Structural flexibility in Trypanosoma brucei enolase
revealed by X-ray crystallography and molecular dynamics
Marcos V. de A. S. Navarro
1,
*
,‡
, Sandra M. Gomes Dias
1,
*
,§
, Luciane V. Mello
2,3,
*,
Maria T. da Silva Giotto
1,†
, Sabine Gavalda
4,
–, Casimir Blonski
4
, Richard C. Garratt
1
and Daniel J. Rigden
2
1 Instituto de Fı
´sica de Sa˜o Carlos, Universidade de Sa˜o Paulo, Sa˜o Carlos SP, Brazil
2 School of Biological Sciences, University of Liverpool, UK
3 Northwest Institute for Bio-Health Informatics, University of Liverpool, UK
4 Groupe de Chimie Organique Biologique, Universite
´Paul Sabatier, Toulouse, France
Enolase (2-phospho-d-glycerate hydrolase, EC 4.2.1.11)
catalyses the reversible dehydration of d-2-phospho-
glycerate to phosphoenolpyruvate (PEP) and partici-
pates in both glycolysis and gluconeogenesis. In
common with most glycolytic enzymes, enolases from a
wide variety of organisms, including Archaea, Bacteria
Keywords
crystal structure; drug design; enolase;
molecular dynamics; structural flexibility
Correspondence
D. J. Rigden, School of Biological Sciences,
Crown Street, University of Liverpool,
Liverpool L69 7ZB, UK
Fax: +44 151 7954406
Tel: +44 151 7954467
E-mail: drigden@liv.ac.uk
Website: http://www.liv.ac.uk/biolsci/
*These authors contributed equally to this
work
†Deceased
Present address
‡Laborato
´rio Nacional de Luz Sı
´ncrotron,
Campinas, SP, Brazil
§Department of Molecular Medicine,
College of Veterinary Medicine, Cornell
University, Ithaca, NY, USA
–Department of Molecular Mechanisms of
Mycobacterial Infections, Institut de Phar-
macologie et de Biologie Structurale, CNRS,
UPS (UMR5089), Toulouse, France
(Received 5 June 2007, revised 25 July
2007, accepted 3 August 2007)
doi:10.1111/j.1742-4658.2007.06027.x
Enolase is a validated drug target in Trypanosoma brucei. To better charac-
terize its properties and guide drug design efforts, we have determined six
new crystal structures of the enzyme, in various ligation states and confor-
mations, and have carried out complementary molecular dynamics simula-
tions. The results show a striking structural diversity of loops near the
catalytic site, for which variation can be interpreted as distinct modes of
conformational variability that are explored during the molecular dynamics
simulations. Our results show that sulfate may, unexpectedly, induce full
closure of catalytic site loops whereas, conversely, binding of inhibitor
phosphonoacetohydroxamate may leave open a tunnel from the catalytic
site to protein surface offering possibilities for drug development. We also
present the first complex of enolase with a novel inhibitor 2-fluoro-2-phos-
phonoacetohydroxamate. The molecular dynamics results further encour-
age efforts to design irreversible species-specific inhibitors: they reveal that
a parasite enzyme-specific lysine may approach the catalytic site more
closely than crystal structures suggest and also cast light on the issue of
accessibility of parasite enzyme-specific cysteines to chemically modifying
reagents. One of the new sulfate structures contains a novel metal-binding
site IV within the catalytic site cleft.
Abbreviations
EV, eigenvector; FPAH, 2-fluoro-2-phosphonoacetohydroxamate; PAH, phosphonoacetohydroxamate; PDB, protein databank; PEP,
phosphoenolpyruvate.
FEBS Journal 274 (2007) 5077–5089 ª2007 The Authors Journal compilation ª2007 FEBS 5077
and Eukarya, are highly conserved [1]. The catalytic site
is particularly well conserved, leading to broadly similar
kinetic parameters for enzymes of different origins [2,3].
The quaternary structure of enolase is typically a
homodimer, although some bacteria apparently contain
octameric enzymes [4,5].
Each subunit of enolase contains an eightfold b⁄a
barrel domain preceded by an N-terminal a+b
domain [6]. The catalytic site is contained completely
within a single subunit and lies at the interface of the
two domains: monomeric enolase is catalytically active
[7]. Catalysis results from acid–base chemistry involving
a Lys-Glu dyad [8,9]. Also essential is the binding of
two divalent metal ions to distinct sites: the first ‘con-
formational’ site being required for substrate binding
and the second ‘catalytic’ site, occupied after substrate
has bound, stabilizing the reaction intermediate [6].
This ordered binding is accompanied by dramatic rear-
rangements of three protein loops lying near the cata-
lytic site. Note that, although the conventional loop
nomenclature is maintained here, regular secondary
structure is sometimes present in these regions. Briefly,
when the catalytic site is occupied by sulfate, phosphate
or phosphoglycolate, all three loops typically adopt an
open conformation, as seen, for example, in our previ-
ous Trypanosoma brucei enolase structure [10]. When
occupied by substrate, or the phosphonoacetohydroxa-
mate (PAH) inhibitor, and two metal ions, the loops
are generally all in a closed conformation, as in some
yeast structures [11]. Intermediate semiclosed confor-
mations have been observed when one metal ion is
absent or in some complexes with PEP [12,13].
As well as its key roles in glycolysis and gluconeo-
genesis, enolase, in common with other glycolytic
enzymes [14], has a remarkable number of ‘moonlight-
ing’ roles in diverse organisms that are unrelated to its
catalytic activity [15]. These include roles in the RNA
degradosome in Escherichia coli [16], as a structure
lens protein (s-crystallin) in the eye [17], as a transcrip-
tion factor in both animals [18] and plants [19] and, on
cell surfaces, as a receptor for plasminogen [15]. In this
last role, the expression of enolase on the surface of
streptococci is particularly interesting, where its inter-
action with host plasminogen is presumed to facilitate
entry of the parasite into host tissues [20]. Very
recently, the enolase of the trypanosomatid parasite
Leishmania mexicana has also been detected on the cell
surface [21]. A role for enolase as plasminogen recep-
tor in this organism is highly plausible because inter-
action between parasite and plasminogen has been
demonstrated [22].
Our interest in T. brucei enolase [2,10] stems from
the promise of the glycolytic pathway as a target for
drugs against parasitic protozoa [23]. With few excep-
tions, homologues of the enzymes involved are present
in the human host, and a premium is placed on seek-
ing and exploiting structural differences between para-
site and host proteins. Irreversible inhibition is
particularly desirable because it would be impervious
to high substrate levels that could displace competitive
inhibitors [23]. Using parasite enzyme-specific residues
(e.g. lysines in both cases), selective inhibitors against
aldolase [24] and phosphofructokinase [25] have been
developed. Despite bearing chemically reactive groups,
by combining high affinity and low reactivity, opti-
mized inhibitors of this kind should have minimal
effects on other proteins in vivo. Indeed, a prodrug
version of an aldolase inhibitor kills parasite cells
without detectable cytotoxicity against human MRC-5
cells [26].
Like other glycolytic enzymes, T. brucei enolase has
been validated as a drug target: RNA interference of
enolase in the bloodstream form of the parasite leads
to an effect on its growth within 24 h and death com-
mences at approximately 48 h [27]. Encouragingly, the
same study also demonstrated that a reduction in eno-
lase activity to approximately 15–20% of its original
level was sufficient for cell death to occur. This sug-
gests that incomplete inhibition of this enzyme in vivo
might prove sufficient for effective treatment. The pres-
ence of homologous enolase isoenzymes in the human
host raises the complication of selectivity. In this
respect, enolase is not the best target because the para-
site and host enzymes share 58% sequence identity.
Nevertheless, modelling showed that there are three
particularly interesting T. brucei enzyme residues, two
cysteines (numbered 147 and 241) and lysine 155, near
to the catalytic site, which are not conserved in the
human enzymes [2] (Fig. S3). The chemical characteris-
tics of the side chains of these residues offer the poten-
tial for species-selective permanent target inactivation
by appropriately designed covalent inhibitors. The
T. brucei crystal structure suggested that the cysteines
were almost entirely solvent inaccessible, yet, most sur-
prisingly, at least Cys147 could be chemically modified
by iodoacetamide with consequent enzyme inhibition
[10]. In that crystal structure, Lys155 is pointed away
from the catalytic site, being unfavourably positioned
to make additional interactions with a catalytic site-
bound inhibitor.
In the present study, we present six new enolase
structures that enhance our understanding of the struc-
tural and dynamic properties of the T. brucei enolase
catalytic site, which is essential for further drug design.
The new structures demonstrate that the enzyme can
adopt three distinct catalytic site structures in the
Structural flexibility in T. brucei enolase M. V. A. S. Navarro et al.
5078 FEBS Journal 274 (2007) 5077–5089 ª2007 The Authors Journal compilation ª2007 FEBS
sulfate-bound form, including one containing a novel
metal binding site. Furthermore, they show structural
heterogeneity in their inhibitor-bound forms, highlight-
ing the potential to extend future inhibitors out of
the ligand-binding pocket. We also present extensive
molecular dynamics simulations aiming to address how
the apparently buried cysteine residues achieve solvent
accessibility and show that Lys155 may indeed offer a
useful alternative possibility for covalent inhibition.
Results and Discussion
Overview of the new structures
Characteristics and statistics of data collection and
refinement for the six new structures are presented in
Table 1. The crystal form is the same in each case,
namely the C222
1
form previously reported [10],
although the precipitant used was PEG 1000 rather
than the PEG monomethylether 550. In the subsequent
analyses, we compare these structures with the previ-
ously published sulfate-bound structure, refined to
2.35 A
˚, and containing Zn
2+
ions bound to sites I and
III [10], which we refer to here as sulfate_1. The new
structures, all obtained by co-crystallization, are all of
significantly better resolution than sulfate_1, in partic-
ular a complex with PAH inhibitor that diffracted well
to 1.65 A
˚. In common with previous structures, from
T. brucei and other organisms, a single Arg residue,
numbered 400 in T. brucei, lies in the disallowed region
of the Ramachandran plot [10]. Among the three
important catalytic site loops previously described (and
discussed further below), there is only one that makes
a crystal contact. This involves Glu272 of loop 3, its
last residue and the most distant from the catalytic
site. Thus, we can be confident that the conformations
observed represent readily achieved structures of the
native enzyme, rather than crystal packing artefacts. In
our initial sulfate_1 structure, density did not allow for
chain tracing of two stretches, from Thr41-Gly42 and
Thr260-Pro266, regions that are frequently poorly
ordered in other enolase structures. With the exception
of sulfate_2, all the structures presented here could be
unambiguously fully traced. In sulfate_2, density did
not allow for the tracing of the polypeptide chain
between Asp251 and Gln273 inclusive. As with sul-
fate_1, one or two artefactual residues preceding the
N-terminal Met of the natural sequence could be
traced in each new structure, and these result from
thrombin cleavage of the His-tag used in purification
(see Experimental procedures). In the three inhibitor-
bound structures, artefactual Zn
2+
ions bound, with
partial occupancy (0.5–0.7), at the crystal packing
interface to residues ‘His0’ and Glu27, and to His283
from a crystal symmetry-related chain.
The new structures are diverse in the contents of their
catalytic sites, both in terms of substrate ⁄inhibitor and
in terms of bound metal (Table 1). The two new sulfate
complexes and the previous sulfate-bound structure
were all achieved at highly similar crystallization condi-
tions (Table 1). As such, there is no clear explanation
why they should differ in conformation (see below) and
we view their structural diversity as being the result of
‘freezing out’ of similarly accessible catalytic site
conformations. Substrate (PEP) and inhibitor (phos-
phonoacetohydroxamate, PAH) [28] bind with their
phospho and phosphono groups, respectively, occupy-
ing the same position as that occupied by sulfate in the
earlier sulfate_1 structure [10]. Schematic diagrams of
the interactions of inhibitors and metal ions with eno-
lase are given in Fig. S4. The binding mode of PEP
seen is essentially the same fully closed conformation as
that seen for yeast enolase [protein databank (PDB)
code 1one][29]. One PAH structure is also fully closed,
as in an earlier yeast complex (PDB code 1ebg) [11],
whereas the second, as discussed below, represents a
novel conformation for bound PAH. Electron density
maps for each complex are given in Fig. S5.
The novel compound 2-fluoro-2-phosphonoaceto-
hydroxamate (FPAH), a derivative with a pK
a
value
more resembling that of the phosphate of substrate
PEP, was also synthesized and its complex determined.
It is a competitive inhibitor of enolase which, despite
its lower pK
a
value compared to PAH, binds more
poorly with a K
i
at pH 7.2 of 1.4 lmcompared to
approximately 15 nmfor PAH (see supplementary
Doc. S1 and Figs S1 and S2) [30]. It binds in the same
way as PEP and PAH with an electron density suggest-
ing that both isomers of the R,S racemic mixture bind
equally well (Fig. S5). Despite the uniformity of ligand
binding, protein structure varies considerably at the
active site in the new set of structures. Rather than try
to explain their differences in the typical qualitative
way (i.e. open, closed, semiopen, loose, etc.), we
attempt a more quantitative description.
As shown in Figs 1 and 2, the principal conforma-
tional differences between the structures lie in three
catalytic site loops, 1–3, corresponding to those high-
lighted in many other studies. However, unlike the
results obtained in a similar analysis for Saccharo-
myces cerevisiae crystal structures (data not shown), a
fourth peak for the region from residues 215–220 is
present. This loop is a neighbour of loop 2 and moves
in a coordinated way in T. brucei but not in yeast
structures. Because loop 4 is distant from the catalytic
site, it is not discussed further.
M. V. A. S. Navarro et al.Structural flexibility in T. brucei enolase
FEBS Journal 274 (2007) 5077–5089 ª2007 The Authors Journal compilation ª2007 FEBS 5079
Table 1. Crystallisation conditions, metal content, data collection statistics and refinement statistics for structures of T. brucei enolase.
Name of structure
Sulfate_1
a
Sulfate_2 Sulfate_3 PEP PAH_1 PAH_2 FPAH
Crystallization conditions 0.1 MMes pH 6.5, 0.1 MMes pH 6.5, 0.1 MMes pH 6.5, 0.1 MMes pH 6.5, 0.1 MMes pH 6.5, 0.1 MMes pH 5.0, 0.1 MMes pH 5.0,
10 mMZnSO
4
,10mMZnSO
4
,10mMZnSO
4
,10mMZnSO
4
,10mMZnSO
4
,10mMZnCl
2
,10mMZnCl
2
,
25% PEGMME550 10% PEG1000 10% PEG1000 10% PEG1000 10% PEG1000 10% PEG1000 10% PEG1000
Metal content
Metal ion sites occupied at catalytic site 1, 3 1, 4 1, 2 1, 2 1, 2 1, 2 1, 2
Occupancy of catalytic site metals 1.0, 0.7 1.0, 1.0 0.9, 0.8 1.0, 1.0 1.0, 1.0 1.0, 1.0 1.0, 1.0
Data collection
Space group C222
1
C222
1
C222
1
C222
1
C222
1
C222
1
C222
1
Unit cell dimensions 74.02 73.62 74.77 73.99 73.86 74.95 74.81
a(A
˚)
b(A
˚) 110.54 111.26 111.17 110.72 109.28 110.76 110.64
c(A
˚) 109.1 109.97 108.98 109.3 107.95 109.22 109.01
Low resolution diffraction limit (A
˚) 38.8 26.0 26.1 25.0 28.0 17.8 21.0
High resolution diffraction limit (A
˚) 2.3 1.90 1.90 2.00 1.65 1.90 1.80
Lower resolution limit of highest
resolution bin
2.38 2.02 2.02 2.11 1.74 2.02 1.99
Completeness (%) 97.2 (93.4)
b
99.0 (96.7) 99.4 (100.0) 99.8 (100.0) 95.5 (91.4) 98.8 (96.1) 99.9 (100)
I⁄r(I) 14.2 (1.9) 17.0 (3.9) 15.2 (2.2) 12.9 (1.9) 26.4 (3.9) 16.1 (3.1) 23.3 (4.3)
Multiplicity 4.7 (4.5) 3.2 (3.2) 4.9 (4.8) 2.9 (2.8) 8.3 (8.3) 4.6 (4.6) 7.6 (7.7)
R
merge
(%) 6.52 (46.8) 5.8 (54.2) 5.7 (49.7) 6.4 (54.5) 5.5 (40.9) 7.0 (40.3) 5.6 (43.5)
Refinement
Number of water molecules 240 308 382 179 421 257 356
Number of reflections 17334 (1942) 33900 (4812) 33961 (4764) 28944 (4161) 47566 (6562) 33851 (4700) 40063 (5784)
R(%) 21.0 (28.5) 21.4 (27.1) 16.4 (25.3) 17.3 (24.7) 16.5 (32.4) 16.5 (20.7) 16.2 (22.2)
R
free
(%) 25.1 (33.7) 25.0 (30.0) 20.6 (33.7) 22.4 (37.1) 20.3 (39.7) 20.5 (30.1) 20.6 (31.3)
Mean temperature factor B (A
˚
2
) 43.7 28.0 21.2 37.2 29.6 24.3 27.9
All atoms
Protein 42.7 28.1 20.1 37.3 28.5 23.9 27.2
Ligand 71.3 33.1 16.1 30.2 22.7 28.9 21.0
Zinc 56.2 39.0 21.6 33.9 27.9 29.2 25.5
Solvent 42.7 26.9 29.9 36.1 38.5 29.9 35.5
rmsd from ideal values 0.007 0.019 0.015 0.018 0.017 0.016 0.018
Bond lengths (A
˚)
Bond angles () 1.4 1.672 1.505 1.680 1.668 1.487 1.611
Ramachandran (%)
c
Most favoured regions 86.4 89.8 89.4 91.1 90.0 90.0 90.5
Additional allowed regions 13.4 9.0 10.3 8.4 9.5 9.8 8.9
Generously allowed regions 0 0.9 0 0.3 0.3 0 0.3
Disallowed regions 0.3 0.3 0.3 0.3 0.3 0.3 0.3
a
See [10]; PDB code 1oep.
b
Values within parentheses are for the highest resolution bin.
c
Calculated with PROCHECK [50].
Structural flexibility in T. brucei enolase M. V. A. S. Navarro et al.
5080 FEBS Journal 274 (2007) 5077–5089 ª2007 The Authors Journal compilation ª2007 FEBS
Borrowing a technique more commonly associated
with molecular dynamics studies, we analysed the con-
formational differences in the new set of structures
using essential dynamics [31]. This also allowed us to
visualize to what extent the resulting modes of confor-
mational variability were explored during molecular
dynamics simulations (see later). The positions of the
six new structures projected onto eigenvectors (EVs) 1
and 2 are shown in Fig. 3A. Visual inspection of the
maximum and minimum projections of EV1 shows
A
B
Fig. 3. (A) Projections of the six new crystal structures and molec-
ular dynamics trajectories on to EVs 1 and 2 resulting from the
essential dynamics analysis. Blue circles are used for sulfate struc-
tures [open for sulfate_1* (see Experimental procedures), filled for
sulfate_3], green triangles for PAH complexes (open for PAH_1,
filled for PAH_2), a magenta square for the PEP structure and an
orange diamond for the FPAH complex. Black dots mark the
PEP + 2 Mg trajectory and red dots the single Mg trajectory start-
ing from the same PEP complex protein conformation. Dots are
shown at 2 ps intervals along the trajectory. (B) Path of the single
Mg trajectory, indicated at 20 ps intervals, showing a structural
switch at approximately 5 ns from a closed (low values for EVs 1
and 2) to an open structure (high EVs). The trajectory start is
marked with a circle whereas the end is indicated by a square.
Fig. 2. Comparison of sulfate_1, sulfate_2, PEP and PAH_1 struc-
tures, coloured, respectively, in shades of green, blue, magenta
and orange. The FPAH ligand position closely resembles that of
PAH_1. A complete cartoon representation of sulfate_2 is shown.
Backbone structure is shown for the other three structures only for
loops 1–4, which are labelled. Note the gaps in loops 1 and 3 of
the sulfate_1 structure and the loop 3 gap in the sulfate_2 struc-
ture. Side chains of Lys155 and His156 are shown as sticks, as are
the structures’ respective ligands showing the overlay of bound sul-
fate with phospho and phosphono groups. Zinc atoms are shown
as spheres occupying the labelled sites I–IV. Black dashes mark
the hydrogen bonding interactions of His156 with PEP or with met-
als in sites III or IV in the sulfate_1 and sulfate_2 structures,
respectively.
Fig. 1. Multi-rms plot of the enolase structures in Table 1 produced
with LSQMAN[56]. The multi-rms value is defined as the rms value of
the distances between all unique pairs of Caatoms for a given resi-
due. Loops 1–4 (see text) are labelled. Note that the value for a
section of loop 3 is artificially low in a stretch, coloured grey, for
which density did not allow tracing of the chain in either of the
open forms, sulfate_1 or sulfate_2.
M. V. A. S. Navarro et al.Structural flexibility in T. brucei enolase
FEBS Journal 274 (2007) 5077–5089 ª2007 The Authors Journal compilation ª2007 FEBS 5081