. Protein Chem. 19, 223±286.