Conformationally constrained analogues of endogenous tripeptide inhibitors of zinc metalloproteinases

Article abstract of DOI:10.1016/S0223-5234(00)01192-2

Two diastereomeric furan-2-carbonylamino-3-oxohexahydroindolizino[8,7-b]indole carboxylates, highly constrained analogues of endogenous pyroglutamyl tripeptide inhibitors of snake venom endopeptidases, have been prepared as potential inhibitors of adamalysin II and matrix metalloproteinases. They proved to be inactive against adamalysin II and weak inhibitors of gelatinase A, gelatinase B, stromelysin 1 and human neutrophyl collagenase. Evaluation of the mode of binding of the (2R,5S,11bR) isomer in the active site of adamalysin II suggests that the decrease of potency may be due to the reorientation of the acylamino chain in three of the heterocyclic nucleus, to a short contact at the entrance of the S′₁ hydrophobic cleft and to the loss of flexibility of the tetracyclic nucleus in the P′₁, P′₂ region of the inhibitor, which prevents optimal arrangement in the S′₁ specificity subsite.

Full text of DOI:10.1016/S0223-5234(00)01192-2

Eur. J. Med. Chem. 36 (2001) 43–53
´
© 2001 Editions scientifiques et me´dicales Elsevier SAS. All rights reserved
PII: S0223-5234(00)01192-2/FLA
Original article
Conformationally constrained analogues of endogenous tripeptide inhibitors of
zinc metalloproteinases
Silvana D’Alessioa, Carlo Gallinab*, Enrico Gavuzzoc, Cesare Giordanod, Barbara Gorinie,
Fernando Mazzae, Mario Paglialunga Paradisid, Gabriella Paninid, Giorgio Pochettic
aPolifarma Research Center, Via Tor Sapienza 138, 00185 Rome, Italy
bDipartimento di Scienze del Farmaco, Istituto di Scienze del Farmaco, Universita` G. d’Annunzio,
Via dei Vestini, 66100 Chieti, Italy
cIstituto di Strutturistica Chimica del CNR, C.P. n. 10, 00016 Monterotondo Stazione, Rome, Italy
dCentro di Studio per la Chimica del Farmaco del CNR, Universita` La Sapienza, P. le A. Moro 5, 00185 Rome, Italy
eDipartimento di Chimica, Ingegneria Chimica e Materiali, Universita` degli Studi, 67010 L’Aquila, Italy
Received 10 September 1999; accepted 11 September 2000
Abstract – Two diastereomeric furan-2-carbonylamino-3-oxohexahydroindolizino[8,7-b]indole carboxylates, highly constrained ana-
logues of endogenous pyroglutamyl tripeptide inhibitors of snake venom endopeptidases, have been prepared as potential inhibitors
of adamalysin II and matrix metalloproteinases. They proved to be inactive against adamalysin II and weak inhibitors of gelatinase
A, gelatinase B, stromelysin 1 and human neutrophyl collagenase. Evaluation of the mode of binding of the (2R,5S,11bR) isomer
in the active site of adamalysin II suggests that the decrease of potency may be due to the reorientation of the acylamino chain in
three of the heterocyclic nucleus, to a short contact at the entrance of the S%1 hydrophobic cleft and to the loss of flexibility of the
tetracyclic nucleus in the P1%, P2% region of the inhibitor, which prevents optimal arrangement in the S1% specificity subsite. © 2001
´
Editions scientifiques et me´dicales Elsevier SAS
2-acylamino-3-oxohexahydroindolizino[8,7-b]indole carboxylates / crystal structure / adamalysin II / matrix metalloproteinases
1. Introduction
which binds in the enzyme active site and a zinc
binding function, mainly hydroxamate, carboxylate,
phosphonate and phosphinate, which interacts with
the catalytic zinc ion. Peptide derived inhibitors, how-
ever, generally suffer from undesirable physical prop-
erties, mainly poor solubility, low stability to
hydrolytic enzymes, fast biliary clearance and poor
oral availability. Therefore, design and synthesis of
non-peptidic MMP inhibitors, suitable as drug candi-
dates, has grown as an active research field. Herein,
we report on structural modifications of endogenous
snake venom peptides, pointing to generate non-pep-
tidic, conformationally constrained inhibitors of ma-
trix and snake venom metalloproteinases.
The matrix metalloproteinases (MMPs) are a fam-
ily of zinc endopeptidases [1] involved in tissue re-
modelling and connective tissue turnover. Aberrant
regulation of these enzymes, particularly collagenases,
stromelysins and gelatinases, has been implicated in
several pathologies [2–5] including rheumatoid
arthritis and tumour invasion and metastasis. Thus,
inhibition of MMPs enzymatic activity has become an
important pharmacological target for the treatment of
a range of diseases for which no specific or effective
drugs are available.
Many structural classes of potent MMP inhibitors
[2–10] incorporate a substrate-like peptide chain,
On the basis of common structural and functional
properties, snake venom hemorragic zinc endopepti-
dases, named reprolysins, have been grouped in the
superfamily of metzincins [11], together with astacins,
* Correspondence and reprints.
E-mail address: cgallina@unich.it (C. Gallina).

44
S. D’Alessio et al. / European Journal of Medicinal Chemistry 36 (2001) 43–53
in the low micromolar range [15] and it has been
proposed that they may function as temporary in-
hibitors of the venom proteinases during storage in
the snake glands. Structural basis for binding interac-
tions of pyroglutamyl tripeptides with metallo-
proteinases has been clarified by solution of the
crystal structure of the pyroGlu–Asn–Trp–OH
atrolysin C complex [16].
Owing to the close resemblance between reprolysins
and MMPs, snake venom pyroglutamyl peptides and
their analogues have been previously studied as in-
hibitors of the human zinc proteinases [15]. The phos-
phonate ligand 1a, as well as the carboxylate
analogues 1b and 2 (figure 1), are examples of metal-
loproteinase inhibitors obtained by structural varia-
tion of pyroGlu–Asn–Trp–OH. Crystal structures of
adamalysin II complexed with phosphonate 1a [17]
and with carboxylates 1b and 2 [18] have been used as
starting models for the conformationally constrained,
non-peptidic analogous 4a, as a possible inhibitor of
adamalysin II and MMPs.
serralysins and MMPs. The crystal structure [12, 13]
of adamalysin II, a reprolysin from Crotalus adaman-
teus, shows highly conserved overall topology with
MMPs, namely MMP-8 and very close similarity at
the active site. The catalytic domains of all these
enzymes contain an essential zinc ion coordinated by
the imidazole groups of three conserved histidine
residues of the consensus sequence HEXXHGXXH,
lying at the bottom of the binding cleft. They present
substantially identical core structures, while impor-
tant variations in the surface loops originate consider-
able differences in the specificity subsites. The S%
1
subsite, generally consisting of hydrophobic residues,
varies considerably in size and shape and is consid-
ered to be the primary specificity site, since it can
strongly influence recognition of substrates and
inhibitors.
Small pyroglutamyl peptides are present in millimo-
lar concentration, together with reprolysins, in the
crude venom of crotalides [14]. Two of these endoge-
nous peptides, pyroGlu–Asn–Trp–OH and py-
roGlu–Glu–Trp–OH, isolated from the venom of C.
atrox, were demonstrated to be inhibitors of atrolysin
Cyclization of the Trp based carboxylate inhibitor
1b into the cyclo-Trp analogue 2, by means of a
Figure 1. Linear (1a, 1b), tricyclic (2) and tetracyclic (4a) inhibitors of zinc-proteinases. Key bonding interactions in the complex of
2 with adamalysin II, involving the large hydrophobic pocket in S%, the catalytic zinc ion and the residues Lys-106, Ile-108 and
1
Leu-170 of the enzyme, are reported in the schematic representation 3.

S. D’Alessio et al. / European Journal of Medicinal Chemistry 36 (2001) 43–53
45
methylene group joining the tryptophan amino group
with the indole 2 position, caused an increase of the
binding affinity for adamalysin II of one order of
magnitude [18]. Since both 1b and 2 inhibitors bind in
the active site of the enzymes, adopting substantially
identical conformations [18], the increase of potency
of the cyclic analogue can be attributed to the de-
crease of conformational entropy for the protease
binding. The schematic representation 3 illustrates the
key enzyme–inhibitor interactions observed in both
the crystal structures [18]. The carboxylate oxygens
legate the catalytic zinc ion in a bidentate mode, while
the Trp side chain partly fills the primary specificity
subsite S1%. Leu CO, Leu NH and furan-2-carbonyl
CO groups form H-bonds with complementary func-
tions of the protein. The Leu side chain and the furan
ring are involved in loose binding interactions with
the surrounding groups.
we decided to embark on the preparation of the
(2R,5S,11bR)-2-(furan-2-carbonyl)amino-5-carboxy-
3-oxo-2,3,5,6,11,11b-hexahydro-1H-indolizino[8,7-b]-
indole 4a. This highly constrained heterocyclic nu-
cleus could supply a new scaffold for non-peptide
inhibitors of adamalysin II and MMPs. The
(2S,5S,11bR) stereo isomer 4b (figure 2) was also
prepared to evaluate the consequence of the change of
configuration at the 2 position.
2. Chemistry
The two desired compounds 4a and 4b were pre-
pared by following previously reported [19] proce-
dures for synthesis of 2-benzyloxycarbonylamino-
3-oxo-11b-hexahydroindolizino[8,7-b]indole-5-car-
boxylate nucleus with complete control of the stereo-
chemistry (figure 2).
The distance between Leu aCH and the bridge
methylene carbon in the binding conformation 3 is ca.
Reaction of N-benzyloxycarbonyl- -homoserine
D
,
2.9 A. Joining of these carbons with a new methylene
aldehyde benzylester 5a with H–L–Trp–OMe, under
conditions of kinetic control [20, 21] of the Pictet–
Spengler reaction, gave a mixture of the trans (6a)
and cis (6c) b-tetrahydrocarbolines, richer in the cis,
less stable isomer. The trans lactame 7a, required for
our purposes, derives from the less abundant, more
stable carboline 6a, which could not be separated
from the prevailing diastereoisomer 6c, by standard
chromatographic procedures. Therefore, the 6ac mix-
ture was directly refluxed in toluene in the presence of
10 molar equiv. of trifluoroacetic acid. Under condi-
tions of acid catalysis, the 6ac kinetic mixture richer
in the cis carboline 6c was transformed into a 7ac
mixture richer in the more stable trans lactame 7a,
which could be secured by silica-gel chromatography.
Tetrahydro b-carboline 6b and lactame 7b were pre-
pared as previously reported [19].
,
bridge shortens the distance to 2.4 A. This cyclization,
combined with deletion of the Leu side chain, leads to
the conformationally constrained analogue 4a, where
the cyclo-Trp unit of 2 is replaced with the highly
constrained 2-amino-3-oxohexahydroindolizino[8,7-
b]indole-5-carboxylate skeleton. In a putative com-
plex of 4a in the adamalysin II active site, all the
binding interactions represented in 3 could be only
slightly modified, provided that the 2, 5 and 11b
stereogenic centres are in the S, R, S configuration.
This additional cyclization points to the preorgani-
zation of the ligand into a conformation allowing zinc
coordination by the carboxylate and burying of the
heterocyclic moiety into the hydrophobic cleft in S1%.
Since this subsite involves a large area of interaction,
presenting different residues and different capacity of
reorganization upon ligand binding in the various
MMPs, it was expected that the increase of rigidity of
the molecule could greatly affect the potency and
selectivity of the inhibitor. It could be anticipated,
however, that the increase in size due to the 1-methyl-
ene group would require a small opening of the
protein chain at the entrance of the hydrophobic cleft
in order to allow the accommodation of the more
bulky heterocyclic nucleus. Additional changes con-
cern the conformational preferences of the furan-2-
carbonylamino chain that would be affected by the
creation of the new ring. In order to evaluate the
effects on binding of this conformational restriction,
Conversion of esters 7a and 7b into the carboxylic
acids 4a and 4b required three steps: (1) hydrogenoly-
sis of the benzyloxycarbonyl group by using ammo-
nium formate in the presence of Pd/C; (2) acylation
of the resulting amines with furan-2-carboxylic acid
by dicyclohexylcarbodiimide/1-hydroxybenzotriazole
activation [22]; and (3) hydrolysis of the methylesters
8a and 8b under mild alkaline conditions [19]. Stereo-
chemistry of intermediate and final compounds has
1
been attributed on the basis of H NMR spectra,
taking into account the assignments for the previously
reported carbolines 6b, 6d and lactames 7b and 7d
[19].

46
S. D’Alessio et al. / European Journal of Medicinal Chemistry 36 (2001) 43–53
Figure 2. Synthesis of the diastereomeric furan-2-carbonylamino-3-oxohexahydroindolizino[8,7-b]indole carboxylates 4a and 4b.
3. X-ray analysis
presents a pseudomirror plane passing through the C5
and C16 atoms; its conformation can be described as
With the aim to unequivocally confirm the configu-
ration at the 2, 5 and 11b stereogenic centres of the
tetracyclic compound 4a, and also to determine its
solid-state conformation for comparison with the bind-
ing conformation of 2 in the active site of adamalysin
II, the present X-ray analysis was performed on crystals
of 4a.
A perspective view of the molecular conformation of
4a, together with adopted numbering scheme (top) and
the relevant internal torsion angles of C and D rings
(bottom) are reported in figure 3. The pentatomic D
ring is characterized by a pseudomirror plane bisecting
the amide bond and passing through the C2 atom; its
conformation can be described by an envelope with the
,
a sofa with the C5 atom displaced of 0.606 A from the
plane of the other five ring atoms, on the same side of
the H bound to C11b. The fusion between C and D
rings is of the quasi-trans type, since the internal torsion
angles of junction have opposite signs, being −18.8 and
34.0° for the penta and hexa-atomic rings, respectively.
The carboxyl group is axial and lies on the same side
of the H bound to C11b. The equatorial character of
the furan-2%-carbonylamino moiety is weakened by the
low puckering of the pentatomic ring; the two torsion
angles preceding and following the carbon bound to the
furan ring are 11.8 and −21.1°, respectively. The
endocyclic amide bond is slightly distorted from trans
planarity, the torsion angle C2–C3–N4–C5 being
−169.7°; this small torsion, however, is not caused by
non-planarity of nitrogen, since the sum of valence
angles around its three substituents is 359.6°.
,
C1 atom at the flap, displaced of 0.369 A from the plane
of the other four ring atoms, on the opposite side of
the H bound to the C11b. The hexatomic C ring

S. D’Alessio et al. / European Journal of Medicinal Chemistry 36 (2001) 43–53
47
ular H-bonds. The first occurs between the carbonyl
oxygen O20 and the indole N15 of a molecule trans-
lated of
1 U along the c axis, with values
,
,
O20···N15=2.822 A, O20···H=1.95
A
and
O20···HꢀN15=167°. The other occurs between the
carbonyl O3 of the reference molecule and the hy-
droxyl O18 of a screw related molecule along c,
translated of 2b–c units, with values O3···O18=2.706
,
,
A, O3···H=1.70 A and O3···HꢀO18=161°. The
crystal packing is completed by van der Waals
contacts.
4. Results and discussion
The new compounds were tested for inhibitory ac-
tivities against adamalysin II, MMP-2, MMP-3,
MMP-8 and MMP-9 by evaluation of the residual
enzyme activity by continuous fluorimetric assays
(table I).
In contrast to the parent carboxylates 1b and 2, the
tetracyclic analogues 4a and 4b did not show inhibit-
ing activity against adamalysin II at the solubility
limit of 13 mM, in the conditions of the assay. Low
levels of inhibition against the tested MMPs, close to
that reported for 1b and 2 [18], were however
retained.
The observed loss of affinity of 4a for adamalysin
II can probably be explained by considering that the
introduction of the fourth condensed ring, in addition
to the desired preorganization of the ligand, bears
some detrimental effects. The best fitting between the
crystal structure of 2 in the complex with adamalysin
II and the small molecule crystal structure of 4a
(figure 4), shows that their hydrophobic heterocyclic
Figure 3. Top: perspective view of the crystal conformation of
4a together with the adopted numbering scheme; bottom:
internal torsional angles (°) of the C and D rings.
The peptide bond connecting the fused ring system
to the furan ring is slightly distorted from trans pla-
narity, the C2–N19–C20–C21 torsion angle being
170.0°; furthermore, it is coplanar with the furan ring,
as can be deduced from the N19–C20–C21–O25
torsion angle of 173.3°, indicating possible electronic
conjugation between the pentatomic double bonds
and the peptide carbonyl bond. The conformation of
the carbonylamino group can be described by the
C3–C2–N19–C20 and N4–C3–C2–N19 torsion an-
gles of −50.3 and 135.5°, respectively.
Crystal packing is characterized by two intermolec-
Table I. Inhibition of adamalysin II, gelatinase A (MMP-2), gelatinase B (MMP-9), human neutrophyl collagenase (MMP-8) and
stromelysin I (MMP-3) by tetracyclic carboxylate ligands 4a and 4b
Inhibitor no.
% Inhibition at 13 mM
Adam. II
MMP-2
MMP-9
MMP-8
MMP-3
4a
4b
1b
2
n.i.a
10
5
5
15
15
30
5
20
30
20
10
n.i.a
n.i.a
10
\50b
\50c
25
\50d
30
a n.i., not inhibitory at 13 mM.
b IC50=10 mM [18].
c IC50=1 mM.
d IC50=1.8 mM [18].

48
S. D’Alessio et al. / European Journal of Medicinal Chemistry 36 (2001) 43–53
structures of the known inhibitors 1a, 1b and 2 in the
adamalysin II active site show that the acylamino
chains, involved in H-bonds with Leu-170 NH and
Lys-106 CO of the protein, adopt extended b-strand
conformations. In a putative complex containing the
crystal conformation of 4a in the active site of
adamalysin II, these H-bonds would be lost and could
only be restored at the energetic cost required by the
reorientation of the chain [24].
The putative complex of the crystal conformation
of 4a in the active site of adamalysin II also shows
,
that the b hydrogen at C1 is too close (ca. 2 A
distance) to the carbonyl oxygen of Pro-168, giving
rise to an unfavourable nonbonded contact. It has
been reported [25] that the carbonyl oxygen of Pro-
238 in MMP-1, homologous of Pro-168 in adamalysin
II, can undergo a small but significant displacement in
order to avoid unacceptable contacts with groups of
the inhibitor, providing evidence for flexibility of the
protein chain in this region. Analogous flexibility of
the protein in MMP-3, MMP-8 and MMP-9 could
possibly explain the residual inhibiting activity shown
by 4a against these enzymes. The loss of measurable
affinity for adamalysin II could be attributed to larger
rigidity of the homologous chain in this enzyme.
Not only the protein chain, but also the inhibitor
flexibility, can strongly affect enzyme affinity. Previ-
ous work, both in the field of metalloproteinases [26]
and streptavidin [27] ligands, has shown that reten-
tion of considerable ligand flexibility in the bound
state is important to achieve optimal binding. In
Figure 4. Best fitting between small-molecule crystal structure
of 4a (filled bonds) and crystal structure of 2 in the complex
with adamalysin II, based on the carboxylate carbon and C5,
C7 and C10 atoms. An arrow indicates the b-hydrogen at C1
,
forming a short contact (ca. 2 A) with the carbonyl oxygen of
Pro-168 in the putative complex with the enzyme.
nuclei and carboxylate groups are practically overlap-
ping. Both the C2 carbon and the adjacent NH of 4a,
are very close to the positions of the corresponding
atoms of 2, confirming the desired preorganization of
the constrained ligand according to the binding con-
formation of 2. These features would guarantee con-
temporary occupation of the S%1 specificity site, zinc
chelation and orientation of the acylamino chain to-
ward the S%,S% sites. However, the C1 methylene
particular, the rigidity of the P%1, P% region would
2
hinder the conformational adjustments of the in-
hibitor necessary to avoid the short contact of the C1
methylene and to optimise the bonding interactions
with the protein.
Contrary to expectation, inhibiting properties of
the diastereomeric analogue 4b were substantially
similar to that of 4a. This result cannot be explained
on the basis of the same mode of binding and could
be attributed to non-specific interactions with the
enzymes.
2
3
bridge, indicated with an arrow, increases size and
rigidity of the molecule in the P%1, P% region and
2
determines a preferred conformation of the furan-2-
carbonylamino chain ca. 90° rotated around the
C2ꢀN19 bond.
5. Conclusions
It has been recently pointed out that metallo-
proteinases, as well as aspartic, serine and cysteine
proteinases, require an extended b-strand conforma-
tion for their active site directed inhibitors and sub-
strate analogues [23]. Accordingly, the crystal
Conformationally constrained analogues 4a and 4b
of endogenous pyroglutamyl tripeptide inhibitors of
snake venom zinc endopeptidases proved to be inac-
tive against adamalysin II and weak inhibitors of

S. D’Alessio et al. / European Journal of Medicinal Chemistry 36 (2001) 43–53
49
MMP-2, MMP-3, MMP-8 and MMP-9. Evaluation
of the hypothetical mode of binding of 4a in the
active site of these enzymes suggests that the decrease
in affinity may be due to the reorientation of the
acylamino chain in 3, as shown in the crystal confor-
mation of the ligand, and to a short contact of the
C11 b-hydrogen. In addition, the rigidity of the hete-
further stirring for 16 h at r.t., the reaction mixture was
cooled to −78°C and an additional amount of trifl-
uoroacetic acid (1.38 g, 12.1 mmol) in CH2Cl2 (34 mL)
was added under stirring. After 4 h at −78°C, the
reaction was quenched by addition of NaHCO3 satu-
rated solution (50 mL). The organic layer was washed
with brine, dried over Na2SO4 and the solvent removed
under reduced pressure. Purification of the crude residue
(6.8 g) by silica-gel chromatography (CHCl3/i-PrOH
99:1 v/v) gave an unresolved mixture (4.6 g, 62%) of the
two (1S,3S,2%R) and (1R,3S,2%R)-1-[2-benzyloxycar-
bonyl-2%-(benzyloxycarbonyl)amino]ethyl-3-methoxycar-
bonyl-1,2,3,4-tetrahydro-b-carbolines 6a and 6c (TLC:
Rf=0.23 and 0.25, respectively, CHCl3/i-PrOH 99:1 as
eluant). This material was employed for the lactamiza-
tion step without further purification.
rocyclic rings in the P1%, P% region of the inhibitor,
2
hinder the conformational adjustments necessary to
avoid the short contact and to optimise the bonding
interactions with the protein. These indications may
be useful for the design of new non-peptidic MMPs
inhibitors based on bulky groups in the P% region.
1
6. Experimental protocols
A solution of the unresolved mixture 6ac (4 g, 7.12
mmol) in toluene (60 mL) was treated with trifl-
uoroacetic acid (8.1 g, 71.2 mmol). The reaction mix-
ture, after refluxing for 4 h under nitrogen, was diluted
with EtOAc and washed with NaHCO3 saturated solu-
tion and brine. After drying over Na2SO4, the solvents
were removed under reduced pressure and the two hex-
ahydroindolizino[8,7-b]indole stereoisomers 7a and 7c
separated by silica-gel chromatography (CHCl3).
6.1. Chemistry
Melting points (Bu¨chi oil bath apparatus) are uncor-
rected. [a]D were determined with a Schmidt–Haensch
1604 digital polarimeter. Molecular sieves type 4A (1.7–
2.4 mm) were from Fluka Chemie AG (Buchs, Switzer-
land). Silica gel 60 (230–400 mesh) for column
chromatography and TLC silica gel 60 F254 plates were
from Merck AG (Darmstadt, Germany). IR spectra
were obtained with a Perkin–Elmer 16 FPC FT-IR
(2R,5S,11bR) - 2 - (Benzyloxycarbonyl)amino - 5 - meth-
oxycarbonyl-3-oxo-2,3,5,6,11,11b-hexahydro-1H-indo-
lizino[8,7-b]indole 7a (1.977 g, 64%) was obtained
(EtOAc) as white crystals: mp 220–222°C; TLC: Rf=
0.13 (CHCl3/i-PrOH 99:1); [a]2D2=34° (c=1, CHCl3);
IR (KBr) main peaks at 3374, 1693, 1518, 1425, 1277,
1
spectrophotometer. H NMR spectra were recorded on
a Varian XL-300 spectrometer operating at 300 MHz.
Chemical shifts are given in ppm relative to TMS as the
internal standard. Analytical HPLC was performed on a
Waters apparatus using a Vydac C18 (4.6×250 mm, 30
mm) column with CH3CN/H2O 25:75 (0.1% TFA) sys-
tem as eluant (flow rate 0.5 mL min−1) with UV detec-
tion (214 nm). Reagent grade materials were from Fluka
Gmbh or Aldrich Chemical Co. and were used without
further purification. Elemental microanalyses (C, H, N)
were within 0.4% of the calculated values. Continuous
fluorimetric assays were performed using a Perkin–
Elmer spectrofluorimeter LS 50B.
1
1017, 745 cm−1; H-NMR (DMSO d6) l1.76 and 2.92
(2H, two m, H-1), 2.94 (1H, dd, J=6.7 and 15.5 Hz,
H-6) 3.36 (1H, d, J=15.5 Hz, H-6), 3.61 (3H, s,
COOCH3), 4.56 (1H, m, H-2), 4.98 (1H, m, H-11b), 5.04
(2H, s, Ph–CH2–O), 5.24 (1H, d, J=6.7 Hz, H-5),
6.93–7.50 (9H, m, aromatic), 7.65 (1H, d, J=8.8 Hz,
NHꢀCO), 11.14 (1H, s, H-11). Anal. C24H23N3O5 (C, H,
N).
(2R,5S,11bS) - 2 - (Benzyloxycarbonyl)amino - 5 - meth-
oxycarbonyl-3-oxo-2,3,5,6,11,11b-hexahydro-1H-indo-
lizino[8,7-b]indole 7c (494 mg, 16%) was obtained
(MeOH) as a white solid: mp 174–176°C; TLC: Rf=
0.11 (CHCl3/i-PrOH 99:1); [a]2D2= −92° (c=0.5, DMF);
IR (CHCl3) main peaks at 1711, 1508, 1450, 1409, 1274,
6.1.1. General procedure for preparation of the
indolizino[8,7-b]indole derivatives 7a and 7c
A solution of trifluoroacetic acid (27 mg, 0.24 mmol)
in anhydrous CH2Cl2 (27 mL) was added dropwise,
under stirring, to a solution of benzyl-2(S)-benzyloxy-
carbonylamino-4-oxo-butyrate [28] (4.70 g, 13.77 mmol)
1
1071, 984 cm−1; H-NMR (DMSO-d6) l2.30 and 2.69
(2H, two m, H-1), 2.89 and 3.02 (2H, A and B of an
ABX, J=5.0, 10.9 and 15.5 Hz, H-6), 3.71 (3H, s,
COOCH3), 3.94 (1H, m, H-2), 4.37 (1H, dd, J=5.0 and
10.9 Hz, H-5), 5.02 (2H, s, Ph–CH2–O), 5.09 (1H, d,
and
L-tryptophan methylester (3.01 g, 13.77 mmol) in
anhydrous CH2Cl2 (160 mL) over 4 A molecular sieves
(2.8 g), cooled to 0°C, under nitrogen atmosphere. After

50
S. D’Alessio et al. / European Journal of Medicinal Chemistry 36 (2001) 43–53
J=7.8 Hz, H-11b), 6.93–7.50 (9H, m, aromatic), 7.84
(1H, d, J=8.9 Hz, NHꢀCO), 11.14 (1H, s, H-11). Anal.
C24H23N3O5 (C, H, N).
(1H, d, J=8.3 Hz, NHꢀCO), 11.15 (1H, s, H-11). Anal.
C21H19N3O5 (C, H, N).
6.1.3. Hydrolysis of the 5-methoxycarbonylhexa-
hydroindolizino[8,7-b]indole derivatives 8a and 8b
6.1.2. General procedure for preparation of the
indolizino[8,7-b]indole derivatives 8a and 8b
A suspension of the required ester (1 mmol) in 3:1
dioxane/methanol (9 mL) was treated with aqueous 2 N
NaOH (1 mL) and stirred overnight at r.t. The reaction
mixture was diluted with water (35 mL) and washed
with EtOAc. The aqueous layer was cooled in ice,
acidified with 2 N HCl to pH 3 and extracted with
CHCl3 (2×20 mL). After washing with brine and drying
over Na2SO4, the organic phase was evaporated under
reduced pressure to give the pure product 4a or 4b.
(2R,5S,11bR)-2-(Furan-2-carbonyl)amino-5-carboxy-
3-oxo-2,3,5,6,11,11b-hexahydro-1H-indolizino [8,7-b]-
indole 4a (239 mg, 70%), mp 195–198°C; HPLC: tR=
19.60 min; [a]2D2=126° (c=0.5, DMF); IR (KBr) main
To a suspension of the required 2-benzyloxycarbony-
lamino derivative 7a or 7b [19] (1 mmol) and 10% Pd/C
(100 mg) in MeOH (10 mL), ammonium formate (4
mmol) was added at r.t., under rapid stirring. After 10
min, the reaction mixture was filtered through a short
celite pad and the solvent removed under reduced pres-
sure. Washing with water of the resulting solid residue
gave the deprotected 2-amino derivative that was dried
in high vacuum and dissolved in anhydrous DMF (4.6
mL) together with furan-2-carboxylic acid (1 mmol) and
1-hydroxybenzotriazole (1 mmol). The mixture was
treated with a solution of dicyclohexylcarbodiimide (1.1
mmol) in DMF (2.3 mL) for 1 h at 0°C and 1 h at 25°C
and the solvent removed under reduced pressure. The
crude residue was stirred in EtOAc (10 mL), the dicyclo-
hexylurea filtered off and the solution washed with 1 M
citric acid, 1 M NaHCO3 and brine. The organic layer
was dried over Na2SO4 and the solvent removed under
reduced pressure. Silica gel chromatography (CHCl3/i-
PrOH 99:1) of the crude residue and crystallization from
MeOH gave the pure products.
peaks at 3394, 1692, 1530, 1428, 1296, 1205, 758, cm−1
;
1H-NMR (DMSO-d6) l1.95 and 2.96 (2H, two m, H-1),
2.98 (1H, dd, J=6.7 and 15.5 Hz, H-6) 3.35 (1H, d,
J=15.5 Hz, H-6), 3.64 (3H, s, COOCH3), 5.01 (1H, m,
H-11b), 5.03 (1H, m, H-2), 5.27 (1H, d, J=6.7 Hz,
H-5), 6.58–7.94 (7H, m, aromatic and furan), 8.71 (1H,
d, J=8.7 Hz, NHꢀCO), 11.14 (1H, s, H-11). Anal.
C20H17N3O5 (C, H, N).
(2R,5S,11bS)-2-(Furan-2-carbonyl)amino-5-carboxy-
3-oxo-2,3,5,6,11,11b-hexahydro-1H-indolizino [8,7-b]-
indole 4b (360 mg, 95%; mp 270–275°C; HPLC: tR=
19.60 min; [a]2D2=48° (c=0.5, DMF); IR (KBr) main
(2R,5S,11bR) - 2 - (Furan - 2 - carbonyl)amino - 5 - meth-
oxycarbonyl - 3 - oxo - 2,3,5,6,11,11b - hexahydro - 1H - in-
dolizino [8,7-b]indole 8a (275 mg, 70%), mp 150–153°C;
TLC: Rf=0.18 (CHCl3/i-PrOH 99:1); [a]2D2=54° (c=1,
CHCl3); IR (CHCl3) main peaks at 3464, 1704, 1595,
peaks at 3271, 1647, 1593, 1527, 1429, 1318, 1203, cm−1
,
1H-NMR (DMSO-d6) l2.48 (2H, two m, H-1), 3.04
(1H, dd, J=7.2 and 15.8 Hz, H-6), 3.22 (1H, d, J=15.8
Hz, H-6), 3.66 (3H, s, COOCH3), 4.52 (1H, m, H-2),
5.27 (2H, m, 5-H and H-11b) 6.60–7.96 (7H, m, aro-
matic and furan), 9.04 (1H, d, J=8.3 Hz, NHꢀCO),
11.15 (1H, s, H-11). Anal. C20H17N3O5 (C, H, N).
1508 1420, 1300, 1010, cm−1 1H-NMR (DMSO-d6)
;
l1.95 and 2.96 (2H, two m, H-1), 2.98 (1H, dd, J=6.7
and 15.5 Hz, H-6) 3.35 (1H, d, J=15.5 Hz, H-6), 3.64
(3H, s, COOCH3), 5.01 (1H, m, H-11b), 5.03 (1H, m,
H-2), 5.27 (1H, d, J=6.7 Hz, H-5), 6.58–7.94 (7H, m,
aromatic and furan), 8.71 (1H, d, J=8.7 Hz, NHꢀCO),
11.14 (1H, s, H-11). Anal. C21H19N3O5 (C, H, N).
(2R,5S,11bS) - 2 - (Furan - 2 - carbonyl)amino - 5 - meth-
oxycarbonyl - 3 - oxo - 2,3,5,6,11,11b - hexahydro - 1H - in-
dolizino [8,7-b]indole 8b (197 mg, 50%); mp 290–294°C;
TLC: Rf=0.16 (CHCl3/i-PrOH 99:1); [a]2D2=20° (c=
0.5, DMF); IR (CHCl3) main peaks at 1702, 1594, 1516,
6.2. Structural studies
6.2.1. X-ray data collection and reduction
Crystals of 4a were obtained from a methanol solu-
tion by slow evaporation. X-ray data were collected at
room temperature on a Rigaku AFC5R diffractometer
with graphite monochromated Cu Ka radiation and a
12 kW rotating anode generator. Cell constants and the
orientation matrix for data collection were obtained
from a least square fit of the angular settings of 20
carefully centred reflections in the range 26.5°<2q<
42.8°. The cell parameters, refined on higher angle
1421, 1304, 1173, 1012 cm−1 1H-NMR (DMSO-d6)
;
l2.48 (2H, two m, H-1), 3.04 (1H, dd, J=7.2 and 15.8
Hz, H-6), 3.22 (1H, d, J=15.8 Hz, H-6), 3.66 (3H, s,
COOCH3), 4.52 (1H, m, H-2), 5.27 (2H, m, H-5 and
H-11b) 6.60–7.96 (7H, m, aromatic and furan), 9.04

S. D’Alessio et al. / European Journal of Medicinal Chemistry 36 (2001) 43–53
51
reflections, are reported in table II. Intensity data were
collected by the ꢀ/2q scan technique with scans of
(1+0.3 tgq)° at variable and appropriate speed to a
maximum 2q of 124°. Stationary background counts
were recorded on each side of the reflection. The peak
counting was twice that of the background. Reflections
with I<25|(I) were rescanned with accumulation of
counts to improve counting statistics. Out of 1681 reflec-
tions, 1099 had I>3|(I) and were used in the refine-
ment. The intensities of three standard reflections,
measured after every 147 reflections declined by 7.7%; a
linear correlation factor was applied to the data to
account for this. An empirical absorption correction,
based on azimuthal scans of several reflections, was
applied resulting in transmission factors ranging from
0.92 to 1.00. The data were corrected for Lorentz and
polarization effects. A correction for secondary extinc-
tion was finally applied (coefficient=2.7×10−6).
Sw(ꢀFoꢀ−ꢀFcꢀ)2 where w=4 Fo2/|2(F2o). All the H atoms
were located at the expected positions except the car-
boxyl group, which was detected from the final differ-
ence Fourier map. The H-atoms were included in the
last structure factor calculation with isotropic thermal
parameters deduced from the carrier atoms. The final R
and Rw are 0.043 and 0.046, respectively. The atomic
scattering factors were those of Cromer and Mann [30].
Anomalous dispersion effects were taken into account
adopting Df % and Df ¦ values of Cromer [31]. The final
fractional coordinates of the non-H atoms together with
their Beq and individual e.s.d.s are given in table III.
All calculations were performed using TEXSAN [32]
crystallographic software package.
A list of H atom coordinates, non-H atom anisotropic
parameters, valence bond lengths and angles, torsion
angles and observed and calculated structure factors are
available on request to the authors.
6.2.3. Matrix metalloproteinases activation and
inhibition assays
6.2.2. Structure solution and refinement
The structure was solved by direct methods using the
program SIR92 [29] and successive Fourier maps. The
non-H atoms were refined anisotropically by full-matrix
least-squares method. The function minimized was
The proenzymes [33] were activated immediately prior
to use with p-aminophenylmercuric acetate (APMA 1
mM for 1 h at 25° for pro-MMP-2 and pro-MMP-9
[34]; with APMA 2 mM for 2 h at 37°C for pro-MMP-8
[35] and with a-chymotrypsin 6 mg mL−1 for 2 h at 37°C
followed by phenylmethylsulfonyl fluoride 0.2 mM for
pro-MMP-3 [36].
Table II. Crystal data of 4aa
For assays measurements, 0.05 mM solutions of the
inhibitors in MeOH were further diluted as required in
the assay buffer, Tris 100 mM, NaCl 100 mM, CaCl2 10
mM, pH 7.5, Brij 35 0.05%. The activated enzyme plus
inhibitor (13 mM final concentration) solution was incu-
bated in the assay buffer for 3 h. The assay temperature
was 25°C for MMP-2 and MMP-9 and 37°C for MMP-
3 and MMP-8. After the addition of 0.1 or 0.05 mM
solution of the appropriate fluorogenic substrate [37] in
DMSO, the hydrolysis was monitored [38] by continu-
ously recording the increase in fluorescence (uex 328 nm,
uem 393 nm) using a Perkin–Elmer spectrofluorimeter
LS 50B. Percent inhibitions were calculated from con-
trol reactions without the inhibitor as means of at least
three independent measurements.
Empirical formula
Formula weight
Crystal system
C20H17N3O5
379.4
Orthorhombic
9.745(2)
27.131(5)
6.735(2)
1780.7(8)
P21 21 21
1.415
4
792
1.5418
0.8
0.2×0.3×0.01
124
1099
0.043, 0.046
4 F2o/|2(F2o)
1.4
,
a (A)
,
b (A)
,
c (A)
3
,
V (A )
Space group
dc (g cm−1
Z
)
F(000)
,
u (Cu Ka) (A)
v (Cu Ka) (mm−1
Crystal size (mm)
2qmax (°)
)
No. reflections [I\3|(I)]
R, Rw
Weighting scheme
S
6.2.4. Adamalysin II inhibition assay
Observation/parameter
4.3
Similarly to the previous procedure, adamalysin II
[33] (6 mg) in the assay buffer (Tris–HCl 50 mM, CaCl2
10 mM, pH 7.5) plus 0.05 mM inhibitor solution in
MeOH, was immediately added with a solution of 2-
aminobenzoyl–Ala–Gly–Leu–Ala-p-nitrobenzylamide
Max, min peak in the final difference map 0.17, −0.19
−3
,
(e A
)
a Maximum resolution of collected data is given by the 2qmax
(°) value.

52
S. D’Alessio et al. / European Journal of Medicinal Chemistry 36 (2001) 43–53
[3] Hagman W.K., Lark M.V., Becker J.W., Ann. Rep. Med.
Chem. 31 (1996) 231–240.
Table III. Fractional coordinates and B(eq) with their e.s.d.s, in
parentheses, for 4a
[4] Ye Q.Z., Hupe D., Johnson L., Curr. Med. Chem. 3 (1996)
407–418.
[5] Levy D.E., Ezrin A., Emerging Drugs 2 (1997) 205–230.
Atom
x
y
z
B(eq)
C(1)
C(2)
C(3)
C(5)
C(6)
C(7)
C(8)
C(9)
C(10)
C(11)
C(12)
C(13)
C(14)
C(16)
C(17)
C(20)
C(21)
C(22)
C(23)
C(24)
N(4)
0.7036 (6)
0.7201 (5)
0.5759 (6)
0.3465 (5)
0.3252 (6)
0.2684 (6)
0.2596 (8)
0.3393 (8)
0.4329 (7)
0.5472 (5)
0.3894 (5)
0.3602 (6)
0.4407 (6)
0.4860 (5)
0.2528 (6)
0.8019 (6)
0.9211 (7)
1.0479 (6)
1.1188 (8)
1.030 (1)
0.9222 (2)
0.9593 (2)
0.9630 (2)
0.9301 (2)
0.8740 (2)
0.7657 (2)
0.7280 (2)
0.7275 (2)
0.7644 (2)
0.9222 (2)
0.8506 (2)
0.8047 (2)
0.8025 (2)
0.8729 (2)
0.9571 (2)
0.9065 (2)
0.8910 (2)
0.9080 (2)
0.8746 (4)
0.8402 (3)
0.9408 (1)
0.8456 (2)
0.9458 (1)
0.9846 (1)
0.9836 (1)
0.9496 (1)
0.8851 (1)
0.8489 (2)
0.9464 (8)
0.7791 (8)
0.686 (1)
0.7507 (8)
0.7462 (8)
0.983 (1)
1.118 (1)
1.287 (1)
1.326 (1)
0.9922 (9)
0.9247 (8)
1.021 (1)
1.192 (1)
1.0345 (8)
0.895 (1)
0.5159 (9)
0.399 (1)
0.384 (1)
0.260 (1)
0.204 (1)
0.8057 (7)
1.1983 (7)
0.6352 (7)
0.5305 (6)
1.0232 (8)
0.8575 (6)
0.5081 (7)
0.2895 (8)
3.4 (3)
2.8 (3)
2.8 (3)
2.9 (3)
3.4 (3)
4.5 (3)
6.2 (4)
6.7 (5)
5.3 (4)
2.8 (2)
2.9 (3)
3.1 (3)
3.4 (3)
2.7 (2)
3.4 (3)
3.1 (3)
3.9 (3)
4.4 (3)
7.2 (5)
8.1 (6)
2.8 (2)
3.4 (2)
2.9 (2)
3.5 (2)
5.8 (2)
3.8 (2)
3.9 (2)
6.5 (3)
[6] Schwartz M.A., Van Wart H.E., in: Ellis G.P., Luscombe D.K.
(Eds.), Progress in Medicinal Chemistry, vol. 29, Elsevier,
Amsterdam, 1991, pp. 239–270.
[7] Porter J.R., Millican T.A., Morphy J.R., Exp. Op. Ther. Pat. 5
(1995) 1287–1296.
[8] Becket R.P., Davidson A.H., Drummond D.H., Huxley P.,
Whittaker M., Drug Discov. Today 1 (1996) 16–26.
[9] Becket R.P., Exp. Op. Ther. Pat. 6 (1996) 1305–1315.
[10] Becket R.P., Whittaker M., Exp. Op. Ther. Pat. 8 (1998)
259–282.
[11] Sto¨cker W., Grams S., Baumann U., Reinemez P., Gomis-Ru¨th
F.X., McKay D.B., Bode W., Prot. Sci. 4 (1995) 823–840.
[12] Gomis-Ru¨th F.X., Kress L.F., Bode W., EMBO J. 12 (1993)
4151–4157.
[13] Gomis-Ru¨th F.X., Kress L.F., Kellerman J., Moyr I., Lee X.,
Huber R., Bode W., J. Mol. Biol. 239 (1994) 513–544.
[14] Fox J.W., Bjarnason J.B., in: Bailey G.S. (Ed.), Enzymes from
Snake Venom, Alaken, Fort Collins, CO, 1998, pp. 599–632.
[15] Robeva A., Politi V., Shannon J.D., Bjarnason J.B., Fox J.W.,
Biomed. Biochim. Acta 50 (1991) 769–773.
0.4886 (4)
N(15) 0.5189 (5)
N(19) 0.8224 (4)
O(3)
O(17)
O(18) 0.1204 (4)
O(20)
O(25)
[16] Zhang D., Botos I., Gomis-Ru¨th F.X., Doll R., Blood C.,
Njoroge F.G., Fox J.W., Bode W., Meyer F., Proc. Natl. Acad.
Sci. USA 91 (1994) 8447–8451.
0.5507 (4)
0.2927 (4)
[17] Cirilli M., Gallina C., Gavuzzo E., Giordano C., Gomis–Ru¨th
F.X., Gorini B., Kress L.F., Mazza F., Paglialunga-Paradisi
M., Pochetti G., Politi V., FEBS Letts. 418 (1997) 319–322.
0.6909 (4)
0.9029 (5)
[18] Gomis-Ru¨th F.X., Meyer E.F., Kress L.F., Bode W., Politi V.,
Prot. Sci. 7 (1998) 283–292.
[19] De la Figuera N., Alkorta I., Garcia Lopez M.T., Herranz R.,
Gonza´les Mun˜iz R., Tetrahedron 51 (1995) 7841–7856.
[39] in MeOH (1.5 mM, final concentration. Hydrolysis
of the fluorogenic substrate was monitored by continu-
ously recording the increase in fluorescence (uex 320 nm,
uem 420 nm) for 30 min at 30°C.
[20] Bailey P.D., Hollinshead S.P., McLay N.R., Morgan K.,
Palmer S.J., Prince S.N., Reynolds C.D., Wood S.D., J. Chem.
Soc. Perkin Trans. 1 (1993) 431–439.
[21] Melnyk P., Ducrot P., Thal C., Tetrahedron 49 (1993) 8589–
8596.
[22] Ko¨nig W., Geiger R., Chem. Ber. 103 (1970) 788–798.
Acknowledgements
[23] Fairlie D.P., Tyndall J.D.A., Reid R.C., Wong A.K.,
Abbenante G., Scanlon M.J., March D.R., Bergman D.A.,
Chai C.L.L., Burkett B.A., J. Med. Chem. 43 (2000) 1271–
1281.
The authors wish to thank M. Petrilli and F. Dianetti,
Servizio Microanalisi del CNR, Area della Ricerca di
Roma, Montelibretti, who performed elemental micro-
analyses. This work was supported by Italian CNR
Targeted Program Biotecnologie, MURST and Cofin.
MURST 97 CFSIB.
[24] As observed by one of the referees, the analogue of 4a with
C11b in the R configuration, would render the C1 atom ‘above’
the plane approximately formed by C11b, N4, C3 and C2 in
ring D (figure 3), allowing more rotational flexibility to bond
C2ꢀN19 and therefore more freedom for the furan-2-carboxam-
ide group. Synthesis of this compound was included in the
original plan of the work and the corresponding methylester
was also prepared. Unfortunately, hydrolysis under the usual,
mild, alkaline conditions caused extensive isomerization (possi-
bly epimerization at C5). From the resulting mixture of car-
boxylic acids, we failed to isolate the pure substances in
adequate amount for characterization.
References
[1] Woessner J.F., FASEB J. 5 (1991) 2145–2154.
[2] Morphy J.R., Millican T.A., Porter J.R., Curr. Med. Chem. 2
(1995) 743–762.
[25] Babine E., Bender S.L., Chem. Rev. 97 (1997) 1359–1472.

S. D’Alessio et al. / European Journal of Medicinal Chemistry 36 (2001) 43–53
53
[26] Yiotakis A., Lecoq A., Vassiliou S., Raynal I., Cuniasse P., J.
Med. Chem. 37 (1994) 2713–2720.
Cambridge, UK). Adamalysin II was a kind gift from L.F.
Kress (Roswell Park Cancer Institute, Buffalo, New York).
[34] Murphy G., Crabbe T., Methods Enzymol. 248 (1995) 470–
484.
[27] Weber P.C., Pantoliano M.W., Simons D.M., Salemme F.R., J.
Am. Chem. Soc. 116 (1994) 2717–2724.
[35] Tschesche H., Methods Enzymol. 248 (1995) 431–439.
[36] Nagase H., Methods Enzymol. 248 (1995) 449–470.
[37] Fluorogenic substrates (7-methoxycoumarin-4-yl)acetyl-Pro-
[28] Prepared by oxydation of N-benzyloxycarbonyl-D-homoserine
according to Keith D.D., Tortora J.A., Ineichen K., Leimgru-
ber W., Tetrahedron 31 (1975) 2633–2636.
[29] Altomare A., Cascarano G., Giacovazzo C., Guagliardi A.,
Burla M.C., Polidori G., Camalli M., J. Appl. Cryst. 27 (1994)
435.
Leu-Gly-Leu-(3-[2,4-dinitro-phenyl]-L-2,3-diaminopropionyl)
×-Ala-Arg-NH2 (MCA-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH2)
Ref. 33 for MMP-2 and MMP-9, MCA-Pro-Cha-Gly-Nva-His-
Ala-Dpa-NH2 for MMP-3 (Knight, C.G., private communica-
tion) and MCA-Pro-Leu-Ala-Nva-Dpa-Ala-Arg-NH2 for
MMP-3 (Knight C.G., private communication) were supplied
by Knight C.G. (Strangeways Research Laboratories, Cam-
bridge, UK). 2-Aminibenzoyl-Ala-Gly-Leu-Ala-p-nitrobenzy-
lamide was purchased from Bachem (Budendorf, Switzerland).
[30] Cromer D.T., Waber J.T., International Tables for X-ray Crys-
tallography, vol. IV, Kynock Press, Birmingham, UK, 1974
Table 2.2A.
[31] Cromer D.T., Waber J.T., International Tables for X-ray Crys-
tallography, vol. IV, Kynock Press, Birmingham, UK, 1974
Table 2.3.1.
[32]
TEXSAN
:
TEXRAY Structure Analysis Package, Molecular Struc-
[38] Knight C.G., Willenbrock F., Murphy G., FEBS Letts. 296
(1992) 263–266.
ture Corporation, 1985.
[33] Pro-MMP-2, pro-MMP-3, pro-MMP-8 and pro-MMP-9 were
supplied by G. Murphy (Strangeways Research Laboratories,
[39] Fox J.W., Campbell R., Beggerly L., Bjarnason J.B., Eur. J.
Biochem. 156 (1986) 65–72.