Inhibitory properties and solution structure of a potent
Bowman–Birk protease inhibitor from lentil
(Lens culinaris, L) seeds
Enzio M. Ragg
1
, Valerio Galbusera
1
, Alessio Scarafoni
1
, Armando Negri
2
, Gabriella Tedeschi
2
,
Alessandro Consonni
1
, Fabio Sessa
1
and Marcello Duranti
1
1 Department of Agri-Food Molecular Sciences, Universita
`degli Studi, Milano, Italy
2 Department of Animal Pathology, Hygiene and Veterinary Public Health-Section of Biochemistry, Universita
`degli Studi, Milano, Italy
Bowman–Birk inhibitor (BBI) proteins are serine prote-
ase inhibitors. First isolated from soybean seeds by
Bowman [1] and subsequently characterized by Birk
et al. [2], BBIs are found in several plant sources, spe-
cially mono- and dicotyledonous seeds [3]. BBIs from
dicots usually have a molecular mass of 7–8 kDa and
are double-headed serine protease inhibitors, while
those from monocots are more variable both in size
and inhibitory sites.
Like many other cotyledonary proteins, BBIs are the
products of a multigene family within the same species
[4–6] and consequently several isoforms have been
Keywords
Bowman–Birk inhibitor; antitryptic activity;
dicotyledonous plant; Lens culinaris; nuclear
magnetic resonance
Correspondence
E. M. Ragg, Department of Agri-Food
Molecular Sciences, Universita
`degli Studi,
via Celoria 2, 20133 Milano, Italy
Fax: +39 0250316801
Tel: +39 0250316800
E-mail: enzio.ragg@unimi.it
(Received 10 May 2006, revised 29 June
2006, accepted 5 July 2006)
doi:10.1111/j.1742-4658.2006.05406.x
Bowman–Birk serine protease inhibitors are a family of small plant pro-
teins, whose physiological role has not been ascertained as yet, while
chemopreventive anticarcinogenic properties have repeatedly been claimed.
In this work we present data on the isolation of a lentil (Lens culinaris, L.,
var. Macrosperma) seed trypsin inhibitor (LCTI) and its functional
and structural characterization. LCTI is a 7448 Da double-headed tryp-
sin ⁄chymotrypsin inhibitor with dissociation constants equal to 0.54 nm
and 7.25 nmfor the two proteases, respectively. The inhibitor is, however,
hydrolysed by trypsin in a few minutes timescale, leading to a dramatic loss
of its affinity for the enzyme. This is due to a substantial difference in the
k
on
and k*
on
values (1.1 lm
)1
Æs
)1
vs. 0.002 lm
)1
Æs
)1
), respectively, for the
intact and modified inhibitor. A similar behaviour was not observed with
chymotrypsin. The twenty best NMR structures concurrently showed a
canonical Bowman–Birk inhibitor (BBI) conformation with two antipodal
b-hairpins containing the inhibitory domains. The tertiary structure is
stabilized by ion pairs and hydrogen bonds involving the side chain and
backbone of Asp10-Asp26-Arg28 and Asp36-Asp52 residues. At physiolo-
gical pH, the final structure results in an asymmetric distribution of oppos-
ite charges with a negative electrostatic potential, centred on the
C-terminus, and a highly positive potential, surrounding the antitryptic
domain. The segment 53–55 lacks the anchoring capacity found in analog-
ous BBIs, thus rendering the protein susceptible to hydrolysis. The inhibi-
tory properties of LCTI, related to the simultaneous presence of two key
amino acids (Gln18 and His54), render the molecule unusual within the
natural Bowman–Birk inhibitor family.
Abbreviations
BApNA, N-benzoyl-DL-arginine-p-nitroanilide; BBI, Bowman–Birk inhibitor; COSY-DQf, two-dimensional correlation spectroscopy double-
quantum filtered; C.S.I., chemical shift index; DSS, 2,2-dimethyl-2-silapentane-5-sulfonate sodium salt; GPpNA, N-glutaryl-L-phenylalanine-p-
nitroanilide; LCTI, Lens culinaris trypsin inhibitor; LCTI*, Lens culinaris trypsin inhibitor hydrolysed form; MD, molecular dynamics; MSTI,
Medicago scutellata trypsin inhibitor; PSTI-IVb, Pisum sativum trypsin inhibitor isoform IVb; SA, simulated annealing; sBBI, soybean
Bowman–Birk inhibitor; SFTI, sunflower trypsin inhibitor.
4024 FEBS Journal 273 (2006) 4024–4039 ª2006 The Authors Journal compilation ª2006 FEBS
identified [7,8]. Despite their pronounced microhetero-
geneity, BBIs share a relatively high degree of sequence
homology, especially in the inhibitory domains, and a
highly conserved disulphide bridge network [9], form-
ing a consensus motif (Prosite code: PDOC00253).
There have been various hypotheses on the physiolo-
gical function of BBIs, including defence and protec-
tion, developmentally regulatory and sulphur-storage
roles, with no conclusive definition as yet [10]. Plant
cell biology data on BBIs biosynthesis and transloca-
tion to the secretory pathway are also missing.
From the inhibitory viewpoint most BBIs, especially
those from dicotyledonous seeds, have a double-
headed structure bearing two independent proteinase
binding sites, often one trypsin and one chymotrypsin
domain. Various synthetic peptides consisting of a sin-
gle inhibitory domain and bearing the inhibitory activ-
ity have been produced and this has served to identify
the role of specific amino acid residues in the protein-
ase inhibition [11].
The renewed interest for this class of protease inhibi-
tors [12] is mainly based on the findings that BBIs may
act as cancer preventive and suppressing agents in a
wide variety of in vitro and in vivo model systems [13].
In some cases, as in the treatment of oral leukoplakia
lesions, the use of BBIs has reached phase II of clinical
trials [14,15]. Besides the anticarcinogenic effects, BBIs
also showed anti-inflammatory activity, by inhibiting
the inflammation-mediating proteases [16]. More
recently, a number of patents on the use of BBIs
against various apparently unrelated diseases have
appeared [17–19]. The molecular basis of these BBI
activities has not been established so far, however,
because a high protease activity has been shown to be
connected with tumour formation and other diseases
associated with angiogenesis; it has been suggested
that the chemopreventive action might be related to
the protease, especially antichymotrypsin, inhibitory
activity [20].
There has been more and more research into the
involvement of specific food proteins and peptides as
causative agents in the prevention and control of
various diseases, many of which are related to the
Western lifestyle, such as obesity, diabetes and cardio-
vascular diseases. Furthermore, the search for novel
biologically active protein molecules and their exploita-
tion as drugs or nutraceutical agents imply their func-
tional and structural characterization. Based on these
considerations, the identification of novel BBI inhibi-
tors, either as natural compounds or synthetic pep-
tides, and the elucidation of their structural and
functional properties, is extremely important. A recent
review dealt with legume-derived inhibitors [21].
We present here our results on the isolation, func-
tional and structural analysis of a BBI from lentil
(Lens culinaris L. var. Macrosperma) seeds. Our isola-
tion procedure yielded a protein in sufficient amounts
and purity to obtain the complete amino acid sequence
and
1
H-NMR chemical shift assignment, as well as the
measurement of interproton distances, by means of
homonuclear correlation and nuclear Overhauser
effect experiments. The experimental values were then
applied as restraints for molecular dynamics calcula-
tions leading to the three-dimensional solution
structure of the protein. Kinetic studies have shown
that the isolated BBI from Lens culinaris seeds (Lens
culinaris trypsin inhibitor; LCTI) is characterized by
unusual inhibitory properties within the family of nat-
ural Bowman–Birk inhibitors.
Results
Purification, mass spectrometry analysis and
primary structure determination of LCTI
The purification of LCTI from lentil seeds involved var-
ious chromatographic steps, including a final affinity
chromatography step on agarose-immobilized trypsin.
The antitrypsin activity was measured at every purifica-
tion step by N-benzoyl-dl-arginine-p-nitroanilide
(BApNA) hydrolysis assays. Purity was greater than
98%, as proved by RP-HPLC and SDS ⁄PAGE (not
shown). The final product was characterized by N-ter-
minal amino acid sequencing, mass spectrometry
(MALDI-TOF) (Fig. 1), amino acid sequence analysis
of Lys-C generated fragments and
1
H-NMR. The isola-
ted 67 amino acid protein had the same primary struc-
ture as a recently published BBI, named LCI1.7,
extracted from Lens culinaris var. Microsperma [22],
with the exception of a C-terminal missing glutamic
acid residue (SwissProt Acc. No. Q8W4Y8). The
molecular mass calculated from the primary structure
(7448.29 Da assuming seven disulfide bonds) agrees
with the one determined by mass spectrometry
(7446.63 Da). In the amino acid sequence (Fig. 2),
several characteristic regions could be identified, inclu-
ding 14 Cys residues and the consensus sequences
CTR(K)SxPPTC and CxY(L ⁄R)SxPxQ(K)C for the
antitrypsin and antichymotrypsin sites, respectively [5].
Figure 2 shows the amino acid sequence alignment of
LCTI with other inhibitors of the Leguminosae family
of known 3D structure. Sequence identity of lentil BBI
ranged from a minimum of 47% with Lima bean BBI
to a maximum of 82% with pea BBI. Major differences
are located at the N- and C-termini. Identities or con-
servative substitutions were observed at the inhibition
E. M. Ragg et al.Inhibitory properties and NMR structure of a lentil BBI
FEBS Journal 273 (2006) 4024–4039 ª2006 The Authors Journal compilation ª2006 FEBS 4025
sites, with the only exception being Medicago scutellata
BBI, which, because it is a double trypsin inhibitor [23],
has an arginine residue instead of a tyrosine or leucine
in the position P1 of the antichymotryptic site (P and P¢
nomenclature according to Schechter and Berger [24]).
Antitrypsin and antichymotrypsin activity assays
The inhibitory activity of LCTI was determined at
pH 8.2, by monitoring the hydrolysis of the chromo-
genic substrates BApNA and N-glutaryl-l-phenylalan-
ine-p-nitroanilide (GPpNA) in the presence of bovine
trypsin and a-chymotrypsin, respectively, and increas-
ing amounts of LCTI. Figure 3 reports the amount of
hydrolysed BApNA as a function of time. In the pres-
ence of LCTI, two distinct kinetic regimes with differ-
ent rate constants were present. This effect was more
evident for equimolar LCTI ⁄trypsin mixtures, whereas
in the case of low amounts of LCTI the first kinetic
phase vanished after a few minutes of the reaction.
N-terminal amino acid sequencing of the proteolytic
fragments (see below) proved that hydrolysis actually
occurred in the antitrypsin site at the cleavable N-ter-
minal P1-P1¢bond (not shown).
The kinetic model assumed (Scheme 1) implies the
formation of a 1 : 1 complex [25] and is the simplest
one able to fit with sufficient accuracy the experimental
results.
The k
cat
⁄K
M
ratio was derived by fitting the experi-
mental data in the absence of inhibitor and agreed
with k
cat
and K
M
independently determined by
means of standard Lineweaver–Burk analysis. At
[LCTI] ⁄[trypsin] ¼0.38, as LCTI is hydrolysed within
a few minutes (Fig. 3, curve 1), its inhibitory activity is
mainly due to Lens culinaris trypsin inhibitor hydro-
Fig. 2. Sequence alignment of LCTI with other inhibitors from Leguminosae family of known 3D structure. Accession numbers are from
Brookhaven Protein Data Bank and refer to the following proteins: LCTI (2AIH_lens, this work); MSTI (1MVZ_Medicago); PSTI-IVb (1PBI_
pea); sBBI (1BBI_soya); lima bean trypsin inhibitor (LBTI) (1H34_lima). T and CT denote P1 residues in the antitrypsin and antichymotrypsin
sites, respectively.
Fig. 3. Hydrolysis of 213 lMBApNa as function of time in the pres-
ence of 0.1 lMtrypsin (pH 8.2, 37 C) and the following amounts
of LCTI: 0 lM(¯, curve 0), 0.038 lM(*, curve 1), 0.11 lM(·, curve
2), 0.225 lM(h, curve 3).
Fig. 1. MALDI-TOF mass spectrum of LCTI. Sin: sinapinic acid. The
insert shows an expansion of the molecular peak.
Scheme 1.
Inhibitory properties and NMR structure of a lentil BBI E. M. Ragg et al.
4026 FEBS Journal 273 (2006) 4024–4039 ª2006 The Authors Journal compilation ª2006 FEBS
lysed form (LCTI*), leading to an accurate measure-
ment of k*
on
and k*
off
. At [LCTI] ⁄[trypsin] ¼1.1 and
2.25 the conversion of LCTI into LCTI* (Fig. 3,
curves 2 and 3), allowed the simultaneous computation
of k
off
,k*
on
and k*
off
. The k
on
value was assumed by
analogy with soybean BBI [26]. The two dissociation
constants (K
d
¼k
off
⁄k
on
and K*
d
¼k*
off
⁄k*
on
, relative
to the virgin and modified inhibitor, respectively) were
calculated on the basis of the derived kinetic constants.
The results, obtained after simultaneous fitting of all
the experimental curves, are reported in Table 1.
The same type of kinetic analysis was applied for
assaying the antichymotryptic activity. In analogy to
the previous experiment, a partial loss of chymotrypsin
inhibitory activity was observed, but it was less evident
due to a lower rate of hydrolysis and, more import-
antly, to a minor difference between K
d
and K*
d
(Table 1).
1
H-NMR sequential assignments and secondary
structure determination
A total of 62 NH-H
a
interactions were detected
through the analysis of the TOCSY and two-dimen-
sional correlation spectroscopy double-quantum
filtered (COSY-DQf) experiments, allowing the identifi-
cation of the characteristic amino acid spin systems.
The arginine residues were identified through the
connectivities with their e-NH protons. Two amide
protons, belonging to spin systems of the type NH-
CH
a
-CHb
2and later identified as Asp10 and Asp36,
were detected at very low field (11.48–11.49 p.p.m.).
Sequential assignments were performed using well
established procedures [27] on the basis of the
d
NN
(i,i+1) and d
aN
(i,i+1) interactions observed in the
NOESY experiments. Other weak connectivities were
detected in the TOCSY and NOESY spectra, where
the sequential assignment pathway between residues 12
and 16 was found split in two, thus suggesting that a
minor form of LCTI (10%) was present in the solu-
tion. As residue 16 is located in the antitrypsin site,
this form was attributed to LCTI*. Additional reso-
nances, attributed to LCTI*, were found for Thr53
and His54. This finding is consistent with the presence
of a minor peak in the mass spectra, which corres-
ponds to a mass increase of 18 Da, as expected from
the hydrolysis of one peptide bond (Fig. 1, insert).
Moreover, minor peaks corresponding to the sequence
starting with Ser17 were detected in the previously
mentioned amino acid sequence analysis (not shown).
Indeed, both NMR and MS spectra showed that the
amount of hydrolysed form increased when the inhib-
itor was kept in solution at pH 3.1 for few days,
suggesting a particular intrinsic lability of the Arg16-
Ser17 bond to hydrolysis at acidic pH.
The sequential inter-residue interactions provided a
means for defining the cis-trans conformation for the
two pairs of contiguous prolines. Thus, Pro20 and
Pro46 were found in trans-conformation, because of
the strong Pro19H
a
-Pro20H
d
and Pro45H
a
-Pro46H
d
interactions, whereas Pro19 and Pro45 were classified
as cis by means of the detected sequential d
aa
(i,i+1)
interactions, respectively, with Gln18 and Asn44.
No d
(i,i+3)
interaction was observed, thus excluding
the presence of any helical segment or type-I ⁄II turn,
within the protein. Figure 4 reports the relevant
sequential NOE interactions for the two inhibitory
regions, located in the Thr11-Val25 and Lys37-Tyr51
segments. They are characterized by clusters of strong
d
aN
(i,i+1) and weak d
aN
(i,i+1) interactions and,
Table 1. Kinetic and thermodynamic parameters for the inhibitory activity of LCTI against bovine trypsin (BT) and a-chymotrypsin (BCT),
measured at pH 8.2. k
on
·10
)6
values taken from [26].
k
on
·10
)6
(M
)1
Æs
)1
)k
off
·10
3
(s
)1
)k*·10
)3
(M
)1
Æs
)1
)k*
off
·10
3
(s
)1
)K
d
·10
9
(M)K*
d
·10
9
(M)K
hyd
BT 1.1 0.60 ± 0.15 2.0 ± 0.8 0.82 ± 0.01 0.54 ± 0.1 410 ± 95 759
BCT 0.2 1.45 ± 0.21 12.5 ± 5.0 0.40 ± 0.05 7.25 ± 1.08 32 ± 13 4.4
Fig. 4. LCTI b-hairpin elements (segments Thr11-Val25 and Lys37-
Tyr51), with observed NOE interactions (double-arrow) and hydro-
gen bonds involving the slowly exchanging amide protons (dotted
line). T and CT denote the antitrypsin and antichymotrypsin sites,
respectively.
E. M. Ragg et al.Inhibitory properties and NMR structure of a lentil BBI
FEBS Journal 273 (2006) 4024–4039 ª2006 The Authors Journal compilation ª2006 FEBS 4027
together with several detected long-range d
aa
and d
aN
interactions, define two b-hairpin secondary structure
elements.
Figure 5A reports the chemical shift index (C.S.I.)
for H
a
, in comparison with the corresponding soybean
BBI values. Random coil values were taken from [28].
Positive values indicate a residue propensity for exten-
ded or b-sheet structure [29]. Thus, the C.S.I. analysis
identifies six b-sheet regions. The LCTI data are very
similar to soybean, with the exception of the 26–29
segment, with positive C.S.I. values more similar to
Medicago scutellata trypsin inhibitor (MSTI) [23] and
the 49–55 segment, characterized by a marked reduc-
tion in propensity for an extended conformation.
At the end of the antitrypsin and antichymotrypsin
b-hairpin, the segments Thr21-Cys22 and Gln47-Gln49
experience long-range interactions, respectively, with
the segment Thr53-Lys55 and Arg28-Glu29. In these
cases, however, the pattern of the observed NOE inter-
actions is not sufficient to indicate the presence of
additional b-strands, but rather a spatial proximity of
these short segments to the b-hairpins.
Measured values of the vicinal coupling constants
provided additional restraints for the corresponding
dihedral angles, to be introduced in the restrained
molecular mechanics and dynamics calculations. The
N- and C-terminus segments appeared rather structure-
less, with no detected long-range NOE up to the
Cys8-Cys61 disulphide bond.
Deuterium exchange experiments and
temperature coefficient measurements
The analysis of the secondary structure suggested the
presence of several hydrogen bonded amide protons,
mainly located near the two inhibitory sites. Deuter-
ium exchange experiments were thus performed, by
directly dissolving the protein in D
2
O and acquiring a
series of one-dimensional spectra at room temperature.
After a few hours after dissolution, 11 amide protons
were still observable and could easily be assigned. In
order to fully characterize the solvent accessibility of
the amide protons, the chemical-shift temperature coef-
ficients (Dd
NH
⁄DT) were determined by performing a
series of TOCSY experiments at various temperatures
(Table 2). As absolute values less than 5 p.p.b.ÆK
)1
indicate solvent protection, the temperature coefficients
are a complementary measurement for the more direct
deuterium-exchange experiments and are particularly
suitable for amide protons in the fast-exchange regime.
The analysis of the experimentally determined values,
and their implication with the peptide tertiary struc-
ture, will be discussed below.
Solution structure of LCTI
The observed NOEs also provided information on the
global protein folding. All the measured vicinal coup-
ling constants and NOE interactions were translated
into restraints for the generation of the solution struc-
ture. Statistics for the total amount of experimental
data are reported in Table 3.
A simulated annealing (SA) procedure was used
starting from a randomly generated linear polypeptide
chain. The actual protocol is described in detail below.
Initially, no disulphide bond definition was introduced
and a limited subset of distances, derived from
NOESY experiments performed at short mixing times
(t
mix
¼80 ms), was utilized for generating a starting
restraints set, together with ideal values for /,wdihed-
ral angles. One hundred and fifty structures were
thus obtained and analysed in terms of total energy,
Fig. 5. (A) Comparison of C.S.I. values for LCTI (white) and
soybean BBI (black) calculated with 3-point smoothing. Data for
soybean BBI were taken from Biological Magnetic Resonance Data
Bank (Acc. no. 1495); (B) Local rmsd values calculated from the
superimposition of the 20 NMR-derived structures. b-hairpin
regions are underlined.
Inhibitory properties and NMR structure of a lentil BBI E. M. Ragg et al.
4028 FEBS Journal 273 (2006) 4024–4039 ª2006 The Authors Journal compilation ª2006 FEBS