Crystallographic studies on the reaction of isopenicillin N synthase with an unsaturated substrate analogue

Article abstract of DOI:10.1039/b212270g

Isopenicillin N synthase (IPNS) catalyses conversion of the linear tripeptie δ-(L-α-aminoadipoyl)-L-cysteinyl-D-valine (ACV) to isopenicillin N (IPN), the central step in biosynthesis of the β-lactam antibiotics. The unsaturated substrate analogue δ-(L-α-aminoadipoyl)-L-cysteinyl-D-vinylglycine (ACvG) has previously been incubated with IPNS and a single product was isolated, a 2-α-hydroxymethyl isopenicillin N (HMPen), formed via a monooxygenase mode of reactivity. ACvG has now been crystallised with IPNS and the structure of the anaerobic IPNS:Fe(II):ACvG complex determined to 1.15 A resolution. Furthermore, by exposing the anaerobically grown crystals to high-pressure oxygen gas, a structure corresponding to the bicyclic product HMPen has been obtained at 1.60 A resolution. In light of these and other IPNS structures, and recent developments with related dioxygenases, the [2 + 2] cycloaddition mechanism for HMPen formation from ACvG has been revised, and a stepwise radical mechanism is proposed. This revised mechanism remains consistent with the observed stereospecificity of the transformation, but fits better with apparent constraints on the coordination geometry around the active site iron atom.

Full text of DOI:10.1039/b212270g

View Article Online / Journal Homepage / Table of Contents for this issue
Crystallographic studies on the reaction of isopenicillin N synthase
with an unsaturated substrate analogue
Jonathan M. Elkins, Peter J. Rutledge,† Nicolai I. Burzlaff,‡ Ian J. Clifton,
Robert M. Adlington, Peter L. Roach§ and Jack E. Baldwin*
The Dyson Perrins Laboratory and the Oxford Centre for Molecular Sciences,
University of Oxford, South Parks Road, Oxford, UK OX1 3QY.
E-mail: jack.baldwin@chem.ox.ac.uk
Received 16th December 2002, Accepted 5th March 2003
First published as an Advance Article on the web 21st March 2003
Isopenicillin N synthase (IPNS) catalyses conversion of the linear tripeptide δ-(-α-aminoadipoyl)--cysteinyl--
valine (ACV) to isopenicillin N (IPN), the central step in biosynthesis of the β-lactam antibiotics. The unsaturated
substrate analogue δ-(-α-aminoadipoyl)--cysteinyl--vinylglycine (ACvG) has previously been incubated with
IPNS and a single product was isolated, a 2-α-hydroxymethyl isopenicillin N (HMPen), formed via a monooxygenase
mode of reactivity. ACvG has now been crystallised with IPNS and the structure of the anaerobic IPNS:Fe():ACvG
complex determined to 1.15 Å resolution. Furthermore, by exposing the anaerobically grown crystals to high-
pressure oxygen gas, a structure corresponding to the bicyclic product HMPen has been obtained at 1.60 Å
resolution. In light of these and other IPNS structures, and recent developments with related dioxygenases, the
[2 ϩ 2] cycloaddition mechanism for HMPen formation from ACvG has been revised, and a stepwise radical
mechanism is proposed. This revised mechanism remains consistent with the observed stereospecificity of the
transformation, but fits better with apparent constraints on the coordination geometry around the active site iron
atom.
Introduction
Isopenicillin N synthase (IPNS) catalyses conversion of the
tripeptide δ-(-α-aminoadipoyl)--cysteinyl--valine (ACV, 1)
to isopenicillin N (IPN, 2), the central step in the biosynthesis
of penicillin and cephalosporin antibiotics. IPNS is a non-haem
iron()-dependent enzyme and utilises one equivalent of
molecular oxygen in the bicyclisation reaction.1 The bicyclis-
ation is believed to proceed via initial β-lactam closure to give a
monocyclic intermediate, associated with
a high valent
iron()–oxo species (Fig. 1a), which then mediates closure of
the five-membered thiazolidine.2,3
Solution-phase experiments with a wide range of substrate
analogues have shown IPNS to exhibit broad substrate speci-
ficity, accepting a large number of modified tripeptides as sub-
strates.4 It is in the position of the valinyl residue that the
enzyme has proved least discriminating. When IPNS was
incubated with the unsaturated substrate analogue δ-(-α-
aminoadipoyl)--cysteinyl--vinylglycine (ACvG, 3), a 2-α-
hydroxymethyl penam (HMPen, 4) was the only product
isolated (Fig. 1b).5 Labelling studies with several related
unsaturated substrates have shown the incorporation of 18O2
into other hydroxylated products,6 and it is probable that this is
also the case with HMPen 4. Formation of the hydroxylated
product 4 from ACvG was rationalised in terms of a mono-
oxygenase (Ϫ2H, ϩO) pathway, with a mechanism involving a
[2πs ϩ 2π] cycloaddition of the iron–oxene to the re face of the
alkene (Fig. 1b).6,7 Incubation of stereospecifically deuterium
labelled isotopomers produced hydroxymethyl-penams in
which the deuterium labels were not scrambled, supporting a
stereospecific mechanism and consistent with the proposed
cycloaddition.8
Fig. 1 The reaction of IPNS with a, its natural substrate ACV 1;
and b, the unsaturated analogue ACvG 3, including an outline
representation of the mechanism previously proposed for this reaction;
L-AA = L-α-aminoadipoyl.
Recently, the application of pseudo-time-resolved crystallo-
graphy to IPNS has allowed structural investigation of incu-
bation results previously observed in solution. By subjecting
IPNS:Fe():substrate crystals to high pressures of oxygen gas,
crystals containing IPNS:Fe():product complexes can be
obtained.3,9,10 We have recently reported solution of a structure
for the IPNS:Fe():IPN complex, obtained from high-pressure
† Current address: Department of Chemistry, University College
Dublin, Belfield, Dublin 4, Ireland.
‡ Current address: Fachbereich Chemie, Universität Konstanz, 78457
Konstanz, Germany.
§ Current address: Department of Chemistry, University of
Southampton, Highfield, Southampton, UK SO17 1BJ.
T h i s j o u r n a l i s © T h e R o y a l S o c i e t y o f C h e m i s t r y 2 0 0 3
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 4 5 5 – 1 4 6 0
1455

View Article Online
Fig. 2 Stereofigures showing IPNS in complex with ACvG and HMPen. a, The anaerobic IPNS:Fe():ACvG complex with a 2mFo Ϫ DFc electron
density map contoured at 1σ around the protein atoms (in blue) and a mFo Ϫ DFc omit map contoured at 3.2σ around ACvG (in red); b, the
IPNS:Fe():HMPen complex with a 2mFo Ϫ DFc electron density map contoured at 1σ around the protein atoms (in blue) and a mFo Ϫ DFc omit
map contoured at 2.5σ around HMPen (in red); c, an overlay of the IPNS:Fe():ACvG and IPNS:Fe():HMPen structures to show the changes that
have occurred in the course of reaction.
oxygenation of IPNS crystals incorporating iron and natural
substrate ACV.3 Attempts to obtain an enzyme:product com-
plex in this manner were complicated by the release of product
from the active site (N. I. Burzlaff, unpublished results). It was
hypothesised that the hydroxymethyl penicillin 4 would be more
firmly tethered to iron in the IPNS active site than the natural
product IPN, by virtue of the additional hydroxyl ligand to the
metal, and thus could allow improved structural characteris-
ation of an IPNS:Fe():product complex.
Here we report the X-ray crystal structure of the IPNS:
Fe():ACvG complex, and a structure for the IPNS:Fe():
HMPen complex, obtained by high-pressure oxygenation
of IPNS:Fe():ACvG crystals. These results, in combination
with previous X-ray crystallographic studies,3,11 allow proposal
of a revised mechanism for the IPNS reaction cycle in its
monooxygenase mode of reactivity.
Results and discussion
Structure of the IPNS:Fe(II):ACvG complex
The structure of the IPNS:Fe():ACvG complex was deter-
mined to 1.15 Å resolution (Fig. 2a). The overall structure of
the protein is essentially identical to the IPNS:Fe():ACV struc-
ture.11 The tripeptide substrate analogue also adopts a very
similar conformation to that of the natural substrate, with only
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 4 5 5 – 1 4 6 0
1456

View Article Online
Table 1 Data collection and statistics
Complex
IPNS:Fe():ACvG
IPNS:Fe():HMPen
X-ray source
Wavelength λ/Å
Space group
ID14-4 ESRF
0.9312
P212121
Station 7.2 SRS
1.488
P212121
Unit cell (a/Å, b/Å, c/Å)
46.401, 71.393, 100.906
46.534, 70.751, 100.876
Resolution shell/Å
Measurements
Unique reflections
Multiplicity
Average I/σ(I )
Completeness (%)
Rmerge (%)a
29.11–1.15 1.21–1.15
41.17–1.60 169–1.60
375623
115554
3.3
23648
13622
1.7
101990
42669
2.4
14387
6077
2.4
6.2
5.3
6.4
1.9
96.2
6.1
78.9
11.9
95.2
7.4
94.2
33.8
Rcryst (%)b
14.3
15.2
0.007 Å (1.27Њ)
15.1
17.8
0.013 Å (1.48Њ)
Rfree (%)c
RMS deviationd
B factorse/Å2
8.6/10.0/9.4/21.5
16.3/18.8/36.5/33.1
Residues
Water molecules
329
586
321
395
PDB ID
1ODM
1ODN
b
c
a Rmerge = ΣjΣh|Ih,j Ϫ 〈Ih〉|/ΣjΣh〈Ih〉 × 100. Rcryst = Σ||Fobs| Ϫ |Fcalc||/Σ|Fobs| × 100. Rfree = based on 4% of the total reflections. d RMS deviation from
ideality for bonds (and angles). e Average B factors in order: main chain, side chain, substrate and solvent.
a small difference in the region of the vinyl group. ACvG is
bound to the protein by electrostatic interactions involving its
carboxylate groups: a salt bridge is formed between the
aminoadipoyl carboxylate and Arg187, and the vinylglycine
carboxylate is in position to hydrogen bond with Tyr189,
Ser281 and to Arg279 via a water molecule. The cysteinyl thiol-
ate of ACvG is a ligand at the active site iron atom, as are the
protein side-chains of His214, Asp216 and His270. As in the
ACV complex, there is a water ligand in the site trans to His214.
However, in contrast to the complex with the natural substrate,
there is also a second water ligand at the metal, in the site
opposite Asp216, near the vinylglycine side-chain. In accom-
modating the extra water ligand at this site, the vinylic side-
chain adopts a conformation which places the terminal CH2
group in the pocket that is occupied by the pro-R methyl group
in the IPNS:Fe():ACV complex (the methyl group further
from iron). In this orientation, the vinyl group of ACvG is close
to the oxygen binding site opposite Asp216, well placed for
reaction with the proposed iron()–oxo species.2 However, the
additional water ligand must presumably be displaced from iron
before oxygen can bind to initiate the oxidative reaction.
similar migration of sulfur around iron was observed to have
occurred in the IPNS:Fe():IPN complex.3 These results are in
contrast to turnover structures in which the reaction cycle has
been intercepted before thiazolidine ring-closure. In the mono-
cyclic β-lactam formed from δ-(-α-aminoadipoyl)--cysteinyl-
-S-methylcysteine (ACmC), the cysteinyl sulfur atom occupies
essentially the same position as it does in the starting IPNS:
Fe():ACmC complex.3 The same is also true of the cysteine-
derived sulfur in the thiocarboxylic acid product arising from
δ-(-α-aminoadipoyl)--cysteine -α-hydroxyisovaleryl ester
(ACOV).9 Movement of the cysteinyl sulfur atom therefore
appears to be indicative of thiazolidine closure.
Connected to the movement of the cysteinyl sulfur, there is
no electron density in the position that the cysteinyl β-carbon
occupies in the anaerobic structure. This change, considered in
combination with the changes to the sulfur-derived electron
density discussed above, is consistent with movement of the
cysteinyl β-carbon to a position much closer to the vinylglycine-
derived nitrogen, in keeping with formation of a covalent bond
between these two atoms.
Secondly, there is strong additional electron density adjacent
to the region previously occupied by the vinylglycine side-chain.
This is consistent with the presence of an extra oxygen atom in
this region, and with appreciable movement of the vinylglycine
CH2 group having occurred. The model of the pendant
hydroxymethyl chain fits well to the observed density. The
stereochemistry of this group seems unequivocal, and consist-
ent with the 2-α-hydroxymethylpenam product 4 observed in
the solution studies.5
The average refined temperature factor (B factor) of the
HMPen molecule is high compared to that of the protein (Table
1), which suggests lower occupancy of the HMPen product.
However, in general the electron density is convincing and in
keeping with the presence of HMPen, the product formed from
the reaction in solution. Only for the β-lactam carbonyl and the
thiazolidine sulfur is the electron density weak. The position of
this carbonyl group suggests that it is similarly orientated to the
equivalent carbonyls in both the IPNS:Fe():IPN structure and
the structure of the monocyclic β-lactam generated from
ACmC.3 In all these cases, the carbonyl oxygen has moved
relative to the starting complexes, rotating ca. 60Њ around the
axis of the cysteinyl C1–Cα bond towards Phe211, into the same
plane as the atoms of the β-lactam ring. The poorer definition
Structure of the IPNS:Fe(II):HMPen complex
Crystals of IPNS:Fe():ACvG were exposed to high pressures
of oxygen gas to initiate turnover in the active site. A crystal
that had been exposed to 20 bar oxygen for 410 minutes allowed
collection of a 1.60 Å resolution dataset, with active site elec-
tron density consistent with an IPNS:Fe():HMPen complex
(Fig. 2b).
There are two key changes in the structure of the substrate-
derived electron density from the IPNS:Fe():ACvG complex,
and it is observation of these changes that form the basis for
assigning the active site electron density as the HMPen mole-
cule (the product observed when ACvG was incubated with
IPNS in solution). Firstly, the cysteinyl sulfur atom has moved
relative to the anaerobic structure: there is weaker electron
density at the position it occupied in the anaerobic structure
(which may correspond to residual ACvG or a low-occupancy
water molecule) and a significant increase in electron density
at a new position, further from the iron. It appears that the
cysteinyl sulfur has moved towards the vinylglycine β-carbon,
leaving its former binding site opposite His270 vacant. A
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 4 5 5 – 1 4 6 0
1457

View Article Online
Fig. 3 a, Mechanism for the conversion of ACV 1 to IPN 2 by IPNS; b, the revised mechanism for conversion of ACvG 3 to HMPen 4 by IPNS.
L-AA = L-α-aminoadipoyl.
of the carbonyl group in this case is consistent with the chang-
ing shape of the molecule overall; from an overlay of the ACvG
and HMPen structures (Fig. 2c) it can be seen that the β-lactam
carbonyl is the region in which greatest movement occurs in the
conversion of ACvG to HMPen. Furthermore, without the
tether provided by ligation of the cysteinyl thiolate to iron, a
greater degree of freedom is imparted to this region of the mol-
ecule. It is interesting to note that the β-lactam carbonyl is also
the least well defined region of the product in the electron dens-
ity map of the IPNS:Fe():IPN complex.3 Thus the original
hypothesis that the α-hydroxymethyl group could aid the cap-
ture of a bicyclic penam by stabilising the IPNS:Fe():product
complex through a bond to iron() is only partly validated,
since the additional link is not in the region that undergoes
greatest movement during substrate turnover.
seen in the IPNS:Mn() structure. The conformation of the
Phe211 side-chain in the unexposed enzyme:substrate com-
plexes is such that it partially obscures the entrance to the active
site. Thus rotation of this side-chain may be necessary in order
to permit departure of the product once formed. It seems plaus-
ible that just as the eight C-terminal residues might move as
part of the product release process, so too this rotation of
Phe211 occurs upon formation of the product, and is linked to
product departure from the active site.
There is a small amount of additional electron density visible
in the IPNS:Fe():HMPen complex, in the space that is occu-
pied by Phe211 in the unexposed complexes. This density
appears to originate as a branch from the aminoadipoyl-
cysteinyl peptide bond of the substrate, so it is possible that it
arises from a small amount of the product on its way out of
the active site. However it is equally possible that this density
corresponds to several poorly ordered water molecules.
Overall protein structure
The overall structure of the IPNS:Fe():HMPen complex
corresponds closely to the unexposed IPNS:Fe():ACvG com-
plex (RMS difference 0.202 Å). Interestingly however, there is
little electron density corresponding to the eight residues at the
protein C-terminus (Leu324–Thr331) in the product structure.
Thus, it was concluded that these residues are disordered and
this region was deleted from the model after the first round of
refinement. On the basis of the previously published IPNS:
Mn()12 and IPNS:Fe():ACV11 structures, it has been proposed
that the substrate and the C-terminus compete as ligands for
the iron binding site trans to His270. In the absence of sub-
strate, the C-terminus enters the active site and Gln330 binds to
the metal; upon substrate binding, Gln330 is displaced from
iron and the eight C-terminal amino acids adopt a position over
the active site entrance.11 It follows that the C-terminus may
become more mobile after the reaction is complete, to allow the
product to leave the active site. The observed lack of discrete
electron density for the C-terminus is consistent with mobility
in this region of the protein, which could in turn be directly
linked to substrate turnover having occurred in the crystal.
Connected to the mobility of the C-terminus, it has been
noted previously that the side-chain of Phe211 occupies differ-
ent conformations in the presence and absence of the natural
substrate, undergoing a rotation of ca.120Њ about the Cα–Cβ
bond when ACV binds.11 The conformation of this side-chain
in the IPNS:Fe():ACvG complex mirrors that seen in the
IPNS:Fe():ACV structure, while in the IPNS:Fe():HMPen
complex Phe211 has rotated 120Њ to return to the orientation
Mechanistic conclusions
In light of the results reported here and other structural
information now available for IPNS,3,11 as well as structural and
spectroscopic information from other related oxygenase
enzymes, it is possible to propose a revised mechanism for the
IPNS reaction with ACvG (Fig. 3b). Rather than the [2πs ϩ 2π]
cycloaddition proposed previously (Fig. 1b),6,7 the iron()–oxo
moiety of intermediate 21 could instead instigate a single-
electron reaction with the double bond. Attack on the less
substituted end would give the secondary radical 22, the more
substituted of the two possible radical intermediates. This
species closely resembles the radical intermediate 18 thought to
be involved in the conversion of ACV to IPN. Attack of the
radical in 22 on the cysteinyl sulfur, with concomitant homoly-
sis of the S–Fe bond, would lead directly to formation of the
HMPen complex 23.
While the cycloaddition alternative cannot be discounted
altogether, the radical mechanism is more in line with that
believed to operate during thiazolidine closure in the reaction
of IPNS with its natural substrate ACV (Fig. 3a). Furthermore,
it avoids the need either to displace the second water ligand
from iron, or to expand the metal’s coordination sphere beyond
six—one of these alternatives would be required for the cyclo-
addition mechanism, which leads to a metallocyclic intermedi-
ate incorporating three bonds from the tripeptide-derived
species to iron. The radical mechanism is also consistent with
the stereoselectivity seen in experiments with stereospecifically
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 4 5 5 – 1 4 6 0
1458

View Article Online
Fig. 4 Synthesis of ACvG 3 from -methionine 5, -α-aminoadipic acid 11 and -cysteine 13. L-AA = L-α-aminoadipoyl.
labelled ACvG isotopomers.8 Since the radical intermediate 22
is tethered to iron through its pendant oxygen, free rotation is
not possible and stereochemical integrity would be preserved.
The reaction of IPNS with ACvG in the crystalline state
provides further structural information on the complexes of
this enzyme with its bicyclic penam products. Moreover,
these crystallographic studies allow revision of the previously
proposed mechanism to better account for the structural
information now available.
10 mM NH4HCO3 in water–methanol as eluant (running time
0–5 minutes: 0% methanol; 5–12 minutes: gradient of 0–25%
methanol; 13–18 minutes: 2.5% methanol; 18–20 minutes: 0%
methanol v/v), 4 ml minϪ1, λ = 254 nm, 5 AUFS (absorbance
units full scale); tr = 11 min].
Crystallisation and turnover experiments
Crystals of the IPNS:Fe():ACvG complex were grown anaero-
bically as previously,18 although careful seeding allowed the
growth of larger crystals than had previously been obtained
(maximum dimensions 1.2 × 0.5 × 0.5 mm). For determination
of the anaerobic IPNS:Fe():ACvG structure, crystals were
flash-frozen in liquid nitrogen directly upon removal from the
anaerobic environment. Crystals for oxygenation experiments
were transferred to the pressurisation device on a glass cover
slip and reaction was initiated at room temperature by exposure
to dioxygen at 20 bar.10 A range of pressurisation times was
investigated, with the best result obtained after a 410 minute
oxygenation. Following oxygen exposure the crystals were
returned to atmospheric pressure and rapidly flash-frozen. The
frozen crystals were stored under liquid nitrogen until required
for data collection.
Experimental
Synthesis of the substrate analogue
-Vinylglycine hydrochloride salt was prepared from N-Z--
methionine methyl ester by periodate oxidation to the sulfoxide,
flash vacuum pyrolysis and acid deprotection (Fig. 4a).13 The
amino acid was protected as its benzhydryl ester,14 then con-
verted to the fully protected tripeptide by standard methods
(Figs. 4b, c).15,16 Finally, acid-mediated deprotection afforded
the tripeptide substrate analogue ACvG 3.17 The crude product
was purified by reversed phase HPLC [ODS (250 × 10 mm),
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 4 5 5 – 1 4 6 0
1459

View Article Online
Data collection
References
Data were collected at 100 K using synchrotron radiation
and an ADSC Quantum 4 detector at beamline ID14-EH4
of the European Synchrotron Radiation Facility, Grenoble,
1 J. E. Baldwin and C. J. Schofield, in The biosynthesis of β-lactams,
ed. M. I. Page, Blackie, Glasgow, 1992, p. 1–78.
2 J. E. Baldwin, Spec. Publ. - R. Soc. Chem., 1985, 52, 62–85.
3 N. I. Burzlaff, P. J. Rutledge, I. J. Clifton, C. M. H. Hensgens,
M. Pickford, R. M. Adlington, P. L. Roach and J. E. Baldwin,
Nature, 1999, 401, 721–724.
4 J. E. Baldwin and M. Bradley, Chem. Rev., 1990, 90, 1079–1088.
5 J. E. Baldwin, R. M. Adlington, A. Basak, S. L. Flitsch,
A. K. Forrest and H.-H. Ting, J. Chem. Soc., Chem. Commun., 1986,
273–275.
6 J. E. Baldwin, R. M. Adlington, S. L. Flitsch, H.-H. Ting and
N. J. Turner, J. Chem. Soc., Chem. Commun., 1986, 1305–1308.
7 J. E. Baldwin, R. M. Adlington, L. G. King, M. F. Parisi,
W. J. Sobey, J. D. Sutherland and H.-H. Ting, J. Chem. Soc., Chem.
Commun., 1988, 1635–1637.
8 S. D. Abbott, 1991, D Phil Thesis, University of Oxford.
9 J. M. Ogle, I. J. Clifton, P. J. Rutledge, J. M. Elkins, N. I. Burzlaff,
R. M. Adlington, P. L. Roach and J. E. Baldwin, Chem. Biol., 2001,
8, 1231–1237.
10 P. J. Rutledge, N. I. Burzlaff, J. M. Elkins, M. Pickford, J. E. Baldwin
and P. L. Roach, Anal. Biochem., 2002, 308, 265–268.
11 P. L. Roach, I. J. Clifton, C. M. H. Hensgens, N. Shibata,
C. J. Schofield, J. Hajdu and J. E. Baldwin, Nature, 1997, 387, 827–
830.
12 P. L. Roach, I. J. Clifton, V. Fülöp, K. Harlos, G. J. Barton, J. Hajdu,
I. Andersson, C. J. Schofield and J. E. Baldwin, Nature, 1995, 375,
700–704.
13 D. H. Rich and J. P. Tam, J. Org. Chem., 1977, 42, 3815–3820.
14 S. Wolfe and M. G. Jokinen, Can. J. Chem, 1977, 57, 1388–1396.
15 J. E. Baldwin, S. R. Herchen, B. L. Johnson, M. Jung, J. J. Usher and
T. Wan, J. Chem. Soc., Perkin Trans. 1, 1981, 2253–2257.
16 J. C. Sheehan, J. Preston and P. A. Cruickshank, J. Am. Chem. Soc.,
1965, 87, 2492–2493.
France (IPNS:Fe():ACvG structure), or
a
300 mm
Mar Research Image Plate at Station 7.2 of the Synchrotron
Radiation Source, Daresbury, UK (IPNS:Fe():HMPen
structure).
Structure determination
Data was processed using MOSFLM19 and the CCP4 suite.20
Refinement was carried out using REFMAC5.21 The program
O22 was used for model building. Initial rigid body refinement
was followed by refinement of all atoms and individual atomic
temperature factors. Manual rebuilding of protein side-chains
was performed as necessary.
For the IPNS:Fe():ACvG structure, electron density
for the substrate ACvG was clearly visible throughout
refinement. Data from the oxygenated crystal gave a struc-
ture with active site electron density consistent with high
occupancy of a bicyclic penam molecule (HMPen), and
possible occupancy of the unreacted substrate ACvG at a low
level.
For the unexposed IPNS:Fe():ACvG structure, the initial
refinements of positional and individual isotropic temper-
ature factors were followed by refinement of individual
anisotropic temperature factors when the model was complete.
For the IPNS:Fe():HMPen structure, the final round of
refinement included refinement of TLS parameters.23 For
the HMPen molecule, coordinate restraints were generated
from selected analogous structures in the Cambridge Struc-
tural Database (CSD) of the Chemical Database Service
(CDS).24 Data collection and refinement statistics are given in
Table 1.
17 M. Bodanzky and A. Bodanzky, in The practice of peptide synthesis,
ed. K. Hafner, C. W. Rees, B. M. Trost, J. M. Lehn, R. v. R. Schleyer
and R. Zahruchik, Springer-Verlag, Berlin, 1984, p. 172.
18 P. L. Roach, I. J. Clifton, C. M. H. Hensgens, N. Shibata, A. J. Long,
R. W. Strange, S. S. Hasnain, C. J. Schofield, J. E. Baldwin and
J. Hajdu, Eur. J. Biochem., 1996, 242, 736–740.
19 A. G. W. Leslie, Acta Crystallogr., Sect D, 1999, 55, 1696–1702.
20 Collaborative Computational Project Number 4, Acta Crystallogr.,
Sect. D, 1994, 50, 760–763.
21 G. N. Murshudov, A. A. Vagin, A. Lebedev, K. S. Wilson and
E. J. Dodson, Acta Crystallogr., Sect. D, 1999, 55, 247–255.
22 T. A. Jones, J. Y. Zou, S. W. Cowan and M. Kjeldgaard,
Acta Crystallogr., Sect. A, 1991, 47, 110–119.
23 M. D. Winn, M. N. Isupov and G. N. Murshudov, Acta Crystallogr.,
Sect. D, 2001, 57, 122–133.
24 D. A. Fletcher, R. F. McMeeking and D. Parkin, J. Chem. Inf.
Comput. Sci., 1996, 36, 746–749.
Acknowledgements
We would like to thank the beamline scientists at the
SRS Daresbury and the ESRF Grenoble for their assis-
tance. This work was supported by the MRC, BBSRC and
EPSRC, UK. P.J.R. was supported by the Rhodes Trust, N.I.B.
by a German DAAD fellowship and P.L.R. by the Royal
Society.
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 4 5 5 – 1 4 6 0
1460