Bowman–Birk protease inhibitor from the seeds of Vigna unguiculata forms a highly stable dimeric structure

Biochimica et Biophysica Acta 1774 (2007) 1264 – 1273 www.elsevier.com/locate/bbapap

Bowman–Birk protease inhibitor from the seeds of Vigna unguiculata forms a highly stable dimeric structure K.N. Rao 1 , C.G. Suresh ⁎ Division of Biochemical Sciences, National Chemical Laboratory, Pune-411008, India Received 1 March 2007; received in revised form 2 July 2007; accepted 16 July 2007 Available online 3 August 2007

Abstract Different protease inhibitors including Bowman–Birk type (BBI) have been reported from the seeds of Vigna unguiculata. Protease isoinhibitors of double-headed Bowman–Birk type from the seeds of Vigna unguiculata have been purified and characterized. The BBI from Vigna unguiculata (Vu-BBI) has been found to undergo self-association to form very stable dimers and more complex oligomers, by sizeexclusion chromatography and SDS-PAGE in the presence of urea. Many BBIs have been reported to undergo self-association to form homodimers or more complex oligomers in solution. Only one dimeric crystal structure of a BBI (pea-BBI) is reported to date. We report the threedimensional structure of a Vu-BBI determined at 2.5 Å resolution. Although, the inhibitor has a monomer fold similar to that found in other known structures of Bowman–Birk protease inhibitors, its quaternary structure is different from that commonly observed in this family. The structural elements responsible for the stability of monomer molecule and dimeric association are discussed. The Vu-BBI may use dimeric or higher quaternary association to maintain the physiological state and to execute its biological function. © 2007 Elsevier B.V. All rights reserved. Keywords: Bowman–Birk protease inhibitor; Vigna unguiculata; Plant protease inhibitor; Protein–protein interaction

1. Introduction Inhibitor proteins of proteases are ubiquitous in nature. Animals, plants and microorganisms contain a number of ‘protease inhibitors’ that form reversible, stoichiometric protein–protein complexes with their cognate proteolytic enzymes [1,2]. Usually, they are present in multiple forms in different tissues of organisms. For the last several years protein inhibitors (PIs) have been investigated for various reasons which also include their utility in the study of protein–protein interactions [3]. Their gross physiological function could be the prevention of unwanted proteolysis and thereby control the protein turnover and Abbreviations: BBI-Bowman, Birk protease inhibitor; PIs, Protein Protease inhibitors;ES-MS,Electrosprayionizationmassspectrometry;SDS-PAGE,Sodium dodecyl sulfate-polyacrylamide gel electrophoresis; FPLC, Fast protein liquid chromatography; BT, Bovine Trypsin; NCS, Non-Crystallographic Symmetry ⁎ Corresponding author. Tel.: +91 20 25902236; fax: +91 20 25902648. E-mail address: [email protected] (C.G. Suresh). 1 Present address: Biology Department, Brookhaven National Laboratory, Upton, NY 11973, USA. 1570-9639/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.bbapap.2007.07.009

metabolism. The most abundant sources of PIs are the plants; their presence in plants is known since long. A large number of PIs have been isolated and characterized; majority of them are serine protease inhibitors [4,5]. Most PIs are of comparatively low molecular weight (4–20 kDa), soluble in water and their polypeptide chains are non-glycosylated. Although, large variation in the overall structure of serine protease inhibitors exist, their insertion loop regions display remarkable similarity in conformation. To ascertain the general principles of inhibition by protein inhibitors, structural study of a variety of complexes will be prudent. Indeed, an increasing number of structures of complexes analyzed using X-ray crystallography have been reported. However, the details of the mechanism and fine specificity of complex formation still remain obscure. In this context it is interesting to note that out of some 70 structures of protease inhibitor–enzyme complexes determined to date only few involve Bowman–Birk protease inhibitors [3,6]. A Bowman–Birk inhibitor (BBI) was first isolated from soybeans by Bowman and its biochemical properties were studied by Birk [7]. Subsequently, many BBIs have been isolated and

K.N. Rao, C.G. Suresh / Biochimica et Biophysica Acta 1774 (2007) 1264–1273

characterized from legumes (only Fabaceae family), Gramineae and many other plants [8]. The special features of dicot members of BBI include: (1) small molecular weights in the range of 6– 9 kDa; (2) occurrence of seven disulfide bridges that stabilize their active configurations; (3) presence of two tandem homologous domains, each bearing an insertion loop and consequently capable of inhibiting two protease molecules simultaneously and independently, making them “double-headed” inhibitors. BBIs from dicotyledonous seeds are of 8 kDa size and doubleheaded. In contrast, the monocots have 8 kDa single-headed and 16 kDa double-headed inhibitors [9]. Dicot BBIs could have evolved from a single-headed ancestral BBI via internal gene duplication, fusion and mutation. It has been suggested that during evolution one of the reactive sites of the 8-kDa doubleheaded BBI became non-functional resulting in an 8-kDa singleheaded BBI in monocots [10]. The 16-kDa double-headed BBI might have evolved from this 8 kDa inhibitor, subsequently, by gene duplication and fusion. The intra molecular sequence homology of BBIs renders credence to this hypothesis. Presently, sequences of more than 100 Bowman–Birk type protease inhibitors from plant sources are available at the PLANTPIs database accessible at http://bighost.area.ba.cnr.it/ PLANT-Pis [11]. Many BBIs from plants have been isolated and characterized. Despite these extensive studies, few three-dimensional structures of these proteins are available. The three-dimensional structures of BBIs from different plant seeds, both in the native state and in complex with trypsin or chymotrypsin, have been reported [10,12–20]. The structural features and the application of BBIs are described in a recent review article [6]. Many BBIs, including the Vu-BBI, have been reported to undergo selfassociation to form homodimers or trimers or more complex oligomers in solution [21,22]. However, adequate structural information to understand the self-association phenomenon of BBIs is Table 1 Structure refinement details Parameter

Value

a

R factor and bR free (%) c Cc and bCcfree (%) Number of reflections for refinement Number of reflections for Rfree Average B-factor for all atoms Number of protein atoms Number of solvent molecules

19.1 and 25.8 87.6 and 92.8 3763 177 46.4 798 58

RMS deviation from ideality (a) bond lengths (Å) (b) bond angles (°)

0.01 3.1

Ramachandran plot Residues in: (a) most favored region (%) (b) additionally allowed region (%) (c) generously allowed region (%) (d) disallowed region (%)

84.4 12.5 3.1 0.0

R-factor = (∑||Fo| − |Fc||)/(∑|Fo|). R-free and Ccfree are calculated for about 5% of the data that was excluded from refinement. c Cc: correlation coefficient = {[∑(Fo − bFoN) (Fc − bFcN)] / [∑(Fo − bFoN)2 (Fc − bFcN)2]½} where Fo and Fc are the observed and calculated structure factor amplitudes and bFoN and bFcN are their averages over all reflections, respectively. a

b

1265

Table 2 The percentage of NaCl (B) at which the isoinhibitors eluted from MonoQ column, the isoelectric pH values estimated using IEF unit, the molecular weights determined by ES-MS method for the six isoinhibitors and their specific activities towards trypsin and chymotrypsin, respectively, are listed here Isoinhibitor

Elution conc. % NaCl (B)

Isoelectric pH values

Mol. wt. in Dalton

Specific activity a for trypsin (units/mg)

Specific activitya for chymotrypsin (units/mg)

PI PII PIII PIV PV PVI

12.5 16.0 17.5 19.0 20.2 21.5

4.45 4.68 4.95 5.16 5.38 5.71

7761 8050 7890 8180 8005 8418

607 413 345 432 390 233

40 57 68 52 132 42

a

Maximum value of estimated standard deviation in activity is 5 units.

lacking. Studies on Vu-BBI including detailed characterization, stability, self-assembling tendency, activity and exposed hydrophobic surface have been reported [21,23–26]. Here we present the characterization and study of the dimeric crystal structure of BBI from the seeds of Vigna unguiculata (‘cowpea’ or ‘black eyed pea’). Since Vu-BBI exists in equilibrium between multimeric states, structural studies have been limited thus far due to difficulty to push Vu-BBI into a single species of molecular association and help crystallization. The crystallization and preliminary X-ray crystallographic analysis of BBI from black eyed pea seeds is reported by us [27] and crystallization of an isoinhibitor from same seeds in complex with bovine trypsin [28] and its threedimensional structure has been reported by Barbosa et al. [29]. 2. Materials and methods The BBIs were purified from dry cowpea seeds by extraction, using ionexchange chromatography (DEAE-Sephadex) and gel filtration (Sephadex-G50) followed by FPLC using MonoQ anion exchanger [27]. The homogeneity of the preparation was checked using native PAGE and SDS-PAGE, isoelectric focusing and X-ray film-contact print technique. In the X-ray film-contact print technique, after running the native PAGE, the gel was incubated in 0.1 mg/ml protease solution; then the gel was washed and placed on an undeveloped X-ray film. The inhibitor bands appear as unhydrolyzed gelatin against the background of hydrolyzed gelatin. Later the rear side of the film can be cleared with protease and then be developed. The specific activity of each isoinhibitor was determined against both trypsin and chymotrypsin. Kunitz's method [30] was used to assay activity on casein substrate by measuring the absorbance of the supernatant at 280 nm after treatment with trichloroacetic acid (TCA). The reaction mixture (2 ml) in 0.1 M potassium phosphate buffer pH 7.5 and casein (10 mg or appropriate concentration) contained the respective protease alone or with inhibitor. The mixture was incubated for 15 min at 37 °C and the reaction was terminated with the addition of 3 ml of 5% TCA. The mixture was centrifuged for 10 min at 10,000 rpm and the absorbance of the clear solution read at 280 nm. The readings were corrected for absorbance due to casein, enzyme and inhibitor. One unit of protease is defined as the amount of protease required for the hydrolysis of 1 μmol/min casein. One unit of inhibitor is defined as the amount required for inhibiting hydrolysis of 1 μmol/min casein. Specific activity of the inhibitor is the activity in units per 1 mg of the pure inhibitor. Protein concentration was estimated by Lowry method using bovine serum albumin (BSA) as standard [31]. Bovine trypsin, porcine chymotrypsin and BSA are purchased from Sigma-Aldrich. We have reported the crystallization and preliminary X-ray studies of the BBI from Vigna unguiculata (Vu-BBI) [27]. The structure of tracey bean BBI (pdb, 1pi2) was used as model for structure determination using molecular replacement

1266

K.N. Rao, C.G. Suresh / Biochimica et Biophysica Acta 1774 (2007) 1264–1273

method implemented in AMoRe [32]. The rigid body refinement of solutions from molecular replacement calculations was carried out using REFMAC program [33] with the data in the resolution range 15–3.0 Å. This was followed by several cycles of positional refinement using data in the same resolution range. Electron density maps were visualized at this stage. Since the molecule was a dimer in the asymmetric unit, molecular averaging was tried, making use of the presence of noncrystallographic symmetry (NCS). The averaged electron density map looked better and the initial model could be improved by averaging. Side chains were progressively fitted as the map quality improved and at positions where the side chain density was ambiguous alanine residues were retained. Solvent molecules were added progressively to the structure using X-SOLVATE module of QUANTA [34]. At the end of the refinement, most of the residues were identified unambiguously except for those at the chymotrypsin binding loop region and terminal residues. The electron density map was poor in one of the active-site loop regions of both monomers in the dimer. The quality of the structure was checked using PROCHECK program [35]. The refinement statistics is shown in Table 1. Structure is deposited in protein data bank, accession code:2OT6.

3. Results and discussion 3.1. Characterization of BBIs In the present study, we were able to separate six isoinhibitors, labeled P-I to P-VI (Table 2) to homogeneity and characterize

them. These isoproteins differed in charge since multiple bands were seen on isoelectric focusing gels and thus were separated on FPLC using MonoQ anion exchange column, eluted with various concentrations of NaCl. One of the reasons for undertaking the separation and characterization of the isoinhibitors is to characterize them based on their physico-chemical properties. Indeed, the six isoforms could be distinguished by their variation in specific activity towards trypsin and chymotrypsin, in terms of their isoelectric points and molecular weights determined by electro-spray mass spectrometry (ES-MS) (Table 2). All the six isoinhibitors showed inhibitory activity towards both trypsin and chymotrypsin and hence are recognized as ‘double-headed’ PIs. However, they differed in terms of their specific activity, against trypsin and chymotrypsin (Table 2). The molecular weight of major isoinhibitor P-IV determined by gel filtration technique using standard molecular weight markers kit was 14.8 kDa and that determined by SDS-PAGE method was 15.2 kDa. However, this estimate was not in agreement with that generally observed for Bowman–Birk type or Kunitz type PIs. But the isotopically averaged masses measured by ES-MS method showed that they were close to 8 kDa, the typical molecular weight of classical

Fig. 1. Multiple sequence alignment using CLUSTALW [45] of several representative ‘double-headed’ dicot PIs of Bowman–Birk family. The ⁎ indicates conserved residues and the two binding site loop residues are denoted by Pn to Pn′. Only those sequences with more than 50% sequence homology are included in the list. The chain length corresponding to that seen in the electron density of Vu-BBI only is listed. The sequences are, V1—electron density derived sequence of Vu-BBI, S1—gi|124042 -BTCI from cow pea, S2—Q4VVG2-CPTI from cow pea, S3—Q9S9H8—Isoinhibitor FIV from cowpea, S4—BBI Pea seeds, PSTI-IVa (precursor), S5—Pea seeds (Chain A, crystal structure), S6—Garden pea, TI12-36, S7—Garden bean, PVI-3, S8—Vicia angustifolia (common vetch), S9—fava bean, S10—Soybean, D-II (Precursor), S11—Tracey bean, PI-IV, S12—Soybean CII (precursor), S13—Macrotyloma axillare seed DE-4, S14—Macrotyloma axillare seed DE-3, S15—adzuki beans PI-I, S16—adzuki beans, PI-II, S17—adzuki bean-IA, S18—horse gram seeds, S19—Lima beans, S20—Mung bean beans, S21—Kidney bean-II, S22—Garden bean-II, S23—Dioclea glabra, DGTI-I, S24—apple leaf seed-DE4, S25—Canavalia lineata, CLTI-II, S26—Erythrina variegata seeds, EBI, S27—snail medic seeds, MSTI, S28—alfalfa leaves, S29—Peanut A-II, S30—Peanut B-II.

K.N. Rao, C.G. Suresh / Biochimica et Biophysica Acta 1774 (2007) 1264–1273

double-headed dicot BBIs [7,9]. The molecular weight of the VuBBI isoinhibitor P-IV estimated using gel filtration is double this value. Thus gel filtration experiment suggests the existence of BBI as dimer in solution. Similarly the molecular weight of the band determined using SDS-PAGE could be thought as of an undissociated dimer. Alternatively, this aberration can also be thought of as a result of abnormal migration, as previously observed in other similar proteins [22,36]. It may be noted that the dimer band remained same even after boiling the sample in presence of βmercaptoethanol (β-me) and sodium dodecyl sulfate (SDS) supporting the possibility of existing as a very stable dimer. In the presence of urea multiple bands corresponding to mixedassociations of different number of monomer chains were observed. Atomic force microscopic studies and modeling of BBIchymotrypsin complex have also suggested equilibrium between monomers and several multimers of Vu-BBI in solution [21,26]. By using partial N-terminal sequencing and with the help of electron density map combined with sequences of isoinhibitors reported previously from the same source, we could infer the full amino acid sequence of the isoinhibitor P-IV that crystallized. There was no electron density in the crystal structure for the first 16 residues at the N-terminus. However, the cleavage between residues 74 and 75 which is known to happen in these type of inhibitors [22] explains the absence of C-terminal residues in the map, which also agrees with the estimated molecular weight (MW) of 8.1 kDa. Thus the inferred full sequence of isoinhibitor P-IV is “SGHHEDSTDEPSESSEPCCDSCVCTKSIPPQCHCTNIRLNSCHSGCKSCLCTFSIPGSCRCLDIANFCYKPCKS” matching the estimated MW. Thus, we believe that P-IV is a different isoinhibitor than those reported by other authors [24,29,37,38]. Physiological role and the reason for several isoinhibitors to be present in the same plant are still not very clear. Properties including molecular weight established by ES-MS indicate that the isoinhibitors are produced either through post-translational modification or gene duplication. Also it has been postulated that in a co-evolving system, plants and insects evolve with new forms of PIs and proteases to counter each other's defense mechanisms [39]. Therefore it is possible that the plants evolved numerous isoinhibitors against diverse digestive enzymes of various parasites in nature [5,40].

1267

3.3. The three-dimensional structure The Vu-BBI (isoinhibitor P-IV) crystallized in space group P21 and the asymmetric unit contained two inhibitor molecules. The secondary structure consists mostly of β-sheets and is devoid of any α-helices. The monomeric structure of BBI can be divided into two domains, an N and a C-domain. Each domain has a two-stranded anti-parallel β-sheet and an associated short strand (Fig. 2). The monomer fold of this BBI, is the same “bowtie motif”, found in other BBI structures [10,12–20]. The two βsheets place the trypsin and chymotrypsin-binding loops towards two opposite ends of the molecule (Fig. 2). The trypsin-binding domain (N domain) is formed from two peptide chain segments (residues Pro17–Thr35 and Asp63–Lys73) separated in the sequence, whereas the chymotrypsin-binding domain (C domain) is formed by a continuous segment of residues Asn36–Leu62. The interactions of the domains are through hydrogen bonds between main chain atoms and hydrophobic contacts between side chains. The individual domains are further stabilized through seven disulfide bridges. There are four such bridges in the N-domain and three in the Cdomain. The connecting disulfide bridges are Cys18–Cys72, Cys19–Cys34, Cys22–Cys68, Cys24–Cys32 in the N domain, and Cys42–Cys49, Cys46–Cys61, Cys51–Cys59 in the C domain (Fig. 2). These seven disulfide bridges provide high

3.2. Sequence comparison of dicot BBIs The sequence of Vu-BBI derived from our electron density (e.d.) map was aligned against sequences of reported isoinhibitors from the same source and various representative sequences of dicot BBIs (Fig. 1). The alignment shows that the sequences are highly homologous, especially the cysteine positions are conserved throughout. The sequence identity among these BBIs is more than 50%, which imply that these BBIs may share a common tertiary fold. Also these proteins display an intra-molecular sequence identity of 55% between the N and C domains. The conservation of residues in the trypsin and chymotrypsin binding loop regions is significantly higher. These residues are important for the inhibitor to achieve the β-turn ‘VI b’ conformation of polypeptide chain.

Fig. 2. The subunit structure of Vu-BBI, disulfide bridges are shown in liquorice yellow color.

1268

K.N. Rao, C.G. Suresh / Biochimica et Biophysica Acta 1774 (2007) 1264–1273

rigidity to the structure and in the event of any proteolytic cleavage of the reactive or insertion sites the conformational changes can confine to insertion loops alone while the rest of the structure can remain intact [41]. The structure has no hydrophobic residues in the core. The side chains of Pro17, Pro29, Ile28, Phe67, Ile64 and Pro71 are found exposed to solvent, and constitute the hydrophobic patches on the surface of the protein. Another notable feature is the buried polar or charged residues Gln31, Asn36, Asp63 and Asn66. Similarly the polar or charged side chains of Asp20, Asn36, Arg60, Asp63 form an electrically charged cluster at the inter-domain region. The interactions of the side chain atoms of the invariant aspartates and an arginine in both monomers

presumably stabilize the association and orientation of the two domains. The exposed hydrophobic patches on the surface of the protein are an unusual structural feature, but common among BBIs [21,22,42]. The nature of the side chains on the surface of the protein is thought to be responsible for the commonly observed phenomenon of self-association in BBIs. The structural patterns observed here are in contrast with that of typical water-soluble proteins. The structural stability of the BBI monomer does not seem to have arisen from the hydrophobicity factor as normally the case with most of the globular proteins. The structural stability appears to be contributed by a combination of numerous hydrogen-bonded contacts, electrostatic attraction at the inter-

Fig. 3. The dimeric structures of (A) Vu-BBI showing the 2-fold NCS symmetry and the four binding sites, also shown are the six invariant water molecules as van der Waals spheres, (B) pea-BBI [12] showing the arrangement of 4 binding sites at the corners of a square, (C) the close view of the dimer interface of the Vu-BBI with the interacting residues shown in ball and stick in yellow. Hydrogen bonds are shown in black with dashed lines. Figure is prepared in MOLSCRIPT [46].

K.N. Rao, C.G. Suresh / Biochimica et Biophysica Acta 1774 (2007) 1264–1273

domain region and by the rigid framework of disulfide bridges along with higher proline content. 3.4. Quaternary association The asymmetric unit in the crystal structure contains two BBI molecules – subunits A and B – that closely interact to form a compact dimer (Fig. 3A and C). A pseudo 2-fold axis whose direction is different from the unit cell axes directions relates the two subunits. The approximate dimensions of the dimer are 40 X 53 X 30 Å. The r.m.s. deviation in Cα positions on superposition of two subunits of the dimer is 0.52 Å. The overlap of the monomers is poor in the chymotrypsin loop region that is anyway poorly defined in the electron density map. The core of the dimer interface involves an extensive network of hydrogen bonds involving residues Gln31, Gly45, Lys47, Asn66, Tyr69 and Cys68 (Fig. 3C). The interface region has a dozen hydrogen bond interactions as shown in Table 3 and Fig. 3C. The strands 64–66 of both monomers come very close to each other forming an anti-parallel β-sheet (Fig. 3C), with amino and carbonyl of Ile64 and Asn66 forming two hydrogen bonds. The two monomers are symmetrically positioned about this β-sheet. Mainly the trypsin-specific loops of the two molecules are positioned on one side of the β-sheet plane and the rest of the molecules on the other side. On the trypsin loop side of the sheet plane, the interactions are between the side chains of Gln31 and Asn66. On the opposite side of the β-sheet, side chains of Lys47 and Tyr69 interact with the carbonyl groups of Cys68 and Gly45, respectively. A dimer structure of BBI was first observed in the crystal structure of pea-BBI [18]. The association of monomers in peaBBI is different from that of Vu-BBI. Since their monomers superposed well, their difference is confined to quaternary structure. Like the pea-BBI structure, Vu-BBI also forms a tight dimer in the asymmetric unit. In pea-BBI the two monomers orient almost perpendicular to each other (Fig. 3B), whereas in Vu-BBI the molecules associate face-to-face (Fig. 3A). In the case of Vu-BBI the C-terminal tail from the two monomers pack together closely to form a 2-stranded β-sheet. In pea-BBI the CTable 3 Possible hydrogen bonded interactions between the subunits A and B of Vu-BBI classified according to the type of interaction Nature of the hydrogen bond

Main chain–Main chain Main chain–Side chain

Side chain–Side chain

Atoms of A subunit

Atoms of B subunit

(residue/No./atom)

(residue/No./atom)

Ile Asn Gly Lys Cys Tyr Gln Gln Gln Asn Asn Asn

Asn Ile Tyr Cys Lys Gly Asn Asn Asn Gln Gln Gln

aA64 A66 A45 A47 A68 A69 A31 A31 A31 A66 A66 A66

O N O NZ O OH OE1 OE1 NE2 OD1 ND2 ND2

B66 B64 B69 B68 B47 B45 B66 B66 B66 B31 B31 B31

Distance (Å)

N O OH O NZ O OD1 ND2 OD1 NE2 NE2 OE1

2.93 2.96 2.56 2.99 2.58 2.76 3.16 2.81 2.48 2.52 3.30 3.10

1269

terminal tail from one monomer crosses over to the other monomer and associates with the N-terminal β-strand to form a continuous 4-stranded β-sheet. The β-sheet in pea-BBI is lengthier compared to that in Vu-BBI and consequently there are two hydrogen bonds observed between the main chain atoms in Vu-BBI and three in pea-BBI. The two monomers of Vu-BBI are arranged more symmetrically at the dimer interface. All the residues interacting at the dimer interface are related by an approximate 2-fold symmetry and a corresponding symmetrical hydrogen bonded network is observed (Table 3). The peaBBI dimer interface lacks the same type of symmetry observed in Vu-BBI. Unlike in the Vu-BBI where the trypsin binding sites come close to each other, the four binding sites of the pea-BBI can be imagined to be placed at the four corners of a square with the molecules oriented along the diagonals. However, it is still unknown whether these dimers are the functional units of the molecule. Since the association of monomers in Vu-BBI is different from that of pea-BBI, the reported interaction of lysine from chymotrypsin binding site with an aspartic acid at the Cterminal, suggested to aid self association to form dimer [22], is not possible in the case of Vu-BBI. The three-dimensional structure of the ternary complex of bi-functional soybean-BBI monomer shows that it can bind two trypsin molecules at the same time [19]. Similarly, a modeled ternary complex of peaBBI with trypsin and chymotrypsin indicates that the simultaneous binding of two proteases is possible only with a monomeric molecule [18]. It is possible that nature has selected the present type of dimer to protect one set of protease inhibitory sites from any unwanted protease attack. 3.5. Topology of the reactive site loops and inhibitory mechanism The two protease-binding sites of BBI are located in the turn region of the two pairs of anti-parallel β-strands that make up the two domains of the molecule (Fig. 2). In each subunit the two reactive site loops are at two symmetrically related ends along the longest dimension of the subunit, 36 Å apart. This arrangement of the reactive sites facilitates simultaneous inhibition of two protease molecules. The two loops project out from the BBI so as to be easily accessible to the active site of proteolytic enzymes. The chymotrypsin insertion loop is poorly defined in the electron density map of both the monomers (Fig. 4A and B). We think the poor density is due to large flexibility and resultant high disorder of these active site loops at their protease insertion regions. Otherwise we expect these loop structures to be more rigid due to disulphide bridges and hydrogen bonds. The trypsin-binding loops are clearly seen in the map. These latter loops in each monomer are stabilized at their tips presumably by the mutually close disposition of them in the dimer. The reactive site loops are constrained by disulfide bridges, Cys24–Cys32 in the N-domain and Cys51–Cys59 in the Cdomain. The disulfide bridges could be playing the role of limiting the conformational freedom of the loops. In addition, the hydrogen bonds and van der Waals contacts between residues at the N-terminal and C-terminal sides of the loops also stabilize the rigid structure of the reactive site loops. The

1270

K.N. Rao, C.G. Suresh / Biochimica et Biophysica Acta 1774 (2007) 1264–1273

Fig. 4. The electron density (2Fo-Fc) is poorly defined for chymotrypsin-binding site (CBS) loops in the two monomers of the dimer in the crystal structure of Vu-BBI. (A) One of the subunits in which the electron density missing for residues 55 and 56 is shown. (B) The second subunit in which the main chain of the polypeptide could be built into the electron density of the same loop, although the density was not excellent. The residue numbers are shown. The residues 51 and 59 are cystines that form disulphide bridge. The electron densities are drawn at contour level of 1σ.

extremities of the two loops are tethered by two symmetry related strong polar interactions, the side chains of Asn36 and Asp63 are hydrogen bonded to the main chain nitrogen atoms of Asp20 and Ser21 and Lys47 and Ser48. It is interesting to note that Asn36 and Asp63 are invariant among BBIs [18], their positions are related by 2-fold pseudo-symmetry, and may have a role in stabilizing the structure. A conserved cis-proline

residue at the P3′ position in each binding region creates a turn classified as type ‘VI b’. The residues at P3′ and P4′ positions and the internal hydrogen bond between Thr at P2 and Ser at P1′ positions help the loop to acquire this conformation. This feature is observed throughout the serine protease inhibitor families, despite possessing different topologies and sequence variation within the reactive sites themselves [1,3]. The pointed shape of

K.N. Rao, C.G. Suresh / Biochimica et Biophysica Acta 1774 (2007) 1264–1273

the turn ‘VI b’ reflects the lock and key motif thought to be the mode of binding of serpins to their cognate enzymes [2,6]. One notable difference between the two reactive loops in the monomer is at P3′–P4′ in the sequence. In domain I these positions are occupied by two proline residues, whereas in domain II it is Pro–Gly in this position as found in most BBIs (Fig. 1). One reason for the observed differences in specificity of loops may be attributed to the conformational differences caused by the above difference in residues. The nature of the residue in the crucial position P1 is selected by the specificity of the loop corresponding to proteases inhibited. The P1 residue is lysine for trypsin binding, while the residue could not be identified in chymotrypsin inhibitory loop because of poor electron density. In addition to the P1 residue, the residues at P2 to P2′ are also important for binding proteases. These positions are occupied by residues Thr at P2, Ile at P2′ and Ser at P1′. The P2 position in natural BBIs is most conserved and occupied by Thr. Usually a large aliphatic side chain, frequently Ile, is found at P2′. The residues of the trypsin-binding loop seem to be more conserved than those of chymotrypsin loop (Fig. 1). The loops are solvent exposed, hence their surface accessibility and B-factors are high. The small size of the reactive site loop of BBI (9 residues) is in contrast with the large reactive site loop of the Kunitz type PI, which contains as many as 50 residues, suggesting that the differences in molecular weights and cysteine content reflect also on the sizes of the reactive site loops in these PIs. 3.6. Solvent structure The final refined structure has 58 ordered solvent molecules. Most of the solvent molecules are found on the surface of the protein. 20 water molecules are found in contact with protein atoms. Rest interact among themselves only and not directly with protein. Also no waters are involved in bridging the two subunits. An analysis of the water molecules common to both subunits was carried out. The structurally significant (invariant)

1271

water molecules that are common to both were identified. A water molecule was considered invariant, if it interacted with at least one common protein atom in the two subunits, and on superposition of the subunits along with their hydration shell, the oxygen atoms of the two waters remained closer than 1.8 Å [43]. The positions of the invariant water molecules W2, W4 and W13 of subunit A were occupied by W15, W44 and W57, respectively, in subunit B (Fig. 3A). The water molecules W2 and W4 interact with atoms N and O of Cys34, whereas W15 and W44 interact with atoms O and N of Cys61. The water molecules W2 and W15 interact with both N of Cys34 and O of Cys61 in the two subunits; whereas W4 and W44 interact with O and N of the same two residues. W4 and W44 also interact with N and O atoms of Asn36. Similarly W2 and W15 interact with O of Asp63 while W13 and W57 interact with O of Gly45. The B-factors of these water molecules are lower than the average value implying that they are tightly bound to protein atoms and may have a role in stabilizing the structure. Similar water mediated interactions involving homologous residues are observed in other BBI structures [19]. Thus, it is reasonable to conclude that these invariant water molecules are an integral part of the structure, essential for preserving the tertiary structure of the BBI. 3.7. Comparison with the reported structure of BTCI–trypsin complex The dimer structure of Vu-BBI described here has been compared with the recently reported structure of a BBI-isoinhibitor (BTCI) from the seeds of Vigna unguiculata complexed with βtrypsin [29]. The overall monomer structure of the inhibitor molecule in both the structures is similar (r.m.s.d. 1.1 and 1.2 Å for superposition of 53 Cα atoms of subunit A and 55 Cα atoms of subunit B, respectively) indicating that only the quaternary structure is affected by trypsin binding. The superposition between BTCI–trypsin complex and the Vu-BBI dimer using the coordinates of inhibitor molecule as reference are shown in Fig. 5.

Fig. 5. Overlap of the structure of BTCI–trypsin complex (2g81) on the dimer structure of Vu-BBI (2ot6). The coordinates of one of the dimers in Vu-BBI and BTCI (red) have been used to overlap the two structures. Overlap shows that trypsin (yellow) binding to one of the monomers in a dimer like Vu-BBI (grey) can have little hindrance from the second monomer.

1272

K.N. Rao, C.G. Suresh / Biochimica et Biophysica Acta 1774 (2007) 1264–1273

The residues 21, 23, 33 and 60 at the interface of BTCI–trypsin complex are different in Vu-BBI. The superposition based on inhibitor coordinates from both the structures shows that the second subunit of Vu-BBI is not involved in any serious steric clashes with trypsin molecule (Fig. 5). It is possible that the inhibition potential of Vu-BBI is unaffected by the dimer formation. Not surprisingly, the samples containing dimers we used for crystallization and other experiments showed trypsin inhibition. 3.8. Comparison with other PI structures When the atomic coordinates of the Vu-BBI were subjected to comparison in DALI database [44], three homologous PI structures were found. They are from the classical dicot BBIs, the monocot BBI from barley and the solution structure of cysteine PI, bromelain inhibitor VI. The Vu-BBI structure was also compared with all the known PI structures of Bowman–Birk family. The program ALIGN was used for the super-position of structures. As expected, the Vu-BBI structure has more similarity within the dicot family than with monocot BBI or bromelain PI. Also within the dicot family Vu-BBI overlaps best with the peaBBI monomer and least with the adzuki bean BBI. 4. Conclusion In the current study, Bowman–Birk type protease inhibitors from the seeds of Vigna unguiculata were purified to homogeneity. The isoinhibitors were characterized and differentiated in terms of molecular weight, isoelectric pH and specific protease activity. One of the isoinhibitor could be crystallized and its three-dimensional structure could be determined. Although more than 100 BBIs have been sequenced, the three-dimensional structures determined thus far are for only a few. Self-association tendency is commonly observed in BBIs. Vu-BBI is the second structure reported to form a very tight dimeric structure. The exposed hydrophobic surface patches in the monomer, strong hydrogen bonded network in the dimer structure explains the reason for BBIs to exist in dimeric as well as other multimeric forms. The comparison between the structures of Vu-BBI dimer and recently reported BTCI–trypsin complex shows that trypsin can bind to one monomer of Vu-BBI dimer without hindrance from the second monomer. The double-headed BBIs are thought to have evolved from a single-headed common ancestral protein. The intra molecular sequence homology of Vu-BBI renders further credence to this hypothesis. Acknowledgements Authors thank Department of Biotechnology, New Delhi, India for graphics and computational facilities. KNR thanks CSIR, India for senior research fellowship. References [1] M. Laskowski, M.A. Qasim, What can the structures of enzyme–inhibitor complexes tell us about the structures of enzyme substrate complexes? Biochim. Biophys. Acta 1477 (2000) 324–337.

[2] W. Bode, R. Huber, Natural protein proteinase inhibitors and their interaction with proteinases, Eur. J. Biochem. 204 (1992) 433–451. [3] R. Helland, J. Otlewski, O. Sundheim, M. Dadlez, A.O. Smalas, The crystal structures of the complexes between bovine beta-trypsin and ten P1 variants of BPTI, J. Mol. Biol. 287 (1999) 923–942. [4] P.R. Shewry, Enzyme inhibitors of seeds: types and properties, in: P.R. Shewry, R. Casey (Eds.), Seed Proteins, 1999, pp. 587–616. [5] V.V. Mosolov, T.A. Valueva, Proteinase inhibitors and their function in plants: a review, Prikl. Biohim. Mikrobiol. 41 (2005) 261–282. [6] R.F. Qi, Z.W. Song, C.W. Chi, Structural features and molecular evolution of Bowman–Birk protease inhibitors and their potential application, Acta. Biochim. Biophys. Sin.(Shanghai) 37 (2005) 283–292. [7] Y. Birk, The Bowman–Birk inhibitor. Trypsin- and chymotrypsin-inhibitor from soybeans, Int. J. Pept. Protein Res. 25 (1985) 113–131. [8] T. Ikenaka, S. Norioka, Bowman–Birk family serine proteinase inhibitors, in: A.J Barrett, G. Salvenson (Eds.), Proteinase Inhibitors, Elsevier, Amsterdam, 1983, pp. 361–374. [9] B. Prakash, S. Selvaraj, M.R. Murthy, Y.N. Sreerama, D.R. Rao, L.R. Gowda, Analysis of the amino acid sequences of plant Bowman–Birk inhibitors, J. Mol. Evol. 42 (1996) 560–569. [10] H.K. Song, Y.S. Kim, J.K. Yang, J. Moon, J.Y. Lee, S.W. Suh, Crystal structure of a 16 kDa double-headed Bowman–Birk trypsin inhibitor from barley seeds at 1.9 Å resolution, J. Mol. Biol. 293 (1999) 1133–1144. [11] F. De Leo, M. Volpicella, F. Licciulli, S. Liuni, R. Gallerani, L.R. Ceci, PLANT-PIs: a database for plant protease inhibitors and their genes, Nucleic Acids Res. 30 (2002) 347–348. [12] M.H. Werner, D.E. Wemmer, Three-dimensional structure of soybean trypsin/chymotrypsin Bowman–Birk inhibitor in solution, Biochemistry 31 (1992) 999–1010. [13] R.H. Voss, U. Ermler, L.O. Essen, G. Wenzl, Y.M. Kim, P. Flecker, Crystal structure of the bifunctional soybean Bowman–Birk inhibitor at 0.28-nm resolution. Structural peculiarities in a folded protein conformation, Eur. J. Biochem. 242 (1996) 122–131. [14] A. Suzuki, T. Yamane, T. Ashida, S. Norioka, S. Hara, T. Ikenazka, Crystallographic refinement of Bowman–Birk type protease inhibitor A-II from peanut (Arachis hypogaea) at 2.3 Å resolution, J. Mol. Biol. 234 (1993) 722–734. [15] E.Y. Park, J.A. Kim, H.W. Kim, Y.S. Kim, H.K. Song, Crystal structure of the Bowman–Birk inhibitor from barley seeds in ternary complex with porcine trypsin, J. Mol. Biol. 343 (2004) 173–186. [16] H.M. Murthy, K. Judge, L. DeLucas, R. Padmanabhan, Crystal structure of Dengue virus NS3 protease in complex with a Bowman–Birk inhibitor: implications for flaviviral polyprotein processing and drug design, J. Mol. Biol. 301 (2000) 759–767. [17] G. Lin, W. Bode, R. Huber, C. Chi, R.A. Engh, The 0.25-nm X-ray structure of the Bowman–Birk-type inhibitor from mung bean in ternary complex with porcine trypsin, Eur. J. Biochem. 212 (1993) 549–555. [18] I. Li de la Sierra, L. Quillien, P. Flecker, J. Gueguen, S. Brunie, Dimeric crystal structure of a Bowman–Birk protease inhibitor from pea seeds, J. Mol. Biol. 285 (1999) 1195–1207. [19] J. Koepke, U. Ermler, E. Warkentin, G. Wenzl, P. Flecker, Crystal structure of cancer chemopreventive Bowman–Birk inhibitor in ternary complex with bovine trypsin at 2.3 Å resolution. Structural basis of Janus-faced serine protease inhibitor specificity, J. Mol. Biol. 298 (2000) 477–491. [20] P. Chen, J. Rose, R. Love, C.H. Wei, B.C. Wang, Reactive sites of an anticarcinogenic Bowman–Birk proteinase inhibitor are similar to other trypsin inhibitors, J. Biol. Chem. 267 (1992) 1990–1994. [21] L.P. Silva, R.B. Azevedo, P.C. Morais, M.M. Ventura, S.M. Freitas, Oligomerization states of Bowman–Birk inhibitor by atomic force microscopy and computational approaches, Proteins 61 (2005) 642–648. [22] P. Kumar, A.G. Rao, S. Hariharaputran, N. Chandra, L.R. Gowda, Molecular mechanism of dimerization of Bowman–Birk inhibitors. Pivotal role of ASP76 in the dimerization, J. Biol. Chem. 279 (2004) 30425–30432. [23] M.M. Ventura, K. Mizuta, H. Ikemoto, Solvent perturbation and surface accessibility of the tryptophyl and tyrosyl groups in black-eyed pea trypsin and chymotrypsin inhibitor, An. Acad. Bras. Cienc. 56 (1984) 217–220.

K.N. Rao, C.G. Suresh / Biochimica et Biophysica Acta 1774 (2007) 1264–1273 [24] L. Morhy, M.M. Ventura, The complete amino acid sequence of the Vigna unguiculata (L.) Walp. seed trypsin and chymotrypsin inhibitor, An. Acad. Bras. Cienc. 59 (1987) 71–81. [25] O.L. Franco, R.C. dos Santos, J.A. Batista, A.C. Mendes, M.A. de Araujo, R.G. Monnerat, M.F. Grossi-de-Sa, S.M. de Freitas, Effects of black-eyed pea trypsin/chymotrypsin inhibitor on proteolytic activity and on development of Anthonomus grandis, Phytochemistry 63 (2003) 343–349. [26] S.M. de Freitas, L.V. de Mello, M.C. da Silva, G. Vriend, G. Neshich, M.M. Ventura, Analysis of the black-eyed pea trypsin and chymotrypsin inhibitor–alpha-chymotrypsin complex, FEBS Lett. 409 (1997) 121–127. [27] K.N. Rao, S.S. Hegde, R.J. Lewis, C.G. Suresh, Crystallization and preliminary X-ray diffraction studies of a Bowman–Birk inhibitor from Vigna unguiculata seeds, Acta Crystallogr. D55 (1999) 1920–1922. [28] J.A.. Barbosa, R.C. Teles, V.P. Forrer, B.G. Guimarães, F.J. Medrano, M.M. Ventura, S.M. Freitas, Crystallization, data collection and phasing of blackeyed pea trypsin/chymotrypsin inhibitor in complex with bovine β-trypsin, Acta Crystallogr. D59 (2003) 1828–1830. [29] J.A. Barbosa, L.P. Silva, R.C. Teles, G.F. Esteves, R.B. Azevedo, M.M. Ventura, S.M. de Freitas, Crystal structure of the Bowman–Birk inhibitor from Vigna unguiculata seeds in complex with {beta}-trypsin at 1.55 Å resolution and its structural properties in association with proteinases, Biophys. J. 92 (2007) 1638–1650. [30] M. Kunitz, Crystalline soybean trypsin inhibitor, J. Gen. Physiol. 30 (1947) 291–310. [31] O.H. Lowry, N.J. Rosebrough, A.L. Farr, R.J. Randall, Protein measurement with the Folin phenol reagent, J. Biol. Chem. 193 (1951) 265–275. [32] J. Navaza, P. Saludjian, AMoRe: an automated molecular replacement program package, Methods Enzymol. 276 (1997) 581–594. [33] G.N. Murshudov, A.A. Vagin, E.J. Dodson, Refinement of macromolecular structures by the maximum-likelihood method, Acta Crystallogr. D53 (1997) 240–255. [34] QUANTA, X-ray structure analysis user's reference San Diego, 1997.

1273

[35] R.A. Laskowski, M.W. MacArthur, D.S. Moss, J.M. Thornton, PROCHECK: a program to check the stereo-chemical quality of protein structures, J. Appl. Crystallogr. 26 (1993) 283–291. [36] A.M. Gatehouse, J.A. Gatehouse, D. Boulter, Isolation and characterization of trypsin inhibitors from cowpea (Vigna unguiculata), Phytochemistry 19 (1980) 751–756. [37] R.H.A. Sammour, Isolation and characterization of four isoinhibitors from cowpea (Vigna unguiculata (L.) Walp Seeds, Turk. J. Biol. 30 (2006) 207–215. [38] V.A. Hilder, R.F. Barker, R.A. Samour, A.M. Gatehouse, J.A. Gatehouse, D. Boulter, Protein and cDNA sequences of Bowman–Birk protease inhibitors from the cowpea (Vigna unguiculata Walp.), Plant Mol. Biol. 13 (1989) 701–710. [39] A.M. Harsulkar, A.P. Giri, A.G. Patankar, V.S. Gupta, M.N. Sainani, P.K. Ranjekar, V.V. Deshpande, Successive use of non-host plant proteinase inhibitors required for effective inhibition of helicoverpa armigera gut proteinases and larval growth, Plant Physiol. 121 (1999) 497–506. [40] C.A. Ryan, Protease inhibitors in plants: genes for improving defense against insects and pathogens, Annu. Rev. Phytopathol. 28 (1990) 425–449. [41] J.M. Thornton, Disulfide bridges in globular proteins, J. Mol. Biol. 151 (1981) 261–287. [42] M.M. Ventura, K. Mizuta, H. Ikemoto, Self-association of the black-eyed pea trypsin and chymotrypsin inhibitor in solution—a study by light-scattering, An. Acad. Bras. Cienc. 53 (1981) 195–201. [43] C. Sadasivan, H.G. Nagendra, M. Vijayan, Plasticity, hydration and accessibility in ribonuclease A. The structure of a new crystal form and its low-humidity variant, Acta Crystallogr. D54 (1998) 1343–1352. [44] L. Holm, C. Sander, Dali/FSSP classification of three-dimensional protein folds, Nucleic Acids Res. 25 (1997) 231–234. [45] R. Chenna, H. Sugawara, T. Koike, R. Lopez, T.J. Gibson, D.G. Higgins, J.D. Thompson, Multiple sequence alignment with the Clustal series of programs, Nucleic Acids Res. 31 (2003) 3497–3500. [46] P.J. Kraulis, MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures, J. Appl. Crystallogr. 24 (1991) 946–950.