Human Δ3,Δ2-enoyl-CoA isomerase, type 2: A structural enzymology study on the catalytic role of its ACBP domain and helix-10

Human D³,D²-enoyl-CoA isomerase, type 2: a structural

enzymology study on the catalytic role of its ACBP

domain and helix-10

Goodluck U. Onwukwe¹, Petri Kursula^1,2, M. Kristian Koski¹, Werner Schmitz³and Rik K. Wierenga¹

1 Biocenter Oulu and Faculty of Biochemistry and Molecular Medicine, University of Oulu, Finland

2 Department of Biomedicine, University of Bergen, Norway

€

3 Theodor Boveri Institute of Biosciences (Biocenter), University of Wurzburg, Germany

The catalytic domain of the trimeric human D³,D²-enoyl-CoA isomerase,

Keywords

crotonase; ECI; helix-10; oxyanion hole;

SAXS

type 2 (HsECI2), has the typical crotonase fold. In the active site of this

fold two main chain NH groups form an oxyanion hole for binding the

thioester oxygen of the 3E- or 3Z-enoyl-CoA substrate molecules. A cata-

lytic glutamate is essential for the proton transfer between the substrate

C2 and C4 atoms for forming the product 2E-enoyl-CoA, which is a key

intermediate in the b-oxidation pathway. The active site is covered by the

C-terminal helix-10. In HsECI2, the isomerase domain is extended at its

N terminus by an acyl-CoA binding protein (ACBP) domain. Small angle

X-ray scattering analysis of HsECI2 shows that the ACBP domain pro-

trudes out of the central isomerase trimer. X-ray crystallography of the

isomerase domain trimer identiﬁes the active site geometry. A tunnel,

shaped by loop-2 and extending from the catalytic site to bulk solvent,

suggests a likely mode of binding of the fatty acyl chains. Calorimetry

data show that the separately expressed ACBP and isomerase domains

bind tightly to fatty acyl-CoA molecules. The truncated isomerase variant

(without ACBP domain) has signiﬁcant enoyl-CoA isomerase activity;

however, the full-length isomerase is more efﬁcient. Structural enzymolog-

ical studies of helix-10 variants show the importance of this helix for efﬁ-

cient catalysis. Its hydrophobic side chains, together with residues from

loop-2 and loop-4, complete a hydrophobic cluster that covers the active

site, thereby ﬁxing the thioester moiety in a mode of binding competent

for efﬁcient catalysis.

Correspondence

R. K. Wierenga, PO Box 5400, University of

Oulu, FI-90014 Oulu, Finland

Fax: +358 8 344 064

Tel: +358 2 94481199

E-mail: rik.wierenga@oulu.ﬁ

(Received 24 September 2014, revised 10

December 2014, accepted 12 December

2014)

doi:10.1111/febs.13179

Database

Structural data are available in the PDB database under the accession numbers

4U18 (ISO-ECI2), 4U19 (ISOA-ECI2), 4U1A (ISOB-ECI2)

Abbreviations

ACBP, acyl-CoA binding protein; ACBP-ECI2, the ACBP domain of HsECI2; DECI, dienoyl-CoA isomerase (mitochondrial and peroxisomal);

DmMFE2, the Drosophila melanogaster MFE2; ECH, enoyl-CoA hydratase-1 (mitochondrial); ECI1, D³,D²-enoyl-CoA isomerase type 1

(mitochondrial); ECI2, D³,D²-enoyl-CoA isomerase, type 2 (mitochondrial and peroxisomal); ECI, D³,D²-enoyl-CoA isomerase; FL-ECI2, the

full-length variant of HsECI2; HsECI2, the human ECI2; ISOA-ECI2, the V349A variant of ISO-ECI2; ISOB-ECI2, the (D350-355) variant of

ISO-ECI2; ISO-ECI2, the isomerase domain variant of HsECI2; ITC, isothermal titration calorimetry; MFE1, multifunctional enzyme type 1

(peroxisomal); MFE2, multifunctional enzyme type 2 (peroxisomal); OAH, oxyanion hole; SAXS, small angle X-ray scattering; SCP2, sterol

carrier protein type 2; SGC, Structural Genomics Consortium; SLS, static light scattering.

746

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

G. U. Onwukwe et al.

Structural enzymological studies of human ECI2

whereas in mammals there are several isoenzymes,

located in both the peroxisomes and mitochondria.

The mammalian ECI type 1 (ECI1) is located only in

mitochondria. The mammalian ECI type 2 (ECI2) is

located in peroxisomes as well as in mitochondria.

Therefore, in mammals, there are at least two isome-

rases in mitochondria (ECI1 and ECI2). Also in per-

oxisomes there are two isomerases: ECI2 and the

multifunctional enzyme type 1 (MFE1). It is not

understood why mammals have several isomerase iso-

zymes. The kinetic properties of each of these isome-

rases from rat have been reported, showing a broad

range of fatty acyl chain length speciﬁcities and differ-

ent kinetic parameters for 3Z- and 3E-enoyl-CoA sub-

strates [6]. Rodents possess a third mitochondrial ECI,

which has been identiﬁed recently and which is homol-

ogous to ECI2 [7]. Structures have been described for

ECI1 from rat mitochondria (1XX4) [8] and human

mitochondria (1SG4) [9], MFE1 from rat peroxisomes

Introduction

D³,D²-enoyl-CoA isomerases (ECIs) catalyze an impor-

tant reaction which allows unsaturated fatty acids with

Z/E double bonds (also known as cis/trans double

bonds) at odd positions, like oleic acid (Z9-octadece-

noic acid) and linoleic acid (Z9,Z12-octadecadienoic

acid) [1], to enter the classical b-oxidation pathway for

the degradation of fatty acids (Fig. 1). A key interme-

diate of the latter pathway is 2E-enoyl-CoA. ECI (EC

5.3.3.8) catalyses the conversion of 3Z-enoyl-CoA and

3E-enoyl-CoA substrates to 2E-enoyl-CoA (Fig. 1).

Other enzymes involved in this auxiliary pathway are

D^3,5D^2,4-dienoyl-CoA isomerase (DECI) and 2,4-die-

noyl-CoA reductase [1,2]. This is an important path-

way, and several studies have reported on disease

symptoms related to enzyme deﬁciencies of mitochon-

drial 2,4-dienoyl-CoA reductase [3] and mitochondrial

ECI [4]. To the best of our knowledge, in the yeast

Saccharomyces cerevisiae there is only one ECI [5],

Fig. 1. The four reactions of the b-oxidation pathway. The D³,D²-enoyl-CoA isomerase reaction is the last reaction of the auxiliary pathway

for the synthesis of 2E-enoyl-CoA from unsaturated fatty acyl-CoA having double bonds at odd positions or having Z double bonds.

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

747

Structural enzymological studies of human ECI2

G. U. Onwukwe et al.

(2X58) [10] and ECI from yeast peroxisomes (1PJH)

[11]. Here, we report on structural enzymological stud-

ies of human ECI2 (HsECI2).

fatty acyl-CoA is known, adopting a bent conforma-

tion (Fig. 2). The CoA moiety binds between helix-10

and b-strands 1–3 of the b-sheet B. The loops emerg-

ing from the b-strands of b-sheet B and leading into

the covering a-helices, in particular loop-1, loop-2,

loop-3 and loop-4, play important roles in the CoA

binding as well as in the catalytic process of these

enzymes. Loop-2 is a long loop with variable length

(Fig. 3A). Loop-3 is very short as strand B3 continues

immediately into helix-3, which is the catalytic helix.

A common feature in the active sites of each of these

enzymes is an oxyanion hole (OAH), formed by two

peptide NH groups of two different loops, loop-2 and

loop-3/helix-3. Loop-2 is solvent exposed and is not

involved in intratrimer subunit–subunit interactions,

whereas the loop-3 region, being the N terminus of

helix-3, is rather buried. In MFE1 and in ECH, loop-2

is involved in binding the fatty acyl tail, whereas in

ECI1 the fatty acyl tail points towards loop-4 (between

loop-4 and helix-9). In ECI1, the C-terminal end of

loop-4, also referred to as the ﬂap region (Fig. 3A),

has adopted another conformation compared to

MFE1 and ECH (and also in yeast ECI) (Fig. 2).

ECI1 and ECI2 therefore form two distinct subfamilies

within the crotonase superfamily, as is also found from

phylogenetic relationships (Fig. 3B), showing also that

the yeast ECI belongs to the ECI2 subfamily. The

Each of the ECIs belongs to the crotonase fold

superfamily of enzymes [12,13]. Crotonase is the origi-

nal name of enoyl-CoA hydratase, type 1 (ECH), the

structure of which was one of the ﬁrst determined

from this superfamily [14], together with the structure

of a dehalogenase [15]. The crotonase fold is formed

by approximately 300 residues. The superfamily

includes many enzymes catalyzing a wide range of

rather different reactions, using CoA thioesters as the

substrate. Usually, these enzymes are trimers or hexa-

mers (dimers of trimers). MFE1 is a monomer with

two domains, and the enoyl-CoA isomerase active site

is in the N-terminal crotonase domain. In this fold,

there are two (mostly) parallel b-sheets (A and B). The

edge strands A5 and B5 run anti-parallel to the other

strands of these sheets (Fig. 2). The central b-sheet

(sheet A) provides

a structural framework. The

strands of this sheet, termed A1–A4, continue into the

b-strands of sheet B, termed B1–B4, and are then

followed by helices H1–H4 forming the bba-spiral

crotonase fold. The b-strands of sheet B are rather

short (Fig. 2).

For MFE1 [16], ECI1 [9] and ECH [17] (3ZWC,

1SG4 and 2DUB, respectively) the mode of binding of

Fig. 2. The crotonase fold. Structural comparison of the superimposed structures of the crotonase domain of rat ECH (2DUB, brown, with

bound octanoyl-CoA), MFE1 (3ZWC, N-terminal domain, lilac, with bound 3S-hydroxydecanoyl-CoA), human ECI1 (1SG4, lemon, with bound

octanoyl-CoA) and yeast ECI2 (1PJH, cyan, unliganded). The labeled helix-5 is at the subunit–subunit interface of the trimers of the

crotonase superfamily. The b-strand between helix-5 and helix-6 is the edge strand of sheet B, whereas some residues of loop-1 and loop-2

are forming the other edge of sheet B. Helix-7 shields b-sheet A from bulk solvent. The acyl-CoA is wedged between helix-10 and sheet B.

The two black arrows show the two different binding modes of the fatty acyl tail. The ‘left’ mode of binding is found in ECI1, whereas the

‘right’ mode of binding is seen in ECH and MFE1. The structures of ECI1 and ECI2 represent the canonical crotonase fold, assembled in a

trimer. MFE1 is a monomer and the additional helices, before helix-9 (helix-9A and the next helix), prevent trimerization. In ECH helices 9

and 10 have been swapped over to the neighboring subunit (not shown in this panel).

748

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

G. U. Onwukwe et al.

Structural enzymological studies of human ECI2

A

B

C

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

749

Structural enzymological studies of human ECI2

G. U. Onwukwe et al.

sequences of the enzymes of the crotonase superfamily

are rather diverse [12,13], as is also evident from the

sequence alignment of the sequences of human ECI2

with the hexameric yeast ECI2 (22% sequence iden-

tity), the monomeric rat MFE1 (19% sequence iden-

tity), the trimeric human ECI1 (17% sequence

identity), the hexameric rat ECH (23% sequence iden-

tity) and the hexameric rat DECI (21% sequence iden-

tity).

some other peroxisomal enzymes, such as in the mam-

malian sterol carrier protein type 2 (SCP2) thiolase

(having an extra SCP2 domain) [23] and in peroxi-

somal multifunctional enzyme type 2 (MFE2) (also

having an SCP2 domain) [24,25]. A common struc-

tural feature of every crotonase enzyme is helix-10.

This is the C-terminal helix, which is anchored at its

N terminus to the rest of the protein at the subunit–

subunit interface. Its C terminus covers the active site

and the CoA binding pocket and points into bulk sol-

vent, being disordered in many structures. A con-

served feature of helix-10 is the presence of one or

two basic residues, which interact with the phosphate

moieties of the substrate in enzyme–substrate com-

plexes [9,16,17].

In ECIs [8,9,18] the catalytic residue is a glutamate,

which abstracts a proton from the substrate C2 atom,

generating a negatively charged enolate [19]. The nega-

tive charge of the thioester oxygen atom is stabilized

by the OAH [20,21]. Subsequently, the abstracted pro-

ton is transferred to the C4 atom, generating the 2E-

enoyl-CoA product (Fig. 1). From sequence (Fig. 3)

and structural alignments (Fig. 2), it can be inferred

that this glutamate in HsECI2 is Glu245, protruding

out of the ﬂap region (end of loop-4, just before helix-

4) into the catalytic site, whereas the OAH is formed

by the peptide NH moieties of Asn165 (loop-2) and

Ile214 (helix-3).

It has been noted that the crotonase fold is fre-

quently used in nature to generate natural enzymes

covering a wide range of substrate and reaction

speciﬁcities, always using the OAH for its biocatalytic

properties [13,19]. This has recently been exploited for

the creation of non-natural enzymes [26–28] catalyzing

the formation of C–C bonds. The current study was

also undertaken to provide better understanding of

the functional importance of various modules, like

helix-10 and the ACBP domain, taking HsECI2 as the

reference molecule. Structural enzymological proper-

ties of HsECI2 were investigated by performing solu-

tion small angle X-ray scattering (SAXS), X-ray

crystallography, calorimetry and enzyme kinetics stud-

ies. The properties of the full-length construct

(FL-ECI2) have been compared with the properties of

a truncated variant lacking the ACBP domain (ISO-

ECI2) as well as with the ACBP domain itself (ACBP-

ECI2). In addition the properties of the truncated

enzyme derived from ISO-ECI2 in which helix-10 has

A peculiar property of mammalian ECI2 is the pres-

ence of the 80 residue long acyl-CoA binding (ACBP)

domain at its N terminus (Fig. 3A), which is con-

nected to the isomerase domain by a linker region.

This extra domain is not present in yeast ECI2. The

NMR structure of the ACBP domain of HsECI2 is

known (2CQU; Tsubota et al.; unpublished), having

the same four helical bundle fold as human liver

ACBP (2CB8) [22]. The four helices of the ACBP of

HsECI2 are labeled in Fig. 3A. This extra domain is

not present in yeast ECI2 and its function in mamma-

lian ECI2 is not clear. Such an extra lipid-molecule

binding domain of unknown function is also found in

Fig. 3. (A) The structure based sequence alignment of isomerases and hydratases of the crotonase superfamily. The secondary structure

above the sequence refers to the human ISO-ECI2 structure as assigned by DSSP. The secondary structure of the ACBP domain concerns

its NMR structure (2CQU). The sequence numbering refers to the Uniprot sequence concerning HsECI2 (O75521) and to the PDB entries of

S. cerevisae peroxisomal ECI2 (ScECI2, 1PJH), the crotonase domain of rat peroxisomal MFE1 (RnMFE1, 3ZWC), human mitochondrial

ECI1 (HsECI1, 1SG4), rat mitochondrial ECH (RnECH, 2DUB) and rat DECI (RnDECI, 1DCI). The fully conserved residues are labeled by an

asterisk. The residues forming the OAH are marked by a plus sign. The black dot (●), at Gly103, marks the ﬁrst residue of the ISO-ECI2,

ISOA-ECI2 and ISOB-ECI2 variants, preceded by the puriﬁcation tag. The C-terminal end of ISO-ECI2 and ISOA-ECI2 is marked with an

arrow (▼). The catalytic residues are highlighted in purple. The residues highlighted in red form the hydrophobic cluster near the thioester

and adenine moiety of the substrate, as known from structures of complexes, ﬁxing helix-10 with respect to loop-2 and loop-4. The

residues highlighted in green form the tunnel of the fatty acyl binding pocket. The basic residues of helix-10 known to interact in complexes

with the phosphate moieties of the ligand are colored in red. (B) The evolutionary relationships of the six sequences aligned in (A). The

percentages of replicate trees in which the associated taxa clustered together in the bootstrap test (10 000 replicates) are shown next to

the branches. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the

phylogenetic tree. (C) The modular structure of ECI2. The constructs used in these studies are visualized. Each of these constructs includes

a puriﬁcation tag just before Gly103 for the ISO-ECI2, ISOA-ECI2 and ISOB-ECI2 constructs, and just before Met1 for FL-ECI2 and ACBP-

ECI2. E245 is the catalytic glutamate. V349 is the mutated residue (V349A) of ISOA-ECI2.

750

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

G. U. Onwukwe et al.

Structural enzymological studies of human ECI2

Fig. 4. The CD melting curves showing the calculated percentages of unfolded protein as a function of temperature. The calculated T_m

values are 57.0 ꢁ 1.4 °C for FL-ECI2 (cyan), 55.4 ꢁ 0.1 °C for ISO-ECI2 (green), 51.4 ꢁ 0.1 °C for ACBP-ECI2 (red), 48.9 ꢁ 0.2 °C for

ISOA-ECI2 (purple) and 47.0 ꢁ 0.3 °C for ISOB-ECI2 (light blue).

been mutated, either by a point mutation V349A

(ISOA-ECI2) or by deleting the C-terminal residues

350–355 (ISOB-ECI2), were studied. The precise con-

structs that have been used are visualized in Fig. 3C.

Crystal structures of the truncated enzyme and its two

helix-10 variants show the active site geometry of

HsECI2 as well as the structural effects of mutations

in helix-10.

Table 1. The Michaelis–Menten parameters of FL-ECI2 and ISO-

ECI2. The standard deviations are based on three independent

experiments.

ꢀ1

)

Enzyme: substrate

FL-ECI2

K_m(lM)

k_cat(s^ꢀ1

)

k_cat/K_m(s^ꢀ1ꢂM

3E-hexenoyl-CoA 138.5 ꢁ 48.7 20.2 ꢁ 3.6 1.5 9 10⁵

3E-decenoyl-CoA

83.4 ꢁ 24.6 56.3 ꢁ 9.3 6.7 9 10⁵

ISO-ECI2

3E-hexenoyl-CoA 446.0 ꢁ 11.2 11.6 ꢁ 5.0 0.3 9 10⁵

3E-decenoyl-CoA

31.0 ꢁ 4.3

7.5 ꢁ 0.3 2.4 9 10⁵

Results and discussion

The puriﬁed FL-ECI2 and ISO-ECI2 behave as tri-

mers in solution as shown by static light scattering

(SLS), having molecular masses of 133.5 and 89 kDa,

respectively. Using circular dichroism (CD) spectros-

copy it was found that these constructs, as well as the

ACBP-ECI2 variant, are properly folded proteins. Sta-

bility studies have been done with FL-ECI2, ISO-ECI2

and ACBP-ECI2 by measuring the CD melting curves

as a function of temperature (Fig. 4). These data show

that ISO-ECI2 is a stable protein (melting temperature

T_m= 55.4 °C), being somewhat less stable than

FL-ECI2 (T_m= 57.0 °C), whereas the T_mof ACBP-

ECI2 is 51.4 °C. Enzyme kinetics measurements show

that ISO-ECI2 can isomerize 3E-hexenoyl-CoA and

3E-decenoyl-CoA to the corresponding 2E-acyl-CoA

molecules (Table 1). The Michaelis–Menten enzyme

kinetics parameters show that the ISO-ECI2 construct

is less active than the wild-type FL-ECI2 enzyme. For

example the k_catfor the 3E-decenoyl-CoA substrate is

56 s^ꢀ1for FL-ECI2 and 7.5 s^ꢀ1for ISO-ECI2,

whereas the K_mvalue changes somewhat, from 83 to

31 l^M. The variation of the kinetic properties of FL-

ECI2 from short-chain to long-chain substrates shows

the same trend as observed for the rat ECI2 [6]: the

k_catvalues increase and the K_mvalues decrease

for longer chain substrates. The differences of the

FL-ECI2 and ISO-ECI2 Michaelis–Menten parameters

indicate that the presence of the ACBP domain inﬂu-

ences the enzyme kinetics properties. Indeed, calori-

metric afﬁnity data using the saturated decanoyl-CoA

(substrate analogue) and 3E-decenoyl-CoA (substrate)

as ligand with the ACBP-ECI2 construct indicate that

the ACBP domain has high afﬁnity for both of these

ligands, with a K_dof 1.5 lM for the saturated deca-

noyl-CoA (substrate analogue) (Table 2) and 5 lM for

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

751

Structural enzymological studies of human ECI2

G. U. Onwukwe et al.

Table 2. The calorimetric data as obtained from the ITC experiments. The error estimates, difference from the mean, are based on two

independent experiments. NB indicates no binding detected.

ꢀ1

Enzyme

FL-ECI2

Substrate analogues K_a9 10⁵(M

)

K_d(lM) ꢀTΔS (kcalꢁmol^ꢀ1

)

ΔH (kcalꢁmol^ꢀ1

)

N (number of sites) ΔG (kcalꢁmol^ꢀ1

)

Octanoyl-CoA

Decanoyl-CoA

Lauroyl-CoA

1.92 ꢂ 0.84

7.10 ꢂ 1.90

27.45 ꢂ 2.01

2.13 ꢂ 0.95

0.032 ꢂ 0.01

0.27 ꢂ 0.16

0.82 ꢂ 0.23

0.65 ꢂ 0.25

1.90 ꢂ 0.86

6.57 ꢂ 0.11

5.21 ꢀ4.60 ꢂ 0.10

1.42 ꢀ3.61 ꢂ 1.10

0.36 ꢀ4.32 ꢂ 0.20

4.69 ꢀ4.23 ꢂ 0.40

ꢀ2.54 ꢂ 0.38 2.19 ꢂ 0.00

ꢀ4.36 ꢂ 0.86 2.03 ꢂ 0.01

ꢀ4.53 ꢂ 0.17 2.02 ꢂ 0.03

ꢀ2.96 ꢂ 0.09 3.87 ꢂ 0.09

ꢀ5.62 ꢂ 0.62 1.05 ꢂ 0.05

ꢀ4.74 ꢂ 0.91 1.33 ꢂ 0.09

ꢀ5.32 ꢂ 0.40 1.35 ꢂ 0.29

ꢀ4.12 ꢂ 0.67 1.69 ꢂ 0.59

ꢀ2.51 ꢂ 0.66 1.14 ꢂ 0.03

ꢀ3.15 ꢂ 0.48 0.84 ꢂ 0.16

ꢀ4.62 ꢂ 1.15 1.30 ꢂ 0.28

ꢀ4.02 ꢂ 0.05 1.52 ꢂ 0.04

ꢀ7.14

ꢀ7.97

ꢀ8.85

ꢀ7.19

ꢀ4.51

ꢀ5.91

ꢀ6.68

ꢀ6.52

ꢀ7.14

ꢀ7.93

ꢀ8.96

ꢀ8.98

NB

Myristoyl-CoA

Octanoyl-CoA

Decanoyl-CoA

Lauroyl-CoA

ISO-ECI2

313.00

1.11 ꢂ 0.22

37.04 ꢀ1.17 ꢂ 0.66

12.20 ꢀ1.36 ꢂ 0.24

15.39 ꢀ2.40 ꢂ 0.43

5.26 ꢀ4.63 ꢂ 0.38

1.52 ꢀ4.78 ꢂ 0.49

0.26 ꢀ4.34 ꢂ 0.97

0.26 ꢀ4.96 ꢂ 0.05

Myristoyl-CoA

ACBP-ECI2 Octanoyl-CoA

Decanoyl-CoA

Lauroyl-CoA

38.35 ꢂ 11.25

38.55 ꢂ 0.35

NB

Myristoyl-CoA

ISOA-ECI2

ISOB-ECI2

Octanoyl-CoA

Decanoyl-CoA

Lauroyl-CoA

NB

1.83 ꢂ 0.03

3.34 ꢂ 0.09

NB

0.40 ꢂ 0.07

0.94 ꢂ 0.04

NB

25.00

10.60

NB

ꢀ8.06 ꢂ 0.08 1.23 ꢂ 0.00

ꢀ10.11 ꢂ 0.14 1.14 ꢂ 0.04

ꢀ6.23

ꢀ6.77

NB

Octanoyl-CoA

Decanoyl-CoA

Lauroyl-CoA

NB

0.99 ꢂ 0.01

1.81 ꢂ 0.05

10.10

5.50

7.7 ꢂ 0.39

ꢀ14.51 ꢂ 0.38 0.66 ꢂ 0.10

ꢀ11.96 ꢂ 0.21 1.10 ꢂ 0.01

ꢀ6.81

ꢀ7.13

4.83 ꢂ 0.18

the substrate 3E-decenoyl-CoA. These dissociation

constants are at least 10-fold lower than the K_mvalues

measured for the 3E-decenoyl-CoA substrate with FL-

ECI2 (K_m= 83.4 lM) and ISO-ECI2 (K_m= 31.0 lM).

The calorimetric data also show that there is one bind-

ing site per ACBP molecule. Calorimetric experiments

with a range of substrate analogues with FL-ECI2

show that there are two binding sites per subunit, in

agreement with the binding of these ligands to both

the ACBP-ECI2 domain and the ISO-ECI2 domain.

The ACBP-ECI2 domain has much higher afﬁnity

than the ISO-ECI2 domain for each of the tested satu-

rated acyl-CoA substrate analogues (Table 2). This is

true for ligands with fatty acyl tails of different

lengths, varying from octanoyl-CoA to myristoyl-CoA,

and it is most pronounced for ligands having the short

tail lengths of octanoyl-CoA and decanoyl-CoA. For

example the K_dvalues of ACBP-ECI2 and ISO-ECI2

for the saturated decanoyl-CoA substrate analogue are

1.5 and 37 l^M, respectively.

The acyl-CoA binding sites of the ACBP and isom-

erase domains are rather different. Crystallographic

[22,29] and NMR [30] binding studies with ACBP have

revealed that the ACBP binding site is a groove, bind-

ing acyl-CoA in such a way that the adenine moiety is

stacked between a tyrosine and the fatty acyl tail,

which subsequently interacts with hydrophobic side

chains. In the spiral-like conformation the pantetheine

part is bulk solvent exposed, being modeled in the

NMR structure [30] but disordered in the crystal struc-

ture of the plasmodium ACBP complex [29]. To the

crotonase fold, acyl-CoA is bound in the characteristic

bent conformation (Fig. 2). The differences in the

mode of binding of acyl-CoA to ACBP and the cro-

tonase domain indeed result in different thermody-

namic proﬁles of acyl-CoA binding to the ACBP-ECI2

domain and the ISO-ECI2 domain. The binding of

acyl-CoA to the ISO-ECI2 domain is mainly enthalpy

driven, whereas the binding to the ACBP-ECI2

domain is mainly entropy driven (Table 2).

Structure of full-length HsECI2 in solution

To gain an insight into the assembly of the full-length

HsECI2 enzyme, structural studies were carried out in

solution using X-ray scattering. Different modeling

approaches were taken, ranging from ab initio bead-

and chain-like models to hybrid modeling utilizing

known structures of the isomerase trimer and the

ACBP domain. In essence, all modeling approaches

gave the same overall results, with very good ﬁts to

the raw X-ray scattering data.

The SAXS experiments show that both FL-ECI2

and ISO-ECI2 are trimers in solution (Table 3, Fig. 5).

The theoretical scattering curve from the ISO-ECI2

crystal structure ﬁts the corresponding SAXS data

well, validating the trimeric conformation observed in

the crystal structure. On the other hand, as expected,

the ACBP domain is a monomer (Table 3, Fig. 5); the

slight misﬁt between the SAXS data and the crystal

structure (Fig. 5A) relates to the fact that our ACBP

construct is signiﬁcantly longer, having an extra 27 res-

752

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

G. U. Onwukwe et al.

Structural enzymological studies of human ECI2

Table 3. SAXS analysis of the HsECI2 constructs. I(0) and R_gwere determined for FL-ECI2 from 4 to 12 samples in the absence and

presence of ligand, respectively. The molecular mass (MM) was calculated from the estimated Porod volume based on the partial speciﬁc

volume of 0.7425 cm³ꢁg^ꢀ1[55]. The chi values for ﬁtting the crystal structures to the raw SAXS data are from CRYSOL.

MM based on Porod

Monomer MM from MM based on

volume (kDa), ratio to Fit of structural model

ꢀ

R_g(A)

ꢀ

D_max(A) MM from sequence

Sample

sequence (kDa)

I(0) (kDa)

to SAXS data (chi)

FL-ECI2 (no ligand)

41.3

93 ꢂ 9

44.8 ꢂ 1.2 160

130, 3.1

0.89 (^BUNCH), 0.84 (CORAL),

1.5 (^GASBOR), 0.97 (DAMMIF)

FL-ECI2 (ligand)

ISO-ECI2

41.3

31.0

11.6

ꢀ

106 ꢂ 12

44.5 ꢂ 2.1

32.3

ꢀ

93

17

ꢀ

120

80

90

100, 3.2

12, 1.0

ꢀ

1.3 (GASBOR)

1.0 (^GASBOR)

1.3

ACBP-ECI2

19.2

ISO-ECI2 crystal

structure (this study)

ACBP crystal

27

ꢀ

13.6

46.2

–

1.8

structure (2FJ9)

idue long puriﬁcation tag, which is predicted to be dis-

ordered. The positions of the ACBP domains in the

full-length enzyme are such that the ACBP domains

do not physically interact with each other. The ACBP

domains extend away from the ISO-ECI2 core and lie

in the same plane as the ﬂattened isomerase domain

trimer, at the corners of a triangle (Fig. 5C & E).

Thus, the full-length enzyme is a highly ﬂattened tri-

mer in solution. Apparently, the ACBP domain adopts

a well deﬁned position with respect to the core trimer;

whether it has additional ﬂexibility or motility cannot

be assessed with the current data. In the FL-ECI2

model, the binding pockets of the ACBP and ISO-

ECI2 domains of the same polypeptide chain are fac-

formational changes in HsECI2 related to acyl-CoA

ligand binding.

The enzymological comparison of FL-ECI2 and

ISO-ECI2 suggests an in vivo function for the ACBP

domain; the ACBP domain of FL-ECI2 makes the

enzyme more proﬁcient compared to ISO-ECI2, espe-

cially for short-chain fatty acyl-CoA substrates. For

example, for the 3E-hexenoyl-CoA substrate, the k_cat

ꢀ1

/

K_mvalue changes from 1.5 9 10⁵

M

ꢁs^ꢀ1for FL-

ECI2 to 0.3 9 10⁵

M

ꢁs^ꢀ1for ISO-ECI2 (Table 1). It

ꢀ1

is not clear from the FL-ECI2 SAXS model how the

presence of the ACBP domain enhances catalytic proﬁ-

ciency, as conformational changes induced by ligand

binding were not detected. The large distance

(Fig. 5G) between the binding pockets of the ACBP

and isomerase domains in the same subunit indicates

that the fatty acyl moiety cannot reach its binding site

in the isomerase domain, while the adenine part is still

bound to ACBP. It is possible that the ACBP domain

initially captures the substrate and subsequently

delivers it via a transient movement towards the active

ꢀ

ing each other, being separated by 40 A, as measured

between the N3 atoms of the adenine moieties of the

superimposed models (Fig. 5G). The SAXS data for

the apo and complexed forms of FL-ECI2 show that

the shape of FL-ECI2 with and without ligand is the

same, as the scattering curves have identical shapes

(Table 3, Fig. 5F). Thus, there are no large-scale con-

Fig. 5. SAXS analysis of HsECI2. (A) SAXS scattering data for ISO-ECI2 (left), FL-ECI2 (middle) and ACBP-ECI2 (right). The shown ﬁts (from

left to right) represent the ISO-ECI2 crystal structure from this study, the ^BUNCHhybrid model of FL-ECI2 (see E) and the crystal structure of

human ACBP (2FJ9). The inset in the right graph shows a superposition of a GASBOR ab initio chain-like model of ACBP-ECI2 (gray) and the

ACBP crystal structure (2FJ9; green). The Guinier plots below each scattering curve indicate good linear ﬁtting for all samples. (B) The

ab initio ^GASBORmodel of ISO-ECI2 (gray) superimposed with the crystal structure of ISO-ECI2. (C) The GASBOR model of FL-ECI2 (gray) with

the superimposed crystal structure of the ISO-ECI2 trimer in the middle surrounded by three ACBP molecules (1ACA). The latter model is

based on the BUNCH model of (E) but the ACBP-ECI2 part was replaced by the coordinates of the NMR structure of bovine ACBP (1ACA)

complexed with palmitoyl-CoA (shown in orange). (D) Distance distribution functions for FL-ECI2 (black), ISO-ECI2 (red) and ACBP-ECI2

(blue). (E) Hybrid model of FL-ECI2 made using ^BUNCH. ACBP and isomerase domains are green and cyan, respectively. The linker regions

(Pro85-Glu105) and N and C termini are shown in magenta. (F) SAXS data for unliganded (topmost curve) and three liganded samples of FL-

ECI2 (top to bottom: octanoyl-CoA, decanoyl-CoA, lauroyl-CoA). (G) Zoomed-in view of the FL-ECI2 model showing the acyl-CoA binding

sites of the isomerase and ACBP domains. The model is the same as shown in (C) except that the dummy atom ^BUNCHmodel of the linker

region is also shown. The acyl-CoA product, 3S-hydroxydecanoyl-CoA, modeled in the isomerase domain is from the superimposed

structure of MFE1 hydratase (3ZWC). The red arrow marks the N terminus of the catalytic helix, helix-3.

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

753

Structural enzymological studies of human ECI2

G. U. Onwukwe et al.

A

B

C

D

E

F

G

754

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

G. U. Onwukwe et al.

Structural enzymological studies of human ECI2

site of ISO-ECI2. Alternatively, this transfer of sub-

strate can happen by diffusion over the surface from

the ACBP binding site to the isomerase binding site,

or by a combination of these two possibilities. Diffu-

sion through bulk solvent would require an additional

solvation and desolvation step. In any case, the rate-

limiting step in the complete isomerase reaction cycle

is unknown, and the function of the ACBP domain

could also be related to product release.

N(Asn165) and N(Ile214), which are the OAH forming

NH groups, while a chloride ion is bound in each of

the three subunits (Fig. 6A).

Extensive structural comparisons were done with the

liganded structures of MFE1, ECH and ECI1. The

active site loops, loop-1, the beginning of loop-2, loop-

3 and the beginning of loop-4, adopt the same confor-

mations as is known from yeast ECI2, MFE1 and

ECH (Figs 6 and 7). The catalytic glutamate points

from the ﬂap region of loop-4 into the active site. The

structures of MFE1, complexed with 3S-hydroxydeca-

noyl-CoA (3ZWC), and ECH, complexed with octa-

noyl-CoA (2DUB), deﬁne the importance of loop-1

and loop-2 for the mode of binding of the 3⁰-phos-

phate-adenosine part of the CoA moiety. The adenine

moiety of acyl-CoA is bound in a bulk solvent exposed

binding pocket, shaped by b-strand B1/loop-1 and b-

strand B2/loop-2 (Fig. 6). The loop-1 residues involved

in this binding pocket are Ala57-Leu58-Asn59-Ala60 in

ECH and Pro20-Val21-Asn22-Ala23 in MFE1. The

corresponding residues in ISO-ECI2 are Lys125-

Lys126-Asn127-Ala128. The loop-2 residues of this

binding pocket are Ala96-Gly97-Ala98-Asp99-Ile100 in

ECH, Ala59-Gly60-Ala61-Asp62-Ile63 in MFE1 and

Ser163-Gly164-Asn165-Asp166-Leu167 in ISO-ECI2

(Fig. 3). Each of these loops has highly conserved fea-

tures such as an asparagine in loop-1 (Asn127 in

HsECI2) and an aspartate in loop-2 (Asp166). The side

chains of these conserved residues stabilize the local

conformation by being hydrogen bonded to main chain

atoms of the respective turns. These two conserved

The ISO-ECI2 crystal structure

The ISO-ECI2 crystals were grown at pH 6.5 in the

presence of 1 ^Mammonium sulfate (Table 4). The

ꢀ

crystal structure of ISO-ECI2, reﬁned at 2.6 A resolu-

tion (Table 5), reveals its trimeric structure (Fig. 6A).

Although the crystals were grown in the presence of

0.5 m^Mdecanoyl-CoA, the ligand is not present in any

of the three active sites. Numerous additional co-crys-

tallization and crystal soaking experiments with vari-

ous acyl-CoA ligands did not result in crystals of the

corresponding ligand ISO-ECI2 complexes. The most

ﬂexible region is loop-2. It is best deﬁned and com-

pletely built in subunit A, but poorly ordered in sub-

unit B in which residues Phe170-Thr171-Asp172-Ile173

could not be built. In subunit C also a complete trace

of loop-2 could be modeled adopting a similar confor-

mation as in subunit A. The side chains of Phe170 and

Leu167 of loop-2 point towards the side chain of

Val349 of helix-10. In subunit A, a glycerol molecule

is bound in the active site, being hydrogen bonded to

Table 4. The buffers used for the puriﬁcation and crystallization experiments. The size-exclusion buffer is also the protein storage buffer.

Puriﬁcation buffers

Crystallization buffers

Afﬁnity chromatography

buffer

Well solution

buffer

Variants

Lysis buffer

Size-exclusion buffer

Cryo buffer

ISO-ECI2

50 m^MHEPES pH 7.5 30 m^MHEPES pH 7.5

30 m^MHEPES pH 7.5 100 m^MMES pH 6.5

100 m^MNaCl 1 ^M(NH₄)₂SO₄

100 mM MES pH 6.5

1 ^M(NH₄)₂SO₄

300 m^MNaCl

150 m^MNaCl

30 mM imidazole (wash)/

500 m^Mimidazole (elution)

0.2 mM decanoyl-CoA

20% glycerol

ISOA-ECI2 50 m^MHEPES pH 7.5 30 mM HEPES pH 7.5

30 m^MHEPES pH 7.5 100 mM HEPES pH 7.5 100 mM HEPES pH 7.5

300 mM NaCl

150 mM NaCl

100 mM NaCl

10% PEG 6000

5% MPD

10% PEG 6000

5% MPD

30 m^Mimidazole (wash)/

500 mM imidazole (elution)

0.2 m^Mlauroyl-CoA

20% glycerol

ISOB-ECI2 50 m^MTris/HCl pH 7.5 30 m^MTris/HCl pH 7.5

30 m^MHEPES pH 7.5 80 m^Msodium

80 m^Msodium

cacodylate pH 6.5

14.4% PEG 8000

160 m^Mcalcium

acetate

300 m^MNaCl

1 m^M

300 m^MNaCl

cacodylate pH 6.5

14.4% PEG 8000

160 m^Mcalcium

acetate

1 m^Mb-mercaptoethanol

10% glycerol

b-mercaptoethanol

10% glycerol

30 m^Mimidazole (wash)/

250 m^Mimidazole (elution)

20% glycerol

MPD: 2-Methyl-2, 4-pentanediol.

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

755

Structural enzymological studies of human ECI2

G. U. Onwukwe et al.

Table 5. The crystallographic statistics of data collection and structure reﬁnement. The Ramachandran values are as calculated by

MOLPROBITY [49].

ISO-ECI2

ISOA-ECI2

ISOB-ECI2

Processing software

Scaling and merging

Beamline

XDS

AIMLESS

XDS

ID29 (ESRF)

0.976

ID29 (ESRF)

0.976

I03 (DLS)

0.976

ꢀ

Wavelength (A)

Spacegroup

ꢀ

Unit cell parameters (A)

P2₁2₁2₁

76.43, 91.92, 130.37

3

47.61, 123.69, 128.93

3

74.44, 95.34, 129.97

3

Number of subunits per a.u

a

ꢀ

Resolution range (A)

49.6–2.64 (2.73–2.64)

98.4 (87.3)

11.5 (1.4)

4.4 (53.0)

27 223 (2323)

7.9 (5.8)

2.5

47.61–1.88 (1.95–1.88)

99.5 (96.8)

12.0 (1.5)

5.4 (57.6)

62 659 (5888)

7.2 (7.0)

34.85–2.85 (3–2.85)

99.6 (97.6)

9.9 (2.8)

6.7 (32.7)^b

22 266 (3146)

5.7 (5.5)

2.6

Completeness (%)

〈I/r〉

R_pim(%)

Number of unique reﬂections

Multiplicity

ꢀ

Matthews coefﬁcient (A /Da)

3

2.1

Solvent content (%)

2

50.8

40.8

52.3

ꢀ

Wilson B-factor (A )

82.1

31.5

49.8

Reﬁnement statistics

ꢀ

Resolution (A)

49.6–2.64

19.43

26.22

25 794

1365

47.6–1.88

18.39

23.44

59 362

3172

34.9–2.85

17.63

23.30

21 084

1113

R_work(%)

R_free(%)

Number of reﬂections (R_work

)

Number of reﬂections (R_free

)

Number of atoms

5773

6072

5746

Number of waters

21

273

37

Number of ligand molecules

Glycerol

Cl^ꢀ

1

3

–

3

1

–

3

–

Ethylene glycol

Geometry statistics

ꢀ

rmsd, bonds (A)

0.013

1.5

0.016

1.7

0.011

1.4

rmsd, angles (°)

2

ꢀ

Average B-factor (A )

Protein atoms

Waters

80.4

59.0

93.8

102.9

–

38.5

41.3

–

52.1

36.1

–

Glycerol

Cl^ꢀ

81.4

48.0

82.7

–

Ethylene glycol

Ramachandran plot (%)

Favored

95.0

99.0

1.0

96.6

99.6

0.4

95.3

99.2

0.8

Allowed

Outliers

PDB ID

4U18

4U19

4U1A

a

b

Values in parentheses are for the high resolution shells. Calculated with AIMLESS [44].

turns are hydrogen bonded to each other via O(Gly60,

loop-2) to N(Val24, loop-1) in MFE1 (3ZWC) and via

O(Gly164, loop-2) to N(Ile129, loop-1) in HsECI2

(Fig. 6). In MFE1, the peptide NH group of the O

(Gly60)–N(Ala61) peptide bond is one of the OAH

hydrogen bond donors. The adenine is wedged between

Ala60 and Ile100 in ECH and Ala23 and Ile63 in

MFE1 (Fig. 6B). The corresponding residues in

ISO-ECI2 are Ala128 and Leu167. The adenine-NH2

moiety is hydrogen bonded to O(Ala96) and O(Ala98)

of loop-2 in ECH. These residues are Ala61 and Ala63

in MFE1 and Ser163 and Asn165 in HsECI2 (Fig. 6).

The catalytic site, near the thioester moiety, is deeply

buried at the N-terminal end of helix-3, as known from

the mode of ligand binding to rat ECH [17] and MFE1

[16] (Fig. 2), having the fatty acyl tail buried in a

tunnel lined by loop-2 residues and extending from the

catalytic site to the bulk solvent.

756

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

G. U. Onwukwe et al.

Structural enzymological studies of human ECI2

A

B

C

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

757

Structural enzymological studies of human ECI2

G. U. Onwukwe et al.

binding pocket for the tail will be shaped by loop-2.

Indeed, in the ISO-ECI2 crystal structure a tunnel is

present between the catalytic site and the bulk solvent,

in each of the three subunits, allowing for fatty acyl

tails of different chain lengths to bind (Fig. 6). This is

in good agreement with the kinetic data for rat ECI2

[6] and the calorimetric data for HsECI2 (Table 2),

showing that this ECI2 can use as substrate fatty acyl-

CoAs with a broad range of fatty acyl chain lengths.

Also in ECI1 the fatty acyl tail is bound in a tunnel,

but in this subfamily the tunnel is shaped by loop-4

(Fig. 2). Figure 6 visualizes the ISO-ECI2 tunnel

shaped by loop-2, in particular the side chains of resi-

dues Asn165, Asn169, Asn184, Leu188 and Leu189.

Also contributing to this tunnel are the side chains of

residues Thr131 and Tyr134 of loop-1, Ile214 of helix-

3 and Trp340 of helix-9. Hydrophobic side chains such

as Leu188 and Leu189 as well as polar side chains

such as Tyr134 and Asn165 contribute to the tunnel.

Also a water and a chloride ion are bound inside the

tunnel (Fig. 6). The water is hydrogen bonded to

Glu245 and the chloride ion contacts at a distance of

The active site geometry of ISO-ECI2

Glu245, the catalytic base, is well deﬁned by its elec-

tron density. It adopts an unusual side chain confor-

mation and it is hydrogen bonded to its own main

chain, N(Glu245), and to OG1(Thr217) via its OE1

(Glu245) carboxylate oxygen (Fig. 6), whereas OE2

(Glu245) points to the catalytic cavity and is

surrounded by the side chains of four hydrophobic

residues: Leu189 (loop-2), Ile214 (helix-3), Phe237

(loop-4) and Leu335 (helix-9). Also in yeast ECI2 the

catalytic glutamate Glu158 is hydrogen bonded to a

side chain, Asn101 [11]. In the structure of ECI1 the

ﬂap region of loop-4 (residues Glu245-Gly246-Gly247-

Cys248 in HsECI2) adopts a different conformation

and in this subfamily of ECIs (Fig. 3B) the catalytic

base Glu136 protrudes into the catalytic site out of a

different position of loop-4 (Fig. 3) [8,9]. This position

is the same as that of the catalytic glutamate Glu164

in the ECH structure (Fig. 3A), which is also proposed

to exchange a proton with the C2 atom [31]. In the

latter structure there are actually two glutamates. The

second glutamate (Glu144) protrudes out of helix-3

and these two glutamates (Glu144 and Glu164) are

both hydrogen bonded to the same catalytic water,

which thereby becomes activated for nucleophilic

attack to the double bond of the enoyl-CoA substrate,

to achieve the chirally speciﬁc hydration of the double

bond [32]. These two glutamates, protruding out of

helix-3 and loop-4, are also present in MFE1, being

Glu103 and Glu123 [16]. MFE1 is an efﬁcient hydra-

tase, which appears to be its main physiological func-

tion. The proposed MFE1 enoyl-CoA isomerase

reaction mechanism, dependent on the catalytic water

[16], is different from the reaction mechanisms of

ECI1 [9] and ECI2 [18].

ꢀ

about 5 A this water and the side chains of Tyr134,

Asn165, Asn185, Leu188, Leu189 and Trp340. Chlo-

ride ions were present in the crystallization buffer at a

concentration of 100 m^M. The rim of the tunnel is

shaped by the side chains of Thr131, Asn169, Lys181,

Asn185 and Leu188. The length of the tunnel

measured from OE2(Glu245) to OG1(Thr131) is about

ꢀ

14 A, which corresponds approximately to the length

of an extended acyl tail of 12 carbon atoms.

Helix-10

The active site is covered by helix-10, which is the

C-terminal helix (Fig. 2). The function of helix-10 in

the crotonase superfamily has not yet been studied. It

A comparison of the structures of ISO-ECI2 with

MFE1 (3ZWC) and ECH (2DUB) suggests that the

Fig. 6. (A) The stereo view of the ISO-ECI2 trimer. The C-terminal helix-10 is in dark blue and the loop-2 is in green. The N terminus and

the C terminus of each of the subunits are labeled with N and C, respectively. In stick model is shown the glycerol as bound to the active

site of the A subunit. Chloride ions in each subunit are shown in yellow. (B) The structural comparison of the loops involved in binding and

catalysis of ISO-ECI2 (brown) and the hydratase domain of MFE1 (3ZWC, light blue) (similar view as in Fig. 2, rotated somewhat about the

horizontal direction). In ISO-ECI2 it concerns loop-1 (Lys125-Ile129), loop-2 (Ser163-Leu167), loop-3 (Pro210-Thr217) and loop-4 (Phe233-

Ser248). In the ISO-ECI2 structure the active site glycerol is also shown, as a stick model. The mode of binding of the 3S-hydroxydecanoyl-

CoA to the MFE1 hydratase active site shows that the adenine ring is wedged between Ala23 (loop-1) and Ile63 (loop-2). Black dotted lines

mark the hydrogen bonds between the adenine amino group and O(Ala59) and O(Ala61) (loop-2). Also highlighted by black dotted lines are

the hydrogen bonds between the thioester oxygen atom of 3S-hydroxydecanoyl-CoA and N(Ala61) (loop-2) and N(Gly100) (loop-3) and the

hydrogen bond between loop-1 and loop-2 (Val24-Gly60) in the MFE1 complex. The hydrogen bonds anchoring the side chain of the catalytic

Glu245 of ISO-ECI2 are shown by red dotted lines. In red is also shown Lys126 of loop-1 of ISO-ECI2 predicted to be important for

substrate binding. (C) The tunnel of ISO-ECI2 which is predicted to bind the acyl moiety (subunit A, similar view to Fig. 2, stereo). The

superimposed 3S-hydroxydecanoyl-CoA product of MFE1 hydratase (3ZWC) is shown as stick representation. The tunnel is formed by loop-

2, loop-1, helix-3 and helix-9. The rim of the tunnel is formed by the side chains Thr131 (loop-1), Asn169 (loop-2), Lys181 (loop-2), Asn185

(loop-2), Leu188 (loop-2) (all shown as black sticks). Also visualized is the chloride ion bound in this tunnel (orange dot).

758

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

G. U. Onwukwe et al.

Structural enzymological studies of human ECI2

A

B

C

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

759

Structural enzymological studies of human ECI2

G. U. Onwukwe et al.

appears to be involved in ligand binding interactions

as at least one basic residue protruding out of helix-10

interacts with the phosphate moieties of the acyl-CoA

substrate in the corresponding structures of these com-

plexes. The presence of basic residues in helix-10 is

seen in each of the sequences listed in Fig. 3. In addi-

tion, from the sequence (Fig. 3) and structural com-

parisons it can be noted that a second conserved

feature of helix-10 concerns the presence of hydropho-

bic residues pointing inwards, towards a hydrophobic

cluster of residues that covers the binding pocket of

the ligand near the thioester moiety. In HsECI2 these

residues are Cys345, Val349 and Phe352 (Figs 3 and

7). Other residues that contribute to this cluster are

protruding out of loop-2 (Leu167 and Phe170) and

loop-4 (Leu240). In Fig. 7 the hydrophobic interac-

tions between helix-10 and loop-2 and loop-4 of ISO-

ECI2 are compared with the corresponding geometry

of the ECH and MFE1 structures. In ECH the corre-

sponding residues are Arg272, Met276 and Phe279 of

helix-10, Ile100 and Met103 of loop-2 and Leu167 of

loop-4, and in MFE1 they are residues Ala264,

Gln268 and Phe271 (helix-10), Ile63 and Phe66 (loop-

2) and Leu126 (loop-4). It can be noted that Ile100 of

ECH also contributes to the binding pocket for the

adenine ring (Fig. 6). Ile100 is in van der Waals con-

tact with the sulfur atom of the thioester moiety, like

Phe66 in MFE1 (Fig. 7). The conserved leucine of

loop-4 also contacts the sulfur of the thioester moiety

in each of the liganded structures at a distance of

the V349A point mutation variant (ISOA-ECI2)

and the helix-10 deletion variant (ISOB-ECI2), in

which the C-terminal residues 350–355 were deleted

(Fig. 3). These variants have somewhat lower stability

than the wild-type ISO-ECI2 trimer (Fig. 4), but crys-

tallization experiments with both variants yielded well-

diffracting crystals.

The afﬁnity of the two variants for acyl-CoA mole-

cules with a saturated chain is somewhat higher com-

pared to wild-type ISO-ECI2. For example, the K_dof

decanoyl-CoA for ISO-ECI2, ISOA-ECI2 and ISOB-

ECI2 is 37, 25 and 10 l^M, respectively, as measured by

calorimetry (Table 2). The enthalpy of binding of these

substrate analogues to ISOA-ECI2 and ISOB-ECI2 is

much higher than for binding to ISO-ECI2. For

the ISOA-ECI2 variant, catalytic activity could be

detected. In the presence of 100 l^M3E-hexenoyl-CoA

and 3E-decenoyl-CoA, the activities are 2.5 and 9.2

lmolꢀmin^ꢁ1ꢀmg^ꢁ1, respectively, whereas for ISO-ECI2

these activities are 4.5 and 18.7 lmolꢀmin^ꢁ1ꢀmg^ꢁ1

.

For the helix-10 deletion mutant ISOB-ECI2, no cata-

lytic activity could be detected with the 3E-hexenoyl-

CoA and 3E-decenoyl-CoA substrates. Clearly, the

C-terminal end of helix-10 is required for efﬁcient

catalysis.

The crystal structures of ISOA-ECI2 and ISOB-

ECI2 were determined (Tables 4 and 5). In each struc-

ture, the asymmetric unit consists of a trimer, as in the

wild-type crystals. Clearly, the mutation does not

affect the trimerization of the subunits.

ꢀ

approximately 4.0 A. In HsECI2 Leu167 of loop-2

In ISOA-ECI2 helix-10 is disordered after the

V349A residue in subunits A and C, but in subunit B

it could be built up to residue Arg355, which is the C-

terminal residue. This is a consequence of crystal pack-

ing, as the C-terminal part of helix-10 is stabilized by

a neighboring molecule. Loop-2 of the ISOA-ECI2 is

only well deﬁned in subunit A, whereas the Phe170-

Thr171-Asp172-Ile173 residues of subunits B and C

were not modeled. In wild-type ISO-ECI2 the Leu167

and Leu240 of loop-4 correspond to these hydropho-

bic residues of MFE1 and ECH.

The crystal structures of ISOA-ECI2 and ISOB-

ECI2

The importance of Val349 and that of helix-10 for

catalysis was investigated by the characterization of

Fig. 7. The hydrophobic cluster which anchors helix-10 to loop-2 and loop-4 of human ISO-ECI2 (light brown, subunit A) aligned with (A) the

hydratase domain of MFE1 with bound 3S-hydroxydecanoyl-CoA (3ZWC, light blue) and (B) ECH with bound octanoyl-CoA (2DUB, light

blue). The following side chains of the ISO-ECI2 crystal structure are shown: Leu167 and Phe170 of loop-2, Leu240 of loop-4, and Cys345

and Val349 of helix-10 (Phe352 of helix-10 is not ordered in this subunit). The glycerol molecule in the active site of ISO-ECI2 and the

chloride ion in the hydrophobic tunnel are also shown as light brown sticks and yellow sphere, respectively. Also shown in each of the

panels is the loop-4 catalytic glutamate, being Glu245 in ECI2-ISO, Glu123 in MFE1 and Glu164 in ECH. In (A) the corresponding

hydrophobic residues of MFE1 are Ile63, Phe66, Leu126, Ala264, Gln268 and Phe271. In (B) the corresponding residues of ECH are Ile100,

Met103, Leu167, Arg272, Met276, Phe279. The van der Waals interactions between the sulfur atom of the acyl-CoA ligand and the side

chains of Phe66, Leu126 (in MFE1) and Ile100, Leu167 (in ECH) are highlighted by black dotted lines (the corresponding residues in ISO-

ECI2 are Leu167, Leu240 respectively). (C) Superposition of ISO-ECI2 (light brown) with ISOA-ECI2 (yellow) and ISOB-ECI2 (cyan),

highlighting the loop-2 changes of Phe170 (in ISOA-ECI2 and ISOB-ECI2) and Leu167 (in ISOB-ECI2). View obtained by rotating around the

horizontal direction with respect to (A) and (B); similar view as in Fig. 2. Key residues of loop-2, loop-4 and helix-10 are shown and labeled.

Also shown is the glycerol molecule bound in the OAH of ISO-ECI2 and the corresponding OAH waters in ISOA-ECI2 and ISOB-ECI2. The

catalytic glutamate (Glu245) and the chloride ion of each of the three structures are shown and labeled.

760

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

G. U. Onwukwe et al.

Structural enzymological studies of human ECI2

and Phe170 side chains of loop-2 are in van der Waals

contact with the Val349 side chain, but in subunit A

of ISOA-ECI2 the Phe170 side chain points towards

the bulk solvent whereas the Leu167 conformation

does not change (Fig. 7). In ISOB-ECI2, helix-10 can

be built up to residue Thr346 in subunit C and up to

residue Cys345 in subunits A and B. Loop-2 adopts a

deﬁned and similar conformation in each of the three

subunits. It is best deﬁned in subunit C. The Phe170

side chain points further inwards and Leu167 points

further outwards, compared to the wild-type enzyme.

Although the mutational differences of ISOA-ECI2

and ISOB-ECI2 concern the helix-10 sequence, there

are no structural changes in this helix. There are also

no structural changes at the catalytic site, near

Glu245, or in the fatty acyl tail binding tunnel, where

the chloride ion binding site is also preserved in ISOA-

ECI2 and ISOB-ECI2 (Fig. 7). There are structural

differences in loop-2, however, in particular in ISOB-

ECI2 (Fig. 7). These structural differences correlate

with differences in catalytic properties, including the

lower speciﬁc activity of ISOA-ECI2 and no catalytic

activity of ISOB-ECI2, although calorimetric data for

both variants show that they still bind the longer satu-

rated substrate analogues tightly (Table 2). In ISOB-

ECI2, loop-2 becomes more rigid in each of the three

subunits and the Phe170 side chain points further

inward, towards the fatty acyl binding tunnel, and it is

in van der Waals contact with the chloride ion that is

bound in this tunnel. In this inward pointing position,

the Phe170 side chain would not block the tunnel,

allowing for binding of saturated substrate analogues

(as detected by calorimetry). In its new position, the

Phe170 side chain has moved into an empty space,

which in ISO-ECI2 connects the catalytic site to bulk

solvent. This somewhat narrow exit tunnel between

catalytic site and bulk solvent, deﬁned by loop-2

(Phe170, Lys181, Asn185), loop-4 (Gln242) and helix-9

(Trp340), is visualized in Fig. 8. The possible func-

tional role of this alternative exit tunnel for binding of

the fatty acyl tails is not clear. In any case, it identiﬁes

an alternative exit path for binding the fatty acyl tail

(between helix-9 and helix-2), which is different from

the exit tunnel in ECI1 (between loop-4 and helix-9;

Fig. 2) and from the MFE1 exit path (between helix-2

and loop-1/helix-1) (Figs 2 and 6). In ECH the exit

path for the fatty acyl tail is made by unfolding of

helix-2 [17]. Possibly more important for catalysis is

the movement of Leu167 of the hydrophobic cluster

away from the thioester moiety. This conserved hydro-

phobic cluster and the catalytic base are on opposite

sides of the thioester moiety: the cluster effectively

shields the reactive thioester moiety from bulk solvent

(Figs 7 and 8). In the mutated variants the positions

of Leu167 (in ISOB-ECI2) and Phe170 (in ISOA-ECI2

and ISOB-ECI2) change (Fig. 8). In particular, in

ISOB-ECI2, being inactive, these structural changes

possibly affect the correct positioning of the thioester

moiety in its binding pocket. These results therefore

suggest that the function of this hydrophobic cluster

could be to ﬁx the reactive thioester moiety rigidly in

its binding pocket. This tight binding then ensures that

the thioester oxygen binds correctly in its OAH allow-

ing for optimal stabilization of the enolate intermedi-

ate when it is formed in the catalytic cycle, facilitating

efﬁcient catalysis.

Concluding remarks

Our SAXS data indicate that the ACBP domains are

placed at the periphery of the isomerase trimer,

extending the ﬂat shape of the trimer (Fig. 5). Hence,

the full-length isomerase is a highly ﬂattened trimer,

with all domains in the same plane. The structure

directly suggests that there is no cross-talk between the

ACBP domains from different chains. The ACBP

domains have high afﬁnity towards saturated acyl-CoA

molecules (which are substrate analogues) for both

short- and long-chain acyl moieties, whereas for the

isomerase domain the afﬁnity depends much more on

the length of the acyl moiety, being signiﬁcantly lower

for short-chain acyl-CoA ligands (Table 2). For short/

medium-chain substrates, like 3E-hexenoyl-CoA and

3E-decenoyl-CoA, the k_cat/K_mvalues of FL-ECI2 are

higher than those for ISO-ECI2 (Table 1), suggesting

that the ACBP domain is important for efﬁcient metab-

olism of short/medium-chain 3E-enoyl-CoA substrates.

The HsECI2 crystal structures suggest that the fatty acyl

tails of the enoyl-CoA substrates bind in a tunnel

accommodating medium-chain as well as long-chain

fatty acyl tails. This tunnel extends from the catalytic

site to bulk solvent and is mainly shaped by residues of

loop-2 and loop-1. Structural comparisons show that

the OAH geometry is a strictly conserved feature of the

catalytic site of this group of enzymes, whereas the cata-

lytic glutamate can protrude into the catalytic cavity

from different positions in loop-4. Also the fatty acyl

tail binding tunnel is shaped in various ways by the

loops following after the b-strands of sheet B. In several

structures of the subfamily of hydratases and isomerases

of the crotonase superfamily, the C-terminal part of

helix-10 is disordered, in particular in unliganded struc-

tures, but our mutagenesis studies clearly show the

importance of helix-10 for catalysis. These studies high-

light also the importance of loop-2. It remains intriguing

that ECI2 can isomerize both 3Z- and 3E-enoyl-CoA

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

761

Structural enzymological studies of human ECI2

G. U. Onwukwe et al.

A

B

C

Fig. 8. Surface representation of the crystal structures of (A) ISO-ECI2 subunit A, (B) ISOA-ECI2 subunit A and (C) ISOB-ECI2 subunit C

showing the narrow exit tunnel from the catalytic site. The view is similar to Fig. 7. The residues lining this tunnel in ISO-ECI2, Phe170,

Lys181, Asn185 (of loop-2), Gln242 (loop-4) and Trp340 (loop-9), are labeled in (A). Also labeled in each of the panels are Leu167 and

Phe170 of loop-2. Catalytic glutamate (not labeled) is faintly visible behind Gln242. Each structure has the chloride ion bound to the wider

exit tunnel (not visible in the ISOB-ECI2 structure in C). A 3S-hydroxyacyl-CoA molecule, as bound to the MFE1 active site, is shown in each

panel. The tail of this molecule ﬁts well into the wider exit tunnel of HsECI2.

762

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

G. U. Onwukwe et al.

Structural enzymological studies of human ECI2

substrates. The presence of the main tunnel (deﬁned by

loop-2 and loop-1) and the narrower exit tunnel (deﬁned

by loop-2 and loop-4) in ISO-ECI2 might be relevant

for its substrate speciﬁcity. Further crystallographic

binding studies with ISO-ECI2, aimed at a better under-

standing of the role of loop-2 for substrate speciﬁcity,

have been initiated.

Protein puriﬁcation

The same general puriﬁcation strategy was adopted for all

constructs. The compositions of the puriﬁcation buffers

used for the crystallization studies are shown in Table 4.

The cell pellet in lysis buffer was thawed and more lysis

buffer was added to give a ﬁnal volume of 40 mL per 1 L

original culture. Lysis was enhanced by adding DNase1,

RNaseA, lysozyme and MgSO₄to the thawed cell suspen-

sion to a ﬁnal concentration of 20 lgꢀmL^ꢁ1, 2 lgꢀmL^ꢁ1

,

Materials and methods

100 lgꢀmL^ꢁ1and 1 mM, respectively. The mixture was

incubated at room temperature (40 min) under constant

stirring and subsequently the soluble fraction was harvested

by centrifugation (35 000 g, 1 h). The clear supernatant

was then passed through a Talon column equilibrated with

the lysis buffer. The recombinant enzymes bound to the

column were eluted with 250–500 m^Mimidazole (Table 4).

The fractions containing the eluted protein were concen-

trated using either a 3 kDa (for ACBP-ECI2) or a 10 kDa

cut-off centrifugal ﬁlter (for the FL-ECI2, ISO-ECI2,

ISOA-ECI2, ISOB-ECI2). In the next step, the enzymes

were passed through the Hiload 16/60 Superdex 200 Prep

Site-directed mutagenesis and protein

expression

For the expression of the FL-ECI2, ISO-ECI2 and ACBP-

ECI2 constructs, the plasmids Pet28a-FL-ECI2 (residues

Met1-Leu359), PNIC28-Bsa4-ISO-ECI2 (Gly103-Arg355)

and Pet28a-ACBP-ECI2 (Met1-Ser89) were used, respec-

tively. The FL-ECI2 and ACBP-ECI2 plasmids (provided

by J. Knudsen) contained an N-terminal puriﬁcation tag

with a thrombin cleavage site (MGSSHHHHHHSSGLV

PRGSHMASSQST)) before Met1 (Fig. 3), while the ISO-

ECI2 plasmid (obtained from the Structural Genomics

Consortium (SGC), Oxford, UK) contained an N-terminal

€

Grade gel ﬁltration column using an Akta Puriﬁer puriﬁca-

tion unit (GE Healthcare, Uppsalla, Sweden). The fractions

containing the recombinant protein were pooled, concen-

trated and stored in small quantities at ꢁ70 °C in the gel

ﬁltration buffer (Table 4). The enzyme concentration was

determined by UV absorption, using extinction coefﬁcients

(His)₆-tag followed by

a TEV protease cleavage site

(MHHHHHHSSGVDLGTENLYFQS). The ISO-ECI2

plasmid was also used as a template for mutagenesis to pre-

pare the helix-10 variants ISOA-ECI2 and ISOB-ECI2.

Mutagenesis was performed with the QuikChange site-

directed mutagenesis kit (Agilent Technologies, Santa

Clara, CA, USA) using the forward primers 5⁰-CAGA

TGAATGCACAAATGCTGCAGTGAACTTCTTATCC-3⁰

(ISOA-ECI2) and 5⁰-CAGATGAATGCACAAATGCT

GTGTGACAGTAAAGGTGGATACGG-3⁰(ISOB-ECI2)

and the reverse primers 5⁰-GGATAAGAAGTTCACTGC

AGCATTTGTGCATTCATCTG-3⁰(ISOA-ECI2) and 5⁰-

CCGTATCCACCTTTACTGTCACACAGCATTTGTGCA

TTCATCTG-3⁰(ISOB-ECI2) (the mutation sites are shown

in bold).

ꢁ1

of 34 295, 19 940 and 15 470 ^Mꢀcm^ꢁ1for FL-ECI2, ISO-

ECI2 (and helix-10 variants) and ACBP-ECI2, respectively.

The storage buffer for the FL-ECI2, ISO-ECI2 and ACBP-

ECI2 used for isothermal titration calorimetry (ITC) and

kinetic experiments was 30 m^MTris/HCl, pH 8.5.

Static light scattering (SLS)

The oligomeric state of FL-ECI2 and ISO-ECI2 in solution

was determined using SLS in the presence of 30 m^Mpotas-

sium phosphate buffer, pH 7.5. Approximately 50 lL of

puriﬁed protein (4 mgꢀmL^ꢁ1) was loaded onto a Super-

The FL-ECI2 and ACBP-ECI2 plasmids were trans-

formed into BL21(DE3) cells, whereas the ISO-ECI2, ISOA-

ECI2 and ISOB-ECI2 plasmids were transformed into BL21

(DE3)pLysS competent cells. The expression of recombinant

proteins was performed in 1 L M9ZB medium (1% N-Z case

plus^ꢀcasein hydrolysate, 85 mM NaCl, 20 mM NH₄Cl,

20 m^MNaH₂PO₄, 1 mM MgSO₄, 0.4% glucose) containing

30 lgꢀmL^ꢁ1kanamycin and 34 lgꢀmL^ꢁ1chloramphenicol.

The culture was grown at 37 °C at 200 r.p.m. until an A 600

of 0.6 was reached, and protein expression was initiated with

IPTG using a ﬁnal concentration of 0.5 m^M(for ISOA-ECI2

and ISOB-ECI2) or 1 mM (for FL-ECI2, ISO-ECI2 and

ACBP-ECI2). After an overnight expression at 18 °C, the

cells were collected using centrifugation and resuspended in

the lysis buffer (Table 4). At the end, the cell suspensions

were stored at ꢁ70 °C until further use.

€

dex200 10/300 GL gel ﬁltration column using an Akta Puri-

ﬁer (GE Healthcare) and subsequently passed through a

Shodex refractive index detector (for measuring the protein

concentration) and a miniDAWN TREOS SLS instrument

(Wyatt, Santa Barbara, CA, USA). The results were ana-

lyzed with ^ASTRA5.3 (Wyatt).

Circular dichroism (CD) and thermal denaturation

For CD measurements the protein was diluted to

30 lgꢀmL^ꢁ1using 30 mM potassium phosphate, pH 7.5.

The wavelength range used was 190–260 nm at a tempera-

ture of 20 °C. The cuvette path length was 0.1 cm. Thermal

denaturation was carried out from 20 to 90 °C using a tem-

perature ramp of 1 °Cꢀmin^ꢁ1. The instrument used was the

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

763

Structural enzymological studies of human ECI2

G. U. Onwukwe et al.

Applied Photophysics Chirascan^TM, Surrey, UK. For all

measurements, buffer spectra were subtracted to give the

enzyme spectrum. For data analysis the wavelength range

used was 200–240 nm. T_mvalues were calculated with GLO-

^BAL3 (Applied Photophysics) by curve ﬁtting calculations

using the spectral changes as a function of temperature. A

two-transition unfolding model for ECI2-FL was used (pre-

dicting the T_mvalue of the intermediate to be

54.3 ꢀ 1.4 °C), whereas for ISO-ECI2, ACBP-ECI2,

ISOA-ECI2 and ISOB-ECI2 a one-transition unfolding

model was used.

at 25 °C for 8–16 min using the ThermoVac instrument

(GE Healthcare). ITC experiments were performed with a

MicroCal ITC200 calorimeter (GE Healthcare). Tempera-

ture equilibration was carried out for 15–20 min before each

experiment. The sample cell was ﬁlled with 200 lL of the

enzyme (40–180 l^Mas calculated for the subunit or mono-

mer), the protein concentrations being dependent on the pro-

tein construct. A total of 38 lL of ligand solution (1–

3.9 m^M) using 1 lL per injection into the sample cell was

used. The time between each injection was 120 s, with an ini-

tial delay of 60 s before the ﬁrst injection. For titration of

ISO-ECI2 with octanoyl-CoA, two data ﬁles were combined

using the ^CONCATsoftware (MicroCal) since saturation could

not be reached with one measurement. All experiments were

carried out at 25 °C under constant stirring at 500 r.p.m.

Data analysis and curve ﬁtting were carried out using ^ORIGIN

7.0 (MicroCal).

Substrate synthesis

Acyl-CoA thioesters were prepared by the mixed anhydride

method [33] and puriﬁed on reverse phase silica gel (RP-18;

ICN Chemicals, Costa Mesa, CA, USA). The substrates

were veriﬁed by determining their masses using an ESI-

FTICR mass spectrometer (APEX II; Bruker, Bremen,

Germany).

Small angle X-ray scattering

Synchrotron SAXS data were collected on beamline I911-4

at MAX-Lab (Lund, Sweden) and beamline P12 at

PETRA III, EMBL/DESY (Hamburg, Germany). For FL-

ECI2 (2.2 mgꢂmL^ꢁ1), the SAXS data were measured in the

presence and absence of acyl-CoA ligands, including octa-

noyl-, decanoyl- and lauroyl-CoA at 50–200 l^M. SAXS

measurements were also carried out for ISO-ECI2

Enzyme activity measurements

Enzyme activities were determined by monitoring the for-

mation of NADH spectrophotometrically at 340 nm in a

coupled assay. The measurements were done in 50 m^M

Tris/HCl, pH 8.5, and the ﬁnal assay mixture (0.5 mL)

was composed of 1 m^MNAD⁺, 25 ng (FL-ECI2) or

50 ng (ISO-ECI2 and its variants) of enzyme and 5–

300 l^Msubstrate. In addition, 1.6 lg of Drosophila mela-

nogaster multifunctional enzyme type 2 (DmMFE2) was

used in the assays as a linker enzyme. This linker enzyme

was puriﬁed as previously described [34] with little modiﬁ-

cation. The substrate concentration was determined by the

Ellman test [35]. The measurements were carried out with

the Jasco V-660 spectrophotometer (Jasco Inc., Easton,

PA, USA). In all experiments the reaction components

were allowed to equilibrate at room temperature for about

10 min before starting the reaction by adding the sub-

strate. NADH formation was monitored for 3 min by

observing the change in absorbance at 340 nm. The

(4.1 mgꢂmL^ꢁ1

)

and ACBP-ECI2 (5.7 mgꢂmL^ꢁ1

) in the

absence of ligands. Exposure times were 2 min at I911-4

ꢀ

(k = 0.91 A) and 20 9 50 ms at P12 (k = 1.24 A). The

corresponding buffers (FL-ECI2, 30 m^MTris, pH 8.5; ISO-

ECI2 and ACBP-ECI2, 30 m^MHEPES, pH 7.5, 100 mM

NaCl) were similarly measured, and buffer contribution

was subtracted from the protein data. For each sample a

50% dilution was also measured, and for all modeling runs

data merged from the two concentrations were used. The

exception was the full-length protein, which always aggre-

gated in the X-ray beam to some extent when it had been

diluted. Hence, only the highest concentration was used

for analyses of FL-ECI2, and no aggregation was evident

in linear Guinier plots.

ꢁ1

extinction coefﬁcient of 6220 ^Mꢂcm^ꢁ1was used for the

Programs of the ^ATSASpackage [36] were used for data

processing and analysis. Distance distributions were calcu-

lated using ^GNOM[37], and ab initio models were built using

DAMMIF [38] and GASBOR [39]. Hybrid modeling employing

rigid body reﬁnement and loop building for FL-ECI2 was

done with ^CORAL[36] and BUNCH [40], respectively. The

model was created using 3-fold symmetry and the crystal

structure of ISO-ECI2 (2F6Q; SGC, Oxford, UK), repre-

sented the isomerase domain (corresponding to residues

Thr106-Asn347) and the NMR structure of the HsECI2

ACBP domain (2CQU, corresponding residues Ser4-Ser84).

Structure superpositions were done using ^SUPCOMB[41].

Theoretical SAXS data for crystal structures were calcu-

lated with ^CRYSOL[42].

conversion calculations. All experiments were performed

at 25 °C and were done at least in duplicate. The data

were ﬁtted to the Michaelis–Menten equation using ^GRAPH-

PAD PRISM software (GraphPad Software Inc., La Jolla,

CA, USA).

Isothermal titration calorimetry (ITC)

ITC experiments were done with protein solutions in 30 mM

Tris/HCl, pH 8.5. The saturated fatty acyl-CoA powder

(Sigma, Helsinki, Finland) was dissolved in the same buffer.

Ligand concentrations were determined using Ellman’s test.

Before each experiment the protein samples were degassed

764

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

G. U. Onwukwe et al.

Structural enzymological studies of human ECI2

A complete trace was built for subunit A of the ISO-

ECI2 structure (from residue Gly103; also present is

Met102 plus seven additional N-terminal tag residues); for

this subunit loop-2 is deﬁned and it has the same confor-

mation as in subunit C. For subunit B the loop-2 resi-

dues Phe170-Val178 of loop-2 could not be built. The C-

terminal residues are Val350 for each of the subunits. For

ISOA-ECI2, the trace of subunit B starts at Met102, the

C-terminal helix-10 is ordered and the V349A mutation is

clearly visible in the map. However, for this subunit loop-

Crystallization and data collection

Crystallization was performed using 96-well plates and the

sitting-drop vapor diffusion method. Crystallization drops

were prepared by the Mosquito crystallization robot (TTP

Labtec, Melbourn, UK) by mixing 0.5 lL of the protein

solution with 0.5 lL well solution (Table 4). The plates

were incubated at room temperature and imaged using the

Formulatrix RI54 plate hotel. ISO-ECI2 (5.2 mgꢀmL^ꢁ1

,

supplemented with 0.5 m^Mdecanoyl-CoA in the protein

solution) crystallized in 100 mM MES pH 6.5, 1 M

(NH₄)₂SO₄, and the crystals grew to their maximum size in

6–10 days. ISOA-ECI2 (4.1 mgꢀmL^ꢁ1, protein solution sup-

plemented with 0.5 mM lauroyl-CoA) and ISOB-ECI2

(3.5 mgꢀmL^ꢁ1) were crystallized and cryo-cooled in differ-

ent conditions but using the same protocol as described for

ISO-ECI2 (Table 4). Before data collection, the crystals

were brieﬂy passed through the crystallization mother

liquor, supplemented with 20% glycerol and a ligand

(Table 4), and subsequently cryo-cooled in liquid nitrogen.

Data collection was carried out on the ID29 beam line at

ESRF (Grenoble, France) and on the beam line I03 of the

Diamond Light Source (DLS), Oxford, UK (Table 5). Data

processing was carried out using ^XDS[43]. Scaling was car-

ried out using ^XSCALE[43] or AIMLESS [44] of the CCP4 suite

[45]. The data processing statistics are listed in Table 5.

2

residues Phe170-Thr171-Asp172-Ile173 could not be

built due to its intrinsic disorder; these residues are only

well deﬁned in subunit A. The C-terminal residues are

Ala349, Arg355 and Ala348 for, respectively, subunits A,

B and C. For the ISOB-ECI2 structure a complete trace

could be built for subunit C from Gly103 to Thr346; also

Met102 was built plus ﬁve additional N-terminal tag resi-

dues. For ISOB-ECI2 loop-2 could be built for each of

the three subunits, having the same conformation; it is

best deﬁned in subunit C. The C-terminal residues are

Cys345, Cys345 and Thr346 for respectively subunits A,

B and C.

The crystal structure ISO-ECI2 (2F6Q) which has been

used for the molecular replacement calculation is an unli-

ganded complex of the SeMet derivative of the ISO-ECI2

construct of HsECI2, crystallized in 0.1 ^MTris, pH 8.5, in

the presence of 20% isopropanol. In this crystal structure

loop-2 is also a high B-factor loop except in subunit B for

which the loop-2 conformation was built in a similar con-

formation as in the ISO-ECI2 subunit A structure. Crystal

structures of rat peroxisomal MFE1 (3ZWC) [16], rat mito-

chondrial ECH (2DUB) [17], mitochondrial ECI1 of

human (1SG4) [9] and rat (1XX4) [8], rat mitochondrial

DECI (1DCI) [50] and yeast (S. cerevisiae) peroxisomal

ECI2 (1PJH) [11] were also used in the comparison studies.

The structures of ACBP used for structural analysis were

the NMR structures of the bovine ACBP (1ACA) [30] com-

plexed with palmitoyl-CoA and the human ACBP-ECI2

domain (2CQU) as well as the crystal structures of human

(2CB8, complexed with myristoyl-CoA, and 2FJ9, unli-

ganded) [22] and Plasmodium falciparum ACBP, complexed

with myristoyl-CoA (1HBK) [29]. The sequence alignment

of human ECI2 with homologous enzymes was performed

using ^CLUSTALW2 [51] on the EBI server (http://www.ebi.

ac.uk/Tools/msa/clustalw2/). The sequence alignment was

improved by structure superpositions carried out using the

SSM [52] tool in COOT [48]. The secondary structure assign-

ment was carried out using ^DSSP[53] from the CMBI server

(http://www.cmbi.ru.nl/dssp.html). Evolutionary history

was inferred using the neighbor-joining method, using

MEGA5 [54]. The evolutionary distances were computed

using the Poisson correction method and are in units of

number of amino acid substitutions per site. The ﬁgures of

the structures were prepared using ^PYMOL(http://www.py

mol.org).

Structure determination, model building and

reﬁnement

The determination of the three structures was done by

molecular replacement using ^PHASER[46]. The structure of

the ISO-ECI2 construct (2F6Q), without ligands, was used

as the search model for all three structure determinations.

There are three subunits per asymmetric unit in each struc-

ture. Reﬁnement was carried out using ^REFMAC5 [47] and

iterative cycles of model building were done using ^COOT

[48]. The REFMAC5 reﬁnement was carried out using non-

crystallographic symmetry restraints. The reﬁnement was

ﬁnalized when no signiﬁcant peaks were found in the

F_oꢁ F_cmap. The models obtained have good reﬁnement

statistics and geometry (Table 5).

Structure and sequence analysis

There is good electron density for most of the residues,

except for some residues of loop-2 which have high B-fac-

tors in most of the chains, in particular residues Phe170-

Thr171-Asp172-Ile173. Also the residues at the N terminus

and C terminus in most chains are disordered. The Raman-

chandran plot calculated by ^MOLPROBITY[49] shows few

outliers in the ﬂexible loops. The only outlier, which is built

in well deﬁned density, is Ser257, adopting φ/w values of

approximately 75°/125° in each of the three chains of each

of the models.

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

765

Structural enzymological studies of human ECI2

G. U. Onwukwe et al.

Mitochondrial 2,4-dienoyl-CoA reductase deﬁciency in

mice results in severe hypoglycemia with stress

intolerance and unimpaired ketogenesis. PLoS Genet 5,

e1000543.

Acknowledgements

We thank Dr Jens Knudsen for the plasmids of FL-

ECI2 and ACBP-ECI2. The plasmid of ISO-ECI2 was

obtained from the SGC (Oxford, UK). The research

leading to these results has received funding from the

European Community’s Seventh Framework Pro-

gramme (FP7/2007-2013) under BioStruct-X (grant

agreement no. 283570) and from the Academy of Fin-

land (243008921). We thank Ville Ratas and Tiila

Kiema for skillfully maintaining our BCO protein

crystallography facility and Piotr Prus for help with

the CD and ITC experiments. We gratefully acknowl-

edge the expert support from the beam line scientists

of the SAXS beam lines I911-4 (MAXlab, Lund, Swe-

den) and P12 (DESY, EMBL Hamburg, Germany)

and the MX beam lines I03 (DLS, Oxford, UK) and

ID29 (ESRF, Grenoble, France). We thank Petri

Pihko for stimulating discussions on the crotonase cat-

alytic properties, Tuomo Glumoff for providing the

DmMFE2 plasmid, and Natalia Markova and Prasad

Kasaragod for expert discussions on the ITC and

enzyme kinetics data, respectively.

4 Janssen U & Stoffel W (2002) Disruption of

mitochondrial b-oxidation of unsaturated fatty acids in

the 3,2-trans-enoyl-CoA isomerase-deﬁcient mouse.

J Biol Chem 277, 19579–19584.

5 Gurvitz A, Mursula AM, Firzinger A, Hamilton B,

€

Kilpelainen SH, Hartig A, Ruis H, Hiltunen JK &

Rottensteiner H (1998) Peroxisomal D³-cis-D²-trans-

enoyl-CoA isomerase encoded by ECI1 is required for

growth of the yeast Saccharomyces cerevisiae on

unsaturated fatty acids. J Biol Chem 273,

31366–31374.

6 Zhang D, Yu W, Geisbrecht BV, Gould SJ, Sprecher H

& Schulz H (2002) Functional characterization of D³,

D²-enoyl-CoA isomerases from rat liver. J Biol Chem

277, 9127–9132.

7 Van Weeghel M, Ofman R, Argmann CA, Ruiter JP,

Claessen N, Oussoren SV, Wanders RJ, Aten J &

Houten SM (2014) Identiﬁcation and characterization

of Eci3, a murine kidney-speciﬁc D3, D2 -enoyl-CoA

isomerase. FASEB J 28, 1365–1374.

8 Hubbard PA, Yu W, Schulz H & Kim JJ (2005)

Domain swapping in the low-similarity isomerase/

hydratase superfamily: the crystal structure of rat

mitochondrial D³, D²-enoyl-CoA isomerase. Protein Sci

14, 1545–1555.

Author contributions

GUO performed the mutation, overexpression, puriﬁ-

cation, enzyme kinetics, calorimetric studies, crystalli-

zation, X-ray data collection, structure determination

and SAXS sample preparation. MKK has been

involved in crystal handling, data collection, data pro-

cessing, structure analysis as well as with preparing the

manuscript. RKW conceived and supervised the pro-

ject while GUO contributed to the experimental

design. PK performed the SAXS data collection and

analysis. WS synthesized the substrates. GUO, MKK

and RKW analyzed the structures and wrote the

paper. All authors ﬁnalized the paper.

9 Partanen ST, Novikov DK, Popov AN, Mursula AM,

ꢀ

Hiltunen JK & Wierenga RK (2004) The 1.3 A crystal

structure of human mitochondrial D³, D²-enoyl-CoA

isomerase shows a novel mode of binding for the fatty

acyl group. J Mol Biol 342, 1197–1208.

10 Kasaragod P, Venkatesan R, Kiema TR, Hiltunen JK

& Wierenga RK (2010) Crystal structure of liganded rat

peroxisomal multifunctional enzyme type 1: a ﬂexible

molecule with two interconnected active sites. J Biol

Chem 285, 24089–24098.

11 Mursula AM, Hiltunen JK & Wierenga RK (2004)

Structural studies on D³, D²-enoyl-CoA isomerase: the

variable mode of assembly of the trimeric disks of the

crotonase superfamily. FEBS Lett 557, 81–87.

12 Holden HM, Benning MM, Haller T & Gerlt JA

(2001) The crotonase superfamily: divergently related

enzymes that catalyze different reactions involving

acyl coenzyme a thioesters. Acc Chem Res 34,

145–157.

13 Hamed RB, Batchelar ET, Clifton IJ & Schoﬁeld CJ

(2008) Mechanisms and structures of crotonase

superfamily enzymes-how nature controls enolate and

oxyanion reactivity. Cell Mol Life Sci 65, 2507–2527.

14 Engel CK, Mathieu M, Zeelen JP, Hiltunen JK &

Wierenga RK (1996) Crystal structure of enoyl-

References

1 Van Weeghel M, Te Brinke H, van Lenthe H, Kulik

W, Minkler PE, Stoll MS, Sass JO, Janssen U, Stoffel

W, Schwab KO et al. (2012) Functional redundancy

of mitochondrial enoyl-CoA isomerases in the

oxidation of unsaturated fatty acids. FASEB J 26,

4316–4326.

2 Hiltunen JK & Qin Y (2000) b-oxidation strategies for

the metabolism of a wide variety of acyl-CoA esters.

Biochim Biophys Acta 1484, 117–128.

3 Miinalainen IJ, Schmitz W, Huotari A, Autio KJ,

Soininen R, Ver Loren van Themaat E, Baes M, Herzig

KH, Conzelmann E & Hiltunen JK (2009)

ꢀ

coenzyme A (CoA) hydratase at 2.5 A resolution: a

766

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

G. U. Onwukwe et al.

Structural enzymological studies of human ECI2

spiral fold deﬁnes the CoA-binding pocket. EMBO J

15, 5135–5145.

15 Benning MM, Taylor KL, Liu R-Q, Yang G, Xiang H,

Wesenberg G, Dunaway-Mariano D & Holden HM

(1996) Structure of 4-chlorobenzoyl coenzyme A

elegans MFE-2 in solution from synchrotron small-

angle X-ray scattering. FEBS Lett 587, 305–310.

26 Hamed RB, Henry L, Gomez-Castellanos JR,

Mecinovic J, Ducho C, Sorensen JL, Claridge TD &

Schoﬁeld CJ (2012) Crotonase catalysis enables ﬂexible

production of functionalized prolines and carbapenams.

J Am Chem Soc 134, 471–479.

27 Hamed RB, Henry L, Gomez-Castellanos JR, Asghar

A, Brem J, Claridge TD & Schoﬁeld CJ (2013)

Stereoselective preparation of lipidated carboxymethyl-

proline/pipecolic acid derivatives via coupling of

engineered crotonases with an alkylmalonyl-CoA

synthetase. Org Biomol Chem 11, 8191–8196.

28 Hamed RB, Mecinovic J, Ducho C, Claridge TD &

Schoﬁeld CJ (2010) Carboxymethylproline synthase

catalysed syntheses of functionalised N-heterocycles.

Chem Commun (Camb) 46, 1413–1415.

29 Van Aalten DM, Milne KG, Zou JY, Kleywegt GJ,

Bergfors T, Ferguson MA, Knudsen J & Jones TA

(2001) Binding site differences revealed by crystal

structures of Plasmodium falciparum and bovine acyl-

CoA binding protein. J Mol Biol 309, 181–192.

30 Kragelund BB, Andersen KV, Madsen JC, Knudsen J

& Poulsen FM (1993) Three-dimensional structure of

the complex between acyl-coenzyme A binding protein

and palmitoyl-coenzyme A. J Mol Biol 230, 1260–

1277.

ꢀ

dehalogenase determined to 1.8 A resolution: an

enzyme catalyst generated via adaptive mutation.

Biochemistry 35, 8103–8109.

16 Kasaragod P, Schmitz W, Hiltunen JK & Wierenga RK

(2013) The isomerase and hydratase reaction

mechanism of the crotonase active site of the

multifunctional enzyme (type-1), as deduced from

structures of complexes with 3S-hydroxy-acyl-CoA.

FEBS J 280, 3160–3175.

17 Engel CK, Kiema TR, Hiltunen JK & Wierenga RK

(1998) The crystal structure of enoyl-CoA hydratase

complexed with octanoyl-CoA reveals the structural

adaptations required for binding of a long chain fatty

acid-CoA molecule. J Mol Biol 275, 859–847.

18 Mursula AM, Van Aalten DM, Hiltunen JK &

Wierenga RK (2001) The crystal structure of D³, D²-

enoyl-CoA isomerase. J Mol Biol 309, 845–853.

19 Pihko P, Rapakko S & Wierenga RK (2009) Oxyanion

holes and their mimics. In Hydrogen Bonding in

Organic Synthesis (Pihko P, ed.), pp. 43–71. Wiley-

VCH Verlag GmbH & Co. KGaE, Weinheim,

Germany.

20 Papai I, Hamza A, Pihko PM & Wierenga RK (2011)

Stereoelectronic requirements for optimal hydrogen-

bond-catalyzed enolization. Chemistry 17, 2859–2866.

21 Bach RD, Thorpe C & Dmitrenko O (2002) C–

Hꢀꢀꢀcarboxylate oxygen hydrogen bonding in substrate

activation by acyl-CoA dehydrogenases: synergy

between the H-bonds. J Phys Chem B 106,

4325–4335.

31 Hofstein HA, Feng Y, Anderson VE & Tonge PJ

(1999) Role of glutamate 144 and glutamate 164 in the

catalytic mechanism of enoyl-CoA hydratase.

Biochemistry 38, 9508–9516.

32 Wakil SJ (1956) Studies on the fatty acid oxidizing

system of animal tissues. IX. Stereospeciﬁcity of

unsaturated acyl CoA hydrase. Biochim Biophys Acta

19, 497–504.

22 Taskinen JP, van Aalten DM, Knudsen J & Wierenga

RK (2007) High resolution crystal structures of

unliganded and liganded human liver ACBP reveal a

new mode of binding for the acyl-CoA ligand. Proteins

66, 229–238.

23 Antonenkov VD, Van Veldhoven PP, Waelkens E &

Mannaerts GP (1997) Substrate speciﬁcities of 3-

oxoacyl-CoA thiolase A and sterol carrier protein 2/3-

oxoacyl-CoA thiolase puriﬁed from normal rat liver

peroxisomes. Sterol carrier protein 2/3-oxoacyl-CoA

thiolase is involved in the metabolism of 2-methyl-

branched fatty acids and bile acid intermediates. J Biol

Chem 272, 26023–26031.

33 Rasmussen JT, Borchers T & Knudsen J (1990)

Comparison of the binding afﬁnities of acyl-CoA-

binding protein and fatty-acid-binding protein for long-

chain acyl-CoA esters. Biochem J 265, 849–855.

34 Haataja TJ, Koski MK, Hiltunen JK & Glumoff T

(2011) Peroxisomal multifunctional enzyme type 2 from

the fruitﬂy: dehydrogenase and hydratase act as

separate entities, as revealed by structure and kinetics.

Biochem J 435, 771–781.

35 Ellman GL (1959) Tissue sulfhydryl groups. Arch

Biochem Biophys 82, 70–77.

36 Petoukhov MV, Franke D, Shkumatov AV, Tria G,

Kikhney AG, Gajda M, Gorba C, Mertens HDT,

Konarev PV & Svergun DI (2012) New developments

in the ATSAS program package for small-angle

scattering data analysis. J Appl Crystallogr 45,

342–350.

24 Koski KM, Haapalainen AM, Hiltunen JK & Glumoff

T (2005) Crystal structure of 2-enoyl-CoA hydratase-2

from human peroxisomal multifunctional enzyme type-

2. J Mol Biol 345, 1157–1169.

€ €

25 Mehtala ML, Haataja TJ, Blanchet CE, Hiltunen JK,

37 Svergun DI (1992) Determination of the regularization

parameter in indirect-transform methods using

perceptual criteria. J Appl Crystallogr 25, 495–503.

Svergun DI & Glumoff T (2013) Quaternary structure

of human, Drosophila melanogaster and Caenorhabditis

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

767

Structural enzymological studies of human ECI2

G. U. Onwukwe et al.

38 Franke D & Svergun DI (2009) DAMMIF, a program

for rapid ab-initio shape determination in small-angle

scattering. J Appl Crystallogr 42, 342–346.

49 Chen VB, Arendall WB III, Headd JJ, Keedy DA,

Immormino RM, Kapral GJ, Murray LW, Richardson

JS & Richardson DC (2010) MolProbity: all-atom

structure validation for macromolecular

39 Svergun DI, Petoukhov MV & Koch MHJ (2001)

Determination of domain structure of proteins from

X-ray solution scattering. Biophys J 80, 2946–2953.

40 Petoukhov MV & Svergun DI (2005) Global rigid body

modeling of macromolecular complexes against small-

angle scattering data. Biophys J 89, 1237–1250.

41 Kozin MB & Svergun DI (2001) Automated matching

of high- and low-resolution structural models. J Appl

Crystallogr 34, 33–41.

42 Svergun D, Barberato C & Koch MHJ (1995)

CRYSOL – a program to evaluate X-ray solution

scattering of biological macromolecules from atomic

coordinates. J Appl Crystallogr 28, 768–773.

43 Kabsch W (2010) XDS. Acta Crystallogr D 66, 125–132.

44 Evans PR & Murshudov GN (2013) How good are my

data and what is the resolution? Acta Crystallogr D 69,

1204–1214.

45 Collaborative Computational Project Number 4 (1994)

The CCP4 suite: programs for protein crystallography.

Acta Crystallogr D 50, 760–763.

crystallography. Acta Crystallogr D 66, 12–21.

50 Modis Y, Filppula SA, Novikov DK, Norledge B,

Hiltunen JK & Wierenga RK (1998) The crystal

ꢀ

structure of dienoyl-CoA isomerase at 1.5 A resolution

reveals the importance of aspartate and glutamate side

chains for catalysis. Structure 6, 957–970.

51 Larkin MA, Blackshields G, Brown NP, Chenna R,

McGettigan PA, McWilliam H, Valentin F, Wallace

IM, Wilm A, Lopez R et al. (2007) Clustal W and

Clustal X version 2.0. Bioinformatics 23, 2947–2948.

52 Krissinel E & Henrick K (2004) Secondary-structure

matching (SSM), a new tool for fast protein structure

alignment in three dimensions. Acta Crystallogr D 60,

2256–2268.

53 Kabsch W & Sander C (1983) Dictionary of protein

secondary structure: pattern recognition of hydrogen-

bonded and geometrical features. Biopolymers 22,

2577–2637.

54 Tamura K, Peterson D, Peterson N, Stecher G, Nei M

& Kumar S (2011) MEGA5: molecular evolutionary

genetics analysis using maximum likelihood,

evolutionary distance, and maximum parsimony

methods. Mol Biol Evol 28, 2731–2739.

55 Mylonas E & Svergun DI (2007) Accuracy of molecular

mass determination of proteins in solution by small-

angle X-ray scattering. J Appl Crystallogr 40,

s245–s249.

46 McCoy AJ, Grosse-Kunstleve RW, Adams PD, Winn

MD, Storoni LC & Read RJ (2007) Phaser

crystallographic software. J Appl Crystallogr 40, 658–674.

47 Murshudov GN, Skubak P, Lebedev AA, Pannu NS,

Steiner RA, Nicholls RA, Winn MD, Long F & Vagin AA

(2011) REFMAC5 for the reﬁnement of macromolecular

crystal structures. Acta Crystallogr D 67, 355–367.

48 Emsley P & Cowtan K (2004) Coot: model-building

tools for molecular graphics. Acta Crystallogr D 60,

2126–2132.

768

FEBS Journal 282 (2015) 746–768 ª 2014 FEBS

Copyright of FEBS Journal is the property of Wiley-Blackwell and its content may not be

copied or emailed to multiple sites or posted to a listserv without the copyright holder's

express written permission. However, users may print, download, or email articles for

individual use.

Article Doi

Human Δ3,Δ2-enoyl-CoA isomerase, type 2: A structural enzymology study on the catalytic role of its ACBP domain and helix-10

DOI: 10.1111/febs.13179

Source and publish data:

Authors:

Article abstract of DOI:10.1111/febs.13179

Full text of DOI:10.1111/febs.13179

Products guided by the article

R&D Labs maybe for 10018-93-6

Relevant to this article

Hot Product