Biochemical characterization of human
3-methylglutaconyl-CoA hydratase and its role in leucine
metabolism
Matthias Mack
1
, Ute Schniegler-Mattox
2
, Verena Peters
3
, Georg F. Hoffmann
3
, Michael Liesert
4
,
Wolfgang Buckel
4
and Johannes Zschocke
2
1 Institut fu
¨
r Technische Mikrobiologie der Hochschule Mannheim, Germany
2 Institut fu
¨
r Humangenetik, Ruprecht-Karls-Universita
¨
t Heidelberg, Germany
3 Abteilung fu
¨
r Allgemeine Pa
¨
diatrie, Ruprecht-Karls-Universita

coenzyme A hydratase; AUH
Correspondence
M. Mack, Institut fu
¨
r Technische
Mikrobiologie der Hochschule Mannheim,
Windeckstr. 110, 68163 Mannheim,
Germany
Fax: +49 6212926420
Tel: +49 6212926496
E-mail:
(Received 14 September 2005, revised
3 March 2006, accepted 7 March 2006)
doi:10.1111/j.1742-4658.2006.05218.x
The metabolic disease 3-methylglutaconic aciduria type I (MGA1) is char-
acterized by an abnormal organic acid profile in which there is excessive
urinary excretion of 3-methylglutaconic acid, 3-methylglutaric acid and
3-hydroxyisovaleric acid. Affected individuals display variable clinical
manifestations ranging from mildly delayed speech development to severe
psychomotor retardation with neurological handicap. MGA1 is caused by
reduced or absent 3-methylglutaconyl-coenzyme A (3-MG-CoA) hydratase
activity within the leucine degradation pathway. The human AUH gene has
been reported to encode for a bifunctional enzyme with both RNA-binding
and enoyl-CoA-hydratase activity. In addition, it was shown that muta-
tions in the AUH gene are linked to MGA1. Here we present kinetic data
of the purified gene product of AUH using different CoA-substrates. The
best substrates were (E)-3-MG-CoA (V
max
¼ 3.9 UÆmg
)1

3-MG-CoA (Fig. 1) whilst 3-hydroxyisovaleric acid is
produced via the enzymatic hydration of 3-methylcro-
tonyl-CoA (crotonase, EC 4.2.1.17) (Fig. 1) [2]. Conse-
quently, humans with reduced or absent 3-MG-CoA
hydratase activity show excessive urinary excretion of
3-methylglutaconic acid, 3-hydroxyisovaleric acid and
Fig. 1. The metabolic pathway of (S)-leucine
(
L-leucine) and isovalerate. Enzymes
involved are as follows: 1, EC 2.6.1.42,
branched chain amino transferase 1; 2, EC
1.2.4.4 ⁄ 2.3.1.168 ⁄ 1.8.1.4, branched chain
2-keto acid dehydrogenase complex; 3, EC
1.3.99.10, isovaleryl-CoA dehydrogenase;
4, EC 6.4.1.4, 3-methylcrotonyl-CoA
carboxylase 1; 5, EC 4.2.1.18, 3-methylgluta-
conyl-CoA hydratase; 6, EC 4.1.3.4,
3-hydroxy-3-methylglutaryl-CoA lyase; 7, EC
2.8.3.–, isovalerate-CoA-transferase; 8, crot-
onase, EC 4.2.1.17. 9, unknown.
Table 1. Enzymes, genes, and associated diseases of the human leucine degradation pathway.
Enzyme name EC Gene OMIM
Branched chain amino transferase 1 2.6.1.42 BCAT1 113520
Branched chain keto acid dehydrogenase E1,
alpha ⁄ beta subunits
1.2.4.4 BCKDHA 608348
BCKDHB 248611
Dihydrolipoamide branched chain transacylase E2 2.3.1.168 DBT 248610
Dihydrolipoamide dehydrogenase E3 1.8.1.4 DLD 246900
Isovaleryl-CoA dehydrogenase 1.3.99.10 IVD 607036

that a second route for 3-MG biosynthesis exists [8].
Interestingly, certain patients with Smith–Lemli–Opitz
syndrome also show abnormally increased plasma
levels of this compound, further challenging our under-
standing of 3-methylglutaconic acid metabolism [9].
Lastly, pregnancy was reported as a possible cause of
MGA [10]. For a long time it had been unclear which
enzyme was responsible for the hydratase step within
leucine degradation. A 3-MG-CoA hydratase was
partially purified from bovine ⁄ ovine liver [11]. It was
established that this enzyme catalyses the syn-addition
of water to (E)-3-MG-CoA leading to (S)-HMG-CoA
[12]. Another enoyl-CoA hydratase, mitochondrial
crotonase, is not active using HMG-CoA and measur-
ing the reverse (dehydration) reaction [13]. Mitochond-
rial trifunctional protein (MTP) is the main enoyl-CoA
hydratase in long chain fatty acid b-oxidation [14].
This enzyme, however, is unlikely to be involved in
leucine degradation since MTP deficiency (MIM
143450, MIM 600890) is not associated with increased
urinary excretion of 3-methylglutaconic acid. A protein
was purified from human brain cells by affinity chro-
matography using the immobilized RNA-oligonucleo-
tide (AUUUA)
5
or ‘AU’ followed by cloning of the
corresponding gene [15]. Interestingly, the gene showed
sequence similarity to enoyl-CoA-hydratases-1 (2-
trans-enoyl-CoA-hydratases; EC 4.2.1.17) and its gene
product had weak enoyl-CoA-hydratase activity using

The present work was initiated to kinetically charac-
terize AUH on its presumed natural substrate 3-MG-
CoA using a new strategy for its synthesis and
developing a new assay. In addition, a mutant form of
AUH (A240V) derived from an MGA1 patient was
tested using 3-MG-CoA.
Results
Overexpression of AUH in Escherichia coli and
purification of the corresponding gene product
The gene for AUH which was cloned from a cDNA
library by Nakagawa et al. [15] encodes 339 amino
acids specifying a 40-kDa protein (AUHp40). Western
blot analysis of brain extracts consistently revealed a
32 kDa AUH protein and it was thus assumed that
the mature form of human AUH in brain has a
molecular weight of 32 kDa (AUHp32) [15]. For the
kinetic characterization of AUH described in the work
at hand, AUH was overproduced in Escherichia coli as
a maltose binding protein fusion (MBP-AUH). The
complete AUH gene (producing MBP-AUHp40 in
E. coli) but also a truncated form of AUH (producing
MBP-AUHp32 in E. coli) were ligated into the
Human 3-methylglutaconyl-CoA hydratase M. Mack et al.
2014 FEBS Journal 273 (2006) 2012–2022 ª 2006 The Authors Journal compilation ª 2006 FEBS
bacterial expression vector pMAL-c2. Consequently,
two different forms of AUH, namely MBP-AUHp40
and MBP-AUHp32 could be isolated from the corres-
ponding E. coli strains. MBP-AUHp40 and MBP-
AUHp32 were purified to apparent homogeneity. In a
subsequent step, the MBP portion of both fusion pro-

CoA and free coenzyme A. The compounds producing
the following signals (peak 3, peak 4 and peak 5) were
found to all have the same relative molecular mass of
893 matching the calculated molecular mass of 3-MG-
CoA (893.647). Thus, three 3-MG-CoA isomers were
produced using the enzyme glutaconate CoA-transferase
(Fig. 4). The three different forms of 3-MG-CoA were
separated by HPLC, collected and their concentration
was determined using an enzymatic 5,5¢-dithiobis-2-ni-
trobenzoate-based assay. Subsequently, the 3-MG-CoA
isomers were tested using AUH (Fig. 3). It was found,
that peak 5 (2 mm) was readily converted to (S)-HMG-
CoA. In addition, free CoA was detected. Peak 4 (2 mm)
produced significantly less HMG-CoA and also, in this
reaction, a substantial amount of free CoA was found.
Peak 3 (0.5 mm) gave mainly free CoA and only small
amounts of HMG-CoA. Peak 5, being the best sub-
strate, should correspond to (E)-3-MG-1-CoA, the inter-
mediate of the leucine degradation pathway (Fig. 4).
Peak 4 is most likely to correspond to (E)-3-MG-5-CoA.
Peak 3 is probably (Z)-3-MG-5-CoA.
Glutaconyl-CoA was prepared accordingly. Also in
this reaction two compounds were produced by glut-
aconate CoA-transferase. The molecules were separ-
ated by HPLC, analyzed by mass spectrometry and
were found to both have the same relative molecular
mass of 881 corresponding to glutaconyl-CoA
(881.247). Peak 1 was dominant and most likely was
glutaconyl-1-CoA. Peak 2 probably was glutaconyl-5-
CoA. The two isomers were separated from each

-COO
–
).
In the case of glutaconate CoA-transferase from A. fermentans,
CoAS
–
transiently is bound to the c-carboxyl group of bE54 [24].
M. Mack et al. Human 3-methylglutaconyl-CoA hydratase
FEBS Journal 273 (2006) 2012–2022 ª 2006 The Authors Journal compilation ª 2006 FEBS 2015
MBP fusion (MBP-AUHp40
mut
)inE. coli according
to the wild-type enzyme and tested using the substrate
(E)-3-MG-1-CoA. In these experiments, the specific
activity of MBP-AUHp40
mut
was 9% (0.068 UÆmg
)1
protein) in comparison to the wild-type enzyme
(0.76 UÆmg
)1
protein). Hence, the mutation A240V cau-
ses a significant loss of enzyme activity.
Direct nonisotopic assay of 3-MG-CoA hydratase
in cultured human skin fibroblasts
The nonisotopic 3-MG-CoA hydratase assay that was
developed during this work was evaluated for its use
in testing homogenates derived from human skin fibro-
blast cultures. Different cell cultures derived from type
I MGA patients and from wild-type controls were

HMG-CoA
800
CoA
(E)-3-methyl-
glutaconyl-1-CoA
AUHAUHAUH
0
10
20
min
0
20
40
60
80
100
120
mAU
CoA
(Z)-3-methyl-
glutaconyl-
5-CoA
010
20 min
200
400
600
800
1000
mAU

the respective isomers of 3-MG-CoA in a total volume of 25 lL. The reaction was started by addition of AUH (1 lg), incubated for 1 h and
the CoA products were HPLC-detected by their absorbance at 260 nm. Peak 3 produced small amounts of HMG-CoA and large amounts of
free CoA. Peak 4 produced HMG-CoA and also large amounts of free CoA. Peak 5 produced large amounts of HMG-CoA, but also free CoA.
Human 3-methylglutaconyl-CoA hydratase M. Mack et al.
2016 FEBS Journal 273 (2006) 2012–2022 ª 2006 The Authors Journal compilation ª 2006 FEBS
readily detectable (Fig. 5). We investigated fibroblast
homogenates from two controls and fibroblast homo-
genates from three patients with established MGA
type 1. Patient 1 [20] was homozygous for a mutation
leading to a stop codon at residue 197 (R197X) and
patient 2 [20] was homozygous for a mutation at the
splice acceptor site of intron 8 (IVS8–1G>A). Patient
3 [20] was compound heterozygous for a missense
mutation A240V (c.719C>T) in exon 7 and an inser-
tion mutation c.613–614insA. This insertion causes a
frameshift that starts at Met205 and leads to the intro-
duction of a stop codon after four amino acids. The
intra-assay variation, estimated by measuring four
fibroblast homogenates in a single experiment, was
3.9%, the interassay variation was 5.4% (n ¼ 3 days).
The fibroblast material from all MGA1 patients pro-
duced significantly less (4–16 mUÆmg
)1
protein,
mean ¼ 8mUÆmg
)1
protein) of 3-hydroxyglutaryl-CoA
as compared to the two controls (72 mUÆmg
)1
protein

hydratase).
Substrate
K
m
(lM)
V
max
a
(UÆmg
)1
)
k
cat
(s
)1
)
k
cat
⁄ K
m
(lM
)1
Æs
)1
)
(E)-3-Methylglutaconyl-1-CoA 8.3 3.9 5.1 0.6
(R,S)-3-Hydroxy-3-
methylglutaryl-CoA
2250 0.2 0.26 1.2
)4

associated with mutations in the AUH gene. This indi-
cated that AUH, in addition to its RNA binding func-
tion, must play an important role in leucine catabolism.
AUH was overproduced in E. coli and characterized
by measuring the reverse reaction, the dehydration of
HMG-CoA to 3-MG-CoA. The reaction was followed
photometrically [4].
In order to measure AUH in the forward reaction
(hydratase activity), it was necessary to synthesize
3-MG-CoA, which is not commercially available. Glut-
aconate CoA-transferase (Gct) from A. fermentans
proved to be useful for the enzymatic production of
this compound. Earlier, Gct was reported to be specific
for (E)-glutaconate and to be completely inactive with
(Z)-glutaconate [21]. The activity of Gct using the
CoA-donor acetyl-CoA and the CoA-acceptor 3-meth-
ylglutaconate was relatively low. Our data suggest that
Gct produces three isomers. The production of
(Z)-3-MG-5-CoA is probably due to the possible
trans-conformation of the C
5
-carboxyl group of (Z)-3-
methylglutaconate (Fig. 4). Most of (Z)-3-MG-5-CoA
was hydrolyzed by AUH to give free CoA and (Z)-3-
methylglutaconate, trace amounts of HMG-CoA,
however, were detected. (E)-3-MG-5-CoA was a better
substrate for hydration with AUH. An explanation for
this may be, that upon binding to AUH, (E)-3-MG-5-
CoA is isomerized to give (E)-3-MG-1-CoA (Fig. 4A).
An intrinsic isomerase activity has also been reported

4
AB C
mAU
0
20
40
60
80
100
120
140
mAU
0
20
40
60
80
100
120
140
0
10
20 min
1
2
3
4
0 10 20 min
1
2

dratase reaction using HMG-CoA. In these experi-
ments, an acyl-CoA-hydrolase activity of AUH was
not detected. This is the first report showing kinetic
data for purified AUH, although a 3-MG-CoA hydra-
tase activity was found earlier in fibroblast and lym-
phocyte lysates measuring the hydration reaction [3].
At that time, the substrate [5-
14
C]3-MG-CoA was pre-
pared by incubation of 3-methylcrotonyl-CoA with
3-methylcrotonyl-CoA-carboxylase in the presence of
NaH
14
CO
3
. In this work, K
m
values for the hydration
of [5-
14
C]3-MG-CoA of 6.9 lm (fibroblast) and 9.4 lm
(lymphocyte), respectively, were reported [3]. The
formation of [5-
14
C]3-methylglutaconate from [5-
14
C]3-
methylglutaconyl-CoA was interpreted as nonspecific
hydrolysis. Our results suggest that CoA-hydrolysis is
an intrinsic function of AUH.

(A240V),
identified in one MGA1 patient, had a clearly reduced
3-MG-CoA hydratase activity (9% of the wild-type
enzyme). This finding provides further evidence con-
firming that AUH is indeed the main hydratase in the
human leucine degradation pathway and that muta-
tions leading to reduced hydratase activity are respon-
sible for the MGA1 phenotype.
The need to differentiate patients with AUH defi-
ciency from patients with other forms of MGA
requires the availability of a sensitive and specific
enzyme assay. Our data show that the hydratase
reaction of AUH is favored over the dehydratase
reaction (factor of 20). Hence, measuring the for-
ward reaction in fibroblast homogenates of patient-
derived cells should increase the sensitivity of an
AUH test. The product of this reaction, however, is
the common intermediate HMG-CoA, which is
quickly degraded by, e.g. 3-hydroxy-3-methylglutaryl-
CoA lyase (EC 4.1.3.4), to give acetyl-CoA and
acetoacetate. As glutaconyl-CoA is a very good
substrate for AUH and since the product of the hy-
dratase reaction, 3-hydroxyglutaryl-CoA, is not an
intermediate within human metabolism, we hypothes-
ized that glutaconyl-CoA may be used as a substrate
for testing AUH activity in a routine setting. Indeed,
we were able to show that AUH activity in fibro-
blasts can be determined by monitoring the forma-
tion of 3-hydroxyglutaryl-CoA. The production of a
small amount of 3-hydroxyglutaryl-CoA in a patient

Mutagenesis Kit and the mismatch oligonucleotides AUH
FW 5¢-AGCTCATATTCTCTGTGCGAGTCCTCGATG
GC-3¢ and AUH RP 5¢-GCCATCGAGGACTCGC
ACA
GAGAATATGAGCT-3¢ (the c.719C>T mutation leading
to the amino acid exchange A240V is underlined). The
AUH genes were proof-sequenced and no secondary muta-
tions were detected.
M. Mack et al. Human 3-methylglutaconyl-CoA hydratase
FEBS Journal 273 (2006) 2012–2022 ª 2006 The Authors Journal compilation ª 2006 FEBS 2019
Mass spectrometry
The CoA-esters were separated by HPLC, ionized by ESI
and detected by TOF. The HPLC system consisted of a
HP1100 series binary-gradient pump, a vacuum degasser (all
from Hewlett-Packard), and a CTC HTS PAL autosampler
(CTC). The dry sample was dissolved in water and 20 lL
injected onto a 4 · 40-mm Grom-Sil120 ODS-4 HE column
(3-lm particle diameter; Grom). The samples were separated
from interfering compounds by a gradient between solution
B (acetonitrile +1 vol% formic acid) and solution A (water
+ 1 volume % formic acid). The gradient (1 mLÆmin
)1
) was
as follows: 0–5 min, 0% B to 83% B; 6–8 min, 100% B. All
gradient steps were linear, and the total analysis time, inclu-
ding equilibration, was 10 min. A splitter between the
HPLC column and the mass spectrometer was used, and
100 lLÆmin
)1
of eluent was introduced into the mass spec-

Waters Sep – Pak tC18 column (1 mL cartridge, 100 mg
sorbent) (Waters, Eschborn, Germany). The column was
first treated with 10 volumes 0.1% (v ⁄ v) trifluoroacetic acid
in H
2
O, 50% acetonitrile (v ⁄ v). The acetonitrile component
also contained 0.1% trifluoroacetic acid. Subsequently, the
column was washed with 10 volumes 0.1% (v ⁄ v) trifluoro-
acetic acid in H
2
O. The samples were loaded and the column
was washed with 10 volumes 0.1% (v ⁄ v) trifluoroacetic acid
in H
2
O. The CoA-ester was eluted with 10 volumes 0.1%
(v ⁄ v) trifluoroacetic acid in H
2
O, 50% acetonitrile (v ⁄ v) con-
taining 0.1% trifluoroacetic acid. After evaporation in vacuo,
the CoA-ester was dissolved in water and stored at )20 °C.
For analytic and preparative purposes a Phenomenex Syn-
ergi 4 l Polar-RP 80 A column (5 lm) was used at a flow
rate of 1 mLÆmin
)1
(Phenomenex, Aschaffenburg, Ger-
many). The eluents were 0.1% trifluoroacetic acid (v ⁄ v) in
H
2
O (solution A) and 0.085% trifluoroacetic acid (v ⁄ v) in
acetonitrile (solution B). The columns were equilibrated for

and 3-hydroxybutyryl-CoA were obtained from Sigma-
Aldrich.
Assay of 3-MG-CoA hydratase
The assay contained in a total volume of 25 lL50mm Tris
HCl pH 7.4, 10 mm EDTA, 1 mgÆmL
)1
bovine serum albu-
min and 0.05–0.2 mm 3-MG-CoA. The reaction was started
by addition of the enzyme (1 lg). The products of the reac-
tion were analyzed by HPLC as described above and mass
spectrometry. The kinetic constants K
m
(lM) and V
max
(UÆmg protein
)1
) were evaluated with the Michaelis-Menten
equation and Lineweaver-Burk plots using the Microsoft
Excel program. The turnover numbers, k
cat
(s
)1
), were
calculated with the subunit molecular mass (78.4 Da) of
MBP-AUHp40.
Direct nonisotopic assay of 3-MG-CoA hydratase
in cultured human skin fibroblasts
Fibroblasts were grown and harvested as described elsewhere
[26]. Cells were suspended in 200 lL phosphate-buffered
Human 3-methylglutaconyl-CoA hydratase M. Mack et al.

References
1 Sweetman L & Williams JC (2001) Branched chain
organic acidurias. The Metabolic and Molecular Bases of
Inherited Disease (Scriver, C R, Beaudet, A L, Sly, W S
& Valle, D, eds), pp. 2125–2163. McGraw-Hill, New
York.
2 Duran M, Beemer FA, Tibosch AS, Bruinvis L, Ketting
D & Wadman SK (1982) Inherited 3-methylglutaconic
aciduria in two brothers – another defect of leucine
metabolism. J Pediatr 101, 551–554.
3 Narisawa K, Gibson KM, Sweetman L, Nyhan WL,
Duran M & Wadman SK (1986) Deficiency of 3-methyl-
glutaconyl-coenzyme A hydratase in two siblings with
3-methylglutaconic aciduria. J Clin Invest 77, 1148–1152.
4 Shoji Y, Takahashi T, Sawaishi Y, Ishida A, Matsumori
M, Enoki M, Watanabe H & Takada G (1999) 3-Methyl-
glutaconic aciduria type I: clinical heterogeneity as a
neurometabolic disease. J Inherit Metab Dis 22, 1–8.
5 Gibson KM, Elpeleg ON, Jakobs C, Costeff H & Kelley
RI (1993) Multiple syndromes of 3-methylglutaconic
aciduria. Pediatr Neurol 9, 120–123.
6 Bione S, D’Adamo P, Maestrini E, Gedeon AK, Bolhu-
is PA & Toniolo D (1996) A novel X-linked gene, G4.5.
is responsible for Barth syndrome. Nat Genet 12, 385–
389.
7 Anikster Y, Kleta R, Shaag A, Gahl WA & Elpeleg O
(2001) Type III 3-methylglutaconic aciduria (optic
atrophy plus syndrome, or Costeff optic atrophy
syndrome): identification of the OPA3 gene and its
founder mutation in Iraqi Jews. Am J Hum Genet 69,

J, Jeno P & Moroni C (1995) AUH, a gene encoding an
AU-specific RNA binding protein with intrinsic enoyl-
CoA hydratase activity. Proc Natl Acad Sci USA 92,
2051–2055.
16 Brennan LE, Nakagawa J, Egger D, Bienz K & Moroni
C (1999) Characterisation and mitochondrial localisa-
tion of AUH, an AU-specific RNA-binding enoyl-CoA
hydratase. Gene 228, 85–91.
17 Nakagawa J & Moroni C (1997) A 20-amino-acid
autonomous RNA-binding domain contained in an
enoyl-CoA hydratase. Eur J Biochem 244, 890–899.
18 Kurimoto K, Fukai S, Nureki O, Muto Y & Yokoyama
S (2001) Crystal structure of human AUH protein, a
single-stranded RNA binding homolog of enoyl-CoA
hydratase. Structure (Cambridge) 9, 1253–1263.
19 Liesert M (2000) Untersuchungen Zur Biochemie der
Vererbbaren Stoffwechselerkrankung Glutarazidurie
Typ I, PhD Thesis, Philipps-Universita
¨
t, Marburg.
20 Ly TBN, Peters V, Gibson KM, Liesert M, Buckel W,
Wilcken B, Carpenter K, Ensenauer R, Hoffmann GF,
Mack M et al. (2003) Mutations in the AUH gene cause
3-methylglutaconic aciduria type I. Hum Mutat 21, 401–
407.
M. Mack et al. Human 3-methylglutaconyl-CoA hydratase
FEBS Journal 273 (2006) 2012–2022 ª 2006 The Authors Journal compilation ª 2006 FEBS 2021
21 Buckel W, Dorn U & Semmler R (1981) Glutaconate
CoA-transferase from Acidaminococcus fermentans. Eur
J Biochem 118, 315–321.