Crystal structure of human gamma-butyrobetaine hydroxylase

Article abstract of DOI:10.1016/j.bbrc.2010.06.121

Gamma-butyrobetaine hydroxylase (GBBH) is a 2-ketoglutarate-dependent dioxygenase that catalyzes the biosynthesis of l-carnitine by hydroxylation of gamma-butyrobetaine (GBB). l-carnitine is required for the transport of long-chain fatty acids into mitochondria for generating metabolic energy. The only known synthetic inhibitor of GBBH is mildronate (3-(2,2,2-trimethylhydrazinium) propionate dihydrate), which is a non-hydroxylatable analog of GBB.To aid in the discovery of novel GBBH inhibitors by rational drug design, we have solved the three-dimensional structure of recombinant human GBBH at 2.0. ? resolution. The GBBH monomer consists of a catalytic double-stranded β-helix (DBSH) domain, which is found in all 2KG oxygenases, and a smaller N-terminal domain. Extensive interactions between two monomers confirm earlier observations that GBBH is dimeric in its biological state. Although many 2KG oxygenases are multimeric, the dimerization interface of GBBH is very different from that of related enzymes.The N-terminal domain of GBBH has a similar fold to the DUF971 superfamily, which consists of several short bacterial proteins with unknown function. The N-terminal domain has a bound Zn ion, which is coordinated by three cysteines and one histidine. Although several other 2KG oxygenases with known structures have more than one domain, none of them resemble the N-terminal domain of GBBH. The N-terminal domain may facilitate dimer formation, but its precise biological role remains to be discovered.The active site of the catalytic domain of GBBH is similar to that of other 2KG oxygenases, and Fe(II)-binding residues form a conserved His-X-Asp-X. _n-His triad, which is found in all related enzymes.

Full text of DOI:10.1016/j.bbrc.2010.06.121

Biochemical and Biophysical Research Communications 398 (2010) 634–639
Contents lists available at ScienceDirect
Biochemical and Biophysical Research Communications
journal homepage: www.elsevier.com/locate/ybbrc
Crystal structure of human gamma-butyrobetaine hydroxylase
a
a
a
a
Kaspars Tars a, , Janis Rumnieks , Andris Zeltins , Andris Kazaks , Svetlana Kotelovica ,
*
Ainars Leonciks a, Jelena Sharipo a, Arturs Viksna b, Janis Kuka c, Edgars Liepinsh c, Maija Dambrova c
a Biomedical Research and Study Centre, Ratsupites 1, LV1067, Riga, Latvia
b Faculty of Chemistry, University of Latvia, Kr. Valdemara 48, LV1013, Riga, Latvia
c Latvian Institute of Organic Synthesis, Aizkraukles 21, LV1006, Riga, Latvia
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 24 June 2010
Available online 1 July 2010
Gamma-butyrobetaine hydroxylase (GBBH) is a 2-ketoglutarate-dependent dioxygenase that catalyzes
the biosynthesis of L-carnitine by hydroxylation of gamma-butyrobetaine (GBB). L-carnitine is required
for the transport of long-chain fatty acids into mitochondria for generating metabolic energy. The only
known synthetic inhibitor of GBBH is mildronate (3-(2,2,2-trimethylhydrazinium) propionate dihydrate),
which is a non-hydroxylatable analog of GBB.
To aid in the discovery of novel GBBH inhibitors by rational drug design, we have solved the three-
dimensional structure of recombinant human GBBH at 2.0 Å resolution. The GBBH monomer consists
of a catalytic double-stranded b-helix (DBSH) domain, which is found in all 2KG oxygenases, and a smal-
ler N-terminal domain. Extensive interactions between two monomers confirm earlier observations that
GBBH is dimeric in its biological state. Although many 2KG oxygenases are multimeric, the dimerization
interface of GBBH is very different from that of related enzymes.
Keywords:
Gamma-buyrobetaine hydroxylase
Dioxygenase
Carnitine
2-Ketoglutarate
Mildronate
The N-terminal domain of GBBH has a similar fold to the DUF971 superfamily, which consists of several
short bacterial proteins with unknown function. The N-terminal domain has a bound Zn ion, which is
coordinated by three cysteines and one histidine. Although several other 2KG oxygenases with known
structures have more than one domain, none of them resemble the N-terminal domain of GBBH. The
N-terminal domain may facilitate dimer formation, but its precise biological role remains to be discov-
ered.
The active site of the catalytic domain of GBBH is similar to that of other 2KG oxygenases, and Fe(II)-
binding residues form a conserved His–X–Asp–Xn–His triad, which is found in all related enzymes.
Ó 2010 Elsevier Inc. All rights reserved.
1. Introduction
important for the enzymatic activity of 2KG oxygenases. Ascorbate
acts as a reducing agent, maintaining the Fe(II) ion in its reduced
Gamma-butyrobetaine hydroxylase (GBBH, EC1.14.11.1) is an
enzyme that catalyzes the last step in the biosynthesis of -carnitine,
state, whereas catalase neutralizes hydrogen peroxide, a toxic by-
product of oxidation reactions. Structurally, the catalytic domains
of 2KG oxygenases share a common double-stranded b-helix fold
(DBSH) and a conserved Fe(II) binding triad His–X–Asp/Glu–Xn –
His. Apart from the DBSH core domain, several 2KG oxygenases
and related enzymes have additional N- or C-terminal domains with
L
which is required for the transport of long-chain fatty acids into the
mitochondrial matrix for subsequent b-oxidation and cellular en-
ergy production [1–3]. GBBH belongs to a class of enzymes called
2-ketoglutarate (2KG) oxygenases, which catalyze a broad range
of oxidation reactions in virtually all living organisms. Hydroxyl-
ation is the most common reaction catalyzed by 2KG oxygenases
and remains the only activity reported for these enzymes in animals
[4]. All 2KG oxygenases utilize 2KG and O2 as substrates and Fe(II)
ion as a cofactor and, in addition to their specific biochemical prod-
uct, release succinate and CO2. Ascorbate and catalase are also
various functions. For example, the N-terminal
hydroxypropylphosphonic acid epoxidase is involved in subunit
oligomerization [5], and the C-terminal -helical bundle domain
a-helical domain of
a
of bacterial proline-3-hydroxylase has been suggested to play a role
in catalysis [6].
Many 2KG oxygenases are monomeric in solution, whereas some
exist as dimers, tetramers or hexamers [4]. GBBH has been shown to
form stable dimers [7]. In general, 2KG oxygenases are structurally
well characterized with more than 50 representatives in the Protein
Data Bank (for review see [4]). Nevertheless, none of the available
structures displays significant sequence identity to GBBH.
Abbreviations: GBBH, gamma-butyrobetaine hydroxylase; GBB, gamma-butyr-
obetaine; 2KG, 2-ketoglutarate; DBSH, double-stranded b-helix.
* Corresponding author. Fax: +371 67442407.
E-mail address: kaspars@biomed.lu.lv (K. Tars).
0006-291X/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.bbrc.2010.06.121

K. Tars et al. / Biochemical and Biophysical Research Communications 398 (2010) 634–639
635
In human tissues, GBBH is expressed only in liver, kidney and to
a lesser extent in the brain [8,9]. The enzymatic activity of GBBH
volume of 0.2 ml, contained the following: 20 mM potassium phos-
phate pH 7.0, 20 mM potassium chloride, 3 mM 2-ketoglutarate,
0.25 mM ferrous ammonium sulfate, 10 mM sodium ascorbate,
has been used to increase the bioavailability of
administering gamma-butyrobetaine (GBB), the substrate for
GBBH and a bio-precursor of -carnitine [1,10–12]. The only known
L-carnitine by
0.16 mg catalase, 3–500 lM GBB and an appropriate amount of
L
GBBH. The reaction was initiated by addition of GBBH, and the mix-
ture was incubated at 37 °C for 120 min. The reaction was stopped
by addition of 0.8 ml of cold acetonitrile:methanol (1:3). The mix-
ture was spun down at 20,000 g for 10 min at 4 °C; the supernatant
was decanted and used to measure the amount of carnitine. The IC50
for mildronate was determined under the same conditions but with
synthetic inhibitor of GBBH is mildronate (3-(2,2,2-trimethylhydr-
azinium) propionate dihydrate), a structural non-hydroxylatable
analog of GBB. Mildronate is clinically used as an anti-ischemic
drug and a regulator of cellular energy metabolism [13,14]. By
inhibiting GBBH and, subsequently, the availability of L-carnitine,
mildronate stimulates glucose metabolism, which can be beneficial
under certain physiological or pathological conditions, such as
ischemia-related pathological states, metabolic syndrome and dia-
betes [13–15].
the addition of mildronate (3 lM to 300 lM). The concentration of L-
carnitine was measured by ultra performance liquid chromatogra-
phy-tandem mass spectrometry (UPLC/MS/MS) in positive ion elec-
trospray mode, as described previously [19].
To find novel pharmacological agents for the regulation of L-car-
nitine pathways and to facilitate rational structure-based drug de-
sign, we have solved the crystal structure of GBBH. The structure
has revealed several unknown features of 2KG oxygenases, such
as an N-terminal zinc-binding domain and a novel dimerization
interface.
2.4. Crystallization and data collection
GBBH was crystallized by a sitting drop vapor technique by
mixing 1
ll protein (7 mg/ml) in 20 mM tris–HCl pH 8.0 with
1
ll of reservoir solution (1.2 M ammonium sulfate, 100 mM so-
dium acetate, pH 4.5). In co-crystallization trials with ligands, each
ligand was added to the reservoir solution to a final concentration
of 5 mM. Needle-shaped crystals appeared after 3–4 h and reached
a maximum size of 0.2–0.4 mm in 24 h. SeMet-labeled protein was
crystallized under the same conditions as above. The crystals were
soaked in the mother liquor with 30% glycerol for approximately
one minute and flash frozen in liquid nitrogen. Data were collected
at ESRF beamlines ID23–1 and BM14u. The best diffracting crystals
were those with mildronate, which diffracted to 2.0 Å resolution.
SeMet-labeled crystals diffracted to 3.5 Å resolution, and crystals
with other ligands diffracted to 3.0–3.2 Å resolution. From the Se-
Met-labeled crystal, two datasets were collected at 0.97856 Å
(peak) and 0.91840 Å (remote) wavelengths at 3.5 Å resolution.
2. Materials and methods
2.1. Cloning, expression and purification
The coding sequence of GBBH with Escherichia coli-optimized
codons was provided by ASLA Ltd. (Riga). The gene was cloned in
a modified pColdI vector (Takara Bio Inc.). As a result, the protein
had the following amino acid sequence before the start codon of
GBBH: MNHKV (pColdI leader sequence) – HHHHHH (His-tag) –
IEGR (Factor Xa cleavage site)-S.
E. coli strain WK6 [16] was co-transformed with two plasmids
encoding (a) the human recombinant gamma-butyrobetaine
hydroxylase (GBBH) and (b) the bacterial chaperones GroES and
GroEL under the control of an IPTG-inducible lac promoter. The
transformed cells were grown in LB medium supplemented with
0.1% glucose, 50 mg/l ampicillin and 50 mg/l kanamycin at 37 °C
until OD600 = 0.6. IPTG was added to a final concentration of
1 mM, and the culture was incubated at 37 °C for 30 min followed
by 16 h at 15 °C. The cells were harvested by centrifugation, resus-
pended in PBS buffer, pH 7.4, lysed by sonication, and the insoluble
fraction was removed by centrifugation. About 10% of the GBBH
was in the soluble fraction. The lysate was applied to a Ni–NTA
agarose column (Qiagen) followed by gel filtration on a Superdex
200 column (GE Healthcare) and ion exchange chromatography
on a Mono Q column (GE Healthcare).
2.5. Structure determination
Indexing by MOSFLM [20] indicated a trigonal lattice, and sub-
sequent scaling by SCALA [21] confirmed space group P6122 or
P6522 as judged by scaling statistics and systematic absences.
Determination of Se atom positions and initial phasing was per-
formed in SHELX C/D/E [22] in both P6122 and P6522 space groups.
The resulting electron density maps were manually inspected, and
protein-like features, such as right-handed
a helices and b strands,
were detected for the P6522 space group. Heavy atom refinement
and phasing was done by SHARP [23], and the resulting map was
improved by solvent flattening with SOLOMON [24]. The initial
model was built automatically in BUCANEER [25]. The resulting
partial model, which covered about 50% of the sequence, was man-
ually inspected and improved in COOT [26]. The coordinates were
used as a molecular replacement model for the 2.0 Å dataset, and
the improved electron density map was used for an additional
BUCANEER round. The remaining parts of the model were built
manually in O [27] and COOT. Structure refinement was done by
REFMAC [28]. Water molecules, zinc and sulfate ions were added
in COOT. All automatically added water molecules were manually
inspected, and some were removed due to bad density, bad con-
tacts or very high B factors.
2.2. Production, purification and verification of SeMet-labeled GBBH
SeMet-labeled protein was produced and purified as above ex-
cept that cells were grown in Overnight Express™ Autoinduction
System 2 media (Novagen).
Initial attempts to obtain SeMet-labeled protein in a Met auxo-
trophic E. coli strain failed due to severely reduced protein expres-
sion levels.
Therefore, SeMet labeling was performed in WK6 cells, which
are not Met auxotrophs. Incorporation of SeMet was verified by
MALDI-TOF mass spectrometry of tryptic peptides as described
previously [17], which indicated close to 100% incorporation of
SeMet.
Data scaling, refinement and validation statistics are shown in
Table 1.
2.6. Determination of Zn presence
2.3. Activity determination
Quantitative determination of Zn was done by electrothermal
atomic absorption spectrometry (ETAAS). Measurements were per-
formed on a Perkin–Elmer Model AAnalyst 600 with AC Zeeman-
effect background correction using a modulated 0.8 Tesla magnetic
Activity was determined by measuring the formation of carni-
tine from GBB according to a slightly modified method from Lind-
stedt and Lindstedt [18]. The complete reaction mixture, in a final

636
K. Tars et al. / Biochemical and Biophysical Research Communications 398 (2010) 634–639
Table 1
Recombinant GBBH, like rat GBBH, was sensitive to inhibition
by mildronate. The IC50 value was found to be 62 M, 3–4 times
higher than the IC50 values of 13–18 M observed in rats [29,31].
Data processing and refinement (only for native dataset) statistics.
l
l
Dataset
Native
SeMet peak
SeMet
remote
These results show that the recombinant human GBBH has similar
or even higher activity than purified native human or rat enzymes.
Space group
P6522
P6522
P6522
Unit cell dimensions
a (Å)
b (Å)
164.8
97.8
165.5
98.2
164.8
97.7
3.2. Crystal packing and quality of the model
Wavelength (Å)
Resolution (Å)
Highest resolution bin (Å)
Rmerge
1.00340
40–2.0
0.97856
40–3.5
3.7–3.5
0.20 (0.54)
3.8 (1.5)
100.0 (100.0)
42.8 (43.9)
0.91840
40–3.5
3.7–3.5
0.17 (0.32)
4.2 (2.3)
100.0 (100.0)
13.6 (14.0)
Cell content analysis by the CCP4 [32] program MATTHEWS
[33] predicted two molecules per asymmetric unit with 97% prob-
ability, which agrees with the previous biochemical data that
GBBH exists as a dimer in solution. However, only one protein mol-
ecule per asymmetric unit was found; as a result, the crystals con-
tain about 71% solvent. Interestingly, most of the solvent was
found in very wide (about 100 Å in diameter) cylindrical channels
centered on the 65 screw axis.
2.1–2.0
0.078 (0.47)*
10.5 (1.5)
96.4 (82.3)
3.4 (2.0)
0.21 (0.31)
0.23 (0.36)
30.9
I/
r
Completeness (%)
Average multiplicity
R-factor
Rfree
Average B-factor(Å2)
Average B-factor from
Wilson plot (Å2)
No. protein atoms
No. ligand atoms
No. solvent molecules
29.9
The GBBH monomer (Fig. 2) contains two domains: a double-
strand b-helix (DBSH) catalytic domain (residues 110–385) and a
smaller N-terminal domain (residues 1–97) of unknown function.
2983
26
197
The domains are connected by an
a helix (residues 98–109). The
model was built for all residues except disordered loop regions,
which include residues 188–198, 223–227, 351–359 and two
C-terminal residues. Also, none of the residues of the purifica-
tion/expression tag, except the last serine (assigned residue num-
ber 0) were visible. Side chains of residues Lys72, Lys 221, Gln
222 and Arg 360 were largely disordered and modeled only up to
the Cb atom.
r.m.s deviations from ideal
Bond lengths (Å)
Bond angles (°)
0.010
1.15
Ramachandran
outliers [38] (%)
2.0
*
Values in parentheses are for the highest resolution bin.
field oriented longitudinally to the optical path. For analysis, 20
of 5 M GBBH in 20 mM tris–HCl pH 8.0 were used.
ll
l
3.3. Oligomerization state
The crystal structure of GBBH confirms the previous biochemi-
cal data that GBBH exists as a dimer in solution. Although the crys-
tallographic asymmetric unit contains a single monomer, the two
monomers in a dimer are related by a crystallographic 2-fold axis.
The extensive contact surface between the monomers, about
3600 Å2 of buried surface as determined by the CCP4 program
AREAIMOL [34], strongly suggests that the dimer is stable in solu-
tion and not merely a crystallization artifact. The contact surface in-
volves both domains (Fig. 2). In the dimer, each catalytic domain
forms contacts with both the catalytic and N-terminal domains of
the other monomer. The two catalytic domains contribute to
1350 Å2 of buried surface, whereas the interaction of the N-terminal
domains with the catalytic domains contribute to 1980 Å2 (990 Å2
3. Results and discussion
3.1. Recombinant GBBH activity
The measured Km for GBB was 27.5 6.5
was 330 16 nmol/min per mg protein (Fig. 1). Previous studies
have reported a Km for GBB of 80 M for purified rat GBBH and a
specific activity of 202.5 nmol/min per mg of protein [29]. The
l
M (S.D.), and the Vmax
l
Km value of GBB for purified human GBBH has been found to be
200
lM [30].
Concentrations of GBB above 200
bit GBBH [29], and a similar tendency was observed with recombi-
nant GBBH (Fig. 1).
l
M have been shown to inhi-
400
300
200
100
0
0
100
200
300
400
500
γ-butyrobetaine (μM)
Fig. 1. The activity of recombinant GBBH at various concentrations of c-butyrobetaine. Values are represented as average S.D. (n = 2).

K. Tars et al. / Biochemical and Biophysical Research Communications 398 (2010) 634–639
637
Fig. 3. Active site of GBBH compared to TauD. GBBH residues are shown as thick
stick models with yellow carbons. Residue numbering is according to GBBH. The
Fig. 2. Overall structure of the BBH dimer. Individual monomers are shown in red
and blue with their N-terminal domains in paler versions of the corresponding
colors. Active site residues are shown as stick models with green carbons, and zinc
ions are shown as gray spheres. Figures 2–4 were prepared using Pymol v. 0.99 [37].
(For interpretation of the references to color in this figure legend, the reader is
referred to the web version of this paper.)
corresponding TauD residues and
aKG (not present in GBBH structure) are shown
as thin stick models with black carbons. Fe(II) ion (present only in TauD structure) is
shown as a gray sphere. The side chain of Arg 360 of GBBH is largely disordered and
therefore modeled only to the Cb atom. (For interpretation of the references to color
in this figure legend, the reader is referred to the web version of this paper.)
3.5. N-terminal domain
each pair). The rest of the buried surface (about 270 Å2) is due to the
interactions of the domain-linking helix with the catalytic domain
of the other monomer. Interactions between monomers are largely
polar in nature, although a number of hydrophobic contacts exist as
well.
Residues 1–97 in the GBBH monomer are folded in a separate N-
terminal domain (Fig. 4). The domain resembles a double b-sand-
wich with two three-stranded antiparallel b-sheets facing each
other. The sheets are connected by a long loop, which includes
two a-helices, a short b-hairpin and a zinc-binding site. The metal
Several other 2-ketoglutarate oxygenases exist in oligomeric
states [4]. However, the interfaces between monomers of previ-
ously described oxygenase multimers appear to be very different
from that of GBBH.
binding site is formed by residues Cys 38, Cys 40, Cys 43 from the
loop and His 82 from the b-sheet. The presence of a metal atom in
the middle of the site is confirmed by a very high electron density
peak. The characteristic tetrahedral coordination geometry and
nature of the interacting protein side chains suggest that the bound
metal is zinc. The zinc determination by ETAAS indicate that pro-
tein and zinc are in roughly equimolar proportions (data not
shown). The interactions with the metal seem to be rather strong
because no zinc or any other similar metal was added to the crys-
tallization mix or during expression or purification. Metallopro-
teins are known to strip nickel ions from chelating agarose
3.4. Catalytic domain and active site
The overall structure of the catalytic DBSH domain of GBBH is
very similar to that of other 2KG oxygenases. According to results
from the DALI server [35], taurine dioxygenase (TauD, pdb code
1GY9, [36]) has the most similar 3D structure (r.m.s.d. of Ca atoms
2.9 Å with 210 residues aligned out of 270, 15% sequence identity).
The active site of GBBH compared to that of TauD is shown in Fig. 3.
Active site residues His 202, Asp 204 and His 347 are located essen-
tially in the same positions in both cases. However, no bound li-
gands or Fe(II) ion could be located in the electron density maps
of our structure, even when added to the crystallization mixture.
Presumably, the low pH and high salt content (pH 4.5, 1.5 M
(NH4)2SO4) in the crystallization mix prevented the ligands from
binding to the active site. In 2KG oxygenases, 2KG is coordinated
by an Fe(II) ion and by a number of polar interactions with protein
side chains. Notably, in all 2KG oxygenases with known structures,
the carboxyl group of the 2KG side-chain forms a salt-bridge with a
nearby arginine or lysine residue [4]. A similar interaction may oc-
cur in GBBH as well. Arg 361 of GBBH is located essentially in the
same place as the carboxylate-interacting Arg 266 of TauD.
Although the side chain of Arg 361 is disordered beyond the Cb
atom in our structure, it might adopt a fixed conformation upon
interaction with 2KG.
Specific substrate-interacting residues in various 2KG oxygen-
ases are very different, reflecting their ability to bind distinct com-
pounds. Additionally, binding of a substrate might cause structural
rearrangements in the active site. For example, loops that are dis-
ordered in our structure might become ordered upon substrate
binding. Therefore, we did not attempt to dock GBB or mildronate
in the active site.
Fig. 4. Zinc-binding site in the N-terminal domain of G-BBH. The zinc atom is
shown as a gray sphere and coordinating residues are labeled. The domain is
rainbow- colored from the N-terminus in blue to the C-terminus in red. (For
interpretation of the references to color in this figure legend, the reader is referred
to the web version of this paper.)

638
K. Tars et al. / Biochemical and Biophysical Research Communications 398 (2010) 634–639
[2] D.W. Foster, The role of the carnitine system in human metabolism, Ann. NY
Acad. Sci. 1033 (2004) 1–16.
during purification; however, they have a different coordination
geometry than observed in our case.
[3] R.R. Ramsay, R.D. Gandour, F.R. Van der Leij, Molecular enzymology of
carnitine transfer and transport, Biochim. Biophys. Acta 1546 (2001) 21–43.
[4] I.J. Clifton, M.A. McDonough, D. Ehrismann, N.J. Kershaw, N. Granatino, C.J.
Schofield, Structural studies on 2-oxoglutarate oxygenases and related double-
stranded beta-helix fold proteins, J. Inorg. Biochem. 100 (2006) 644–669.
[5] L.J. Higgins, F. Yan, P. Liu, H.W. Liu, C.L. Drennan, Structural insight into
antibiotic fosfomycin biosynthesis by a mononuclear iron enzyme, Nature 437
(2005) 838–844.
[6] I.J. Clifton, L.C. Hsueh, J.E. Baldwin, K. Harlos, C.J. Schofield, Structure of proline
3-hydroxylase. Evolution of the family of 2-oxoglutarate dependent
oxygenases, Eur. J. Biochem. 268 (2001) 6625–6636.
[7] S. Lindstedt, I. Nordin, Multiple forms of gamma-butyrobetaine hydroxylase
(EC 1.14.11.1), Biochem. J. 223 (1984) 119–127.
[8] C. Rigault, F. Le Borgne, J. Demarquoy, Genomic structure, alternative
maturation and tissue expression of the human BBOX1 gene, Biochim.
Biophys. Acta 1761 (2006) 1469–1481.
[9] F.M. Vaz, S. van Gool, R. Ofman, L. Ijlst, R.J. Wanders, Carnitine biosynthesis:
identification of the cDNA encoding human gamma-butyrobetaine
hydroxylase, Biochem. Biophys. Res. Commun. 250 (1998) 506–510.
[10] Y. Higashi, K. Yokogawa, N. Takeuchi, I. Tamai, M. Nomura, N. Hashimoto, J.I.
Hayakawa, K.I. Miyamoto, A. Tsuji, Effect of gamma-butyrobetaine on fatty
liver in juvenile visceral steatosis mice, J. Pharm. Pharmacol. 53 (2001)
527–533.
[11] C.J. Rebouche, A.G. Engel, Tissue distribution of carnitine biosynthetic enzymes
in man, Biochim. Biophys. Acta 630 (1980) 22–29.
[12] F.M. Vaz, R.J. Wanders, Carnitine biosynthesis in mammals, Biochem. J. 361
(2002) 417–429.
[13] M. Dambrova, E. Liepinsh, I. Kalvinsh, Mildronate: cardioprotective action
through carnitine-lowering effect, Trends Cardiovasc. Med. 12 (2002)
275–279.
[14] E. Liepinsh, R. Vilskersts, L. Zvejniece, B. Svalbe, E. Skapare, J. Kuka, H. Cirule, S.
Grinberga, I. Kalvinsh, M. Dambrova, Protective effects of mildronate in an
experimental model of type 2 diabetes in Goto-Kakizaki rats, Br. J. Pharmacol.
157 (2009) 1549–1556.
To our knowledge, a separate, non-catalytic metal-binding do-
main has never been observed in 2KG oxygenases. Because the
N-terminal zinc-binding domain of GBBH does not appear to have
any assigned function in the literature, we performed a search in
the DALI server to identify proteins with a similar fold. The domain
appears to be related to the DUF971 superfamily, which consists of
several short bacterial proteins with unknown functions. Some of
the DUF971 proteins have conserved cysteines in the same posi-
tions as in the GBBH zinc-binding motif. In the Protein Data Bank,
the only structure assigned to the DUF971 family is the entry
3LUU, which contains coordinates of a protein of unknown func-
tion from Acidithiobacillus ferroxidans. 3LUU has a very similar fold
and 18% sequence identity to the GBBH N-terminal domain, and
the Ca atom r.m.s.d. is 2.3 Å, with 83 residues aligned out of 89.
Interestingly, 3LUU also has a bound zinc ion but in a completely
different position and involving different residues compared to
GBBH. The zinc binding residues in GBBH are partially conserved
in 3LUU; one cysteine is replaced with an isoleucine, which seems
to be enough to prevent the coordination of a zinc ion. Apparently,
the zinc ion has an important role in stabilizing the b-sheet-inter-
connecting loop because it is ordered in GBBH but largely disor-
dered (unmodelled) in 3LUU. The closest relative to 3LUU (46%
sequence identity) with a known function is phosphoribosylformi-
mino-5-aminoimidazole carboxamide ribotide isomerase from
Ralstonia solanacearum. It remains to be investigated whether the
actual function of the GBBH N-terminal domain is related to the
above-mentioned proteins.
The GBBH residues involved in zinc binding are evolutionary
conserved in all sequenced eukaryotes more complex than Caeno-
rhabditis elegans. The zinc-binding motif is also conserved in some
yeast and bacterial enzymes, for example, GBBHs from Candida
albicans and Pseudomonas sp. A related enzyme, trimethyllysine
hydroxylase (EC 1.14.11.8) has a partially conserved zinc-binding
site, where one cysteine has been replaced by a serine.
[15] E. Liepinsh, R. Vilskersts, E. Skapare, B. Svalbe, J. Kuka, H. Cirule, O. Pugovics,
I. Kalvinsh, M. Dambrova, Mildronate decreases carnitine availability and up-
regulates glucose uptake and related gene expression in the mouse heart, Life
Sci. 83 (2008) 613–619.
[16] R. Zell, H.J. Fritz, DNA mismatch-repair in Escherichia coli counteracting the
hydrolytic deamination of 5-methyl-cytosine residues, EMBO J.
1809–1815.
6 (1987)
[17] P. Plevka, A. Kazaks, T. Voronkova, S. Kotelovica, A. Dishlers, L. Liljas, K. Tars,
The structure of bacteriophage phiCb5 reveals a role of the RNA genome
and metal ions in particle stability and assembly, J. Mol. Biol. 391 (2009) 635–
647.
[18] G. Lindstedt, S. Lindstedt, Cofactor requirements of gamma-butyrobetaine
hydroxylase from rat liver, J. Biol. Chem. 245 (1970) 4178–4186.
[19] E. Liepinsh, J. Kuka, B. Svalbe, R. Vilskersts, E. Skapare, H. Cirule, O. Pugovics,
I. Kalvinsh, M. Dambrova, Effects of long-term mildronate treatment on
cardiac and liver functions in rats, Basic Clin. Pharmacol. Toxicol. 105 (2009)
387–394.
[20] A.G.W. Leslie, Recent changes to the MOSFLM package for processing film and
image plate data. Joint CCP4 + ESF-EAMCB Newsletter on Protein
Crystallography 26 (1992).
[21] P.R. Evans, Scala. Joint CCP4 + ESF-EAMCB Newsletter on Protein
Crystallography 33 (1997) 22-24.
4. Conclusions
The crystal structure of GBBH apo form has been solved at 2.0 Å
resolution. The GBBH macromolecule contains two identical
monomers with an extensive dimerization surface. The structure
of the catalytic domain of GBBH is similar to other related enzymes
and possesses conserved features, such as Fe(II) and 2KG binding
residues. The fold of the N-terminal domain is novel to 2KG oxy-
genases and contains a zinc-binding site.
[22] G.M. Sheldrick,
112–122.
A short history of SHELX, Acta Crystallogr. A 64 (2008)
[23] G. Bricogne, C. Vonrhein, C. Flensburg, M. Schiltz, W. Paciorek, Generation,
representation and flow of phase information in structure determination:
recent developments in and around SHARP 2.0, Acta Crystallogr.
Crystallogr. 59 (2003) 2023–2030.
D Biol.
Accession numbers
[24] J.P. Abrahams, A.G. Leslie, Methods used in the structure determination of
The coordinates and the structure factors have been deposited
in the Protein Data Bank with the accession number 3N6 W.
bovine mitochondrial F1 ATPase, Acta Crystallogr.
(1996) 30–42.
D Biol. Crystallogr. 52
[25] K. Cowtan, The Buccaneer software for automated model building. 1 Tracing
protein chains, Acta Crystallogr. D Biol. Crystallogr. 62 (2006) 1002–1011.
[26] P. Emsley, K. Cowtan, Coot: model-building tools for molecular graphics, Acta
Crystallogr. D Biol. Crystallogr. 60 (2004) 2126–2132.
Acknowledgments
[27] T.A. Jones, M. Bergdoll, M. Kjeldgaard, O:
A macromolecule modeling
This work was supported by the Latvian State Research Program
4VPP-2010-6/3.3 and by ESF grant 1DP/1.1.1.2.0/09/APIA/VIAA/
150. We thank Anna Janson and Ana-Laura Stern for their assis-
tance in data collection. We thank the personnel for their help dur-
ing our stay at ESRF, in particular Hassan Belrhali for his assistance
in quickly obtaining the first SAD map. We thank also Prof. Lars Lil-
jas for critical reading of this manuscript.
environment, in: C. Bugg, S. Ealick (Eds.), Crystallographic and Modeling
Methods in Molecular Design, Springer-Verlag, New York, 1990, pp. 189–
199.
[28] G.N. Murshudov, A.A. Vagin, E.J. Dodson, Refinement of macromolecular
structures by the maximum-likelihood method, Acta Crystallogr.
Crystallogr. 53 (1997) 240–255.
[29] S. Galland, F. Le Borgne, D. Guyonnet, P. Clouet, J. Demarquoy, Purification and
characterization of the rat liver gamma-butyrobetaine hydroxylase, Mol. Cell
Biochem. 178 (1998) 163–168.
D Biol.
[30] G. Lindstedt, S. Lindstedt, I. Nordin, Gamma-butyrobetaine hydroxylase in
human kidney, Scand. J. Clin. Lab. Invest. 42 (1982) 477–485.
[31] M. Spaniol, H. Brooks, L. Auer, A. Zimmermann, M. Solioz, B. Stieger,
S. Krahenbuhl, Development and characterization of an animal model of
carnitine deficiency, Eur. J. Biochem. 268 (2001) 1876–1887.
References
[1] J. Bremer, Carnitine–metabolism and functions, Physiol. Rev. 63 (1983)
1420–1480.

K. Tars et al. / Biochemical and Biophysical Research Communications 398 (2010) 634–639
639
[32] N. Collaborative Computational Project, The CCP4 suite: Programs for protein
crystallography, Acta Cryst. D50 (1994) 760–763.
[33] K.A. Kantardjieff, B. Rupp, Matthews coefficient probabilities: improved
estimates for unit cell contents of proteins, DNA, and protein–nucleic acid
complex crystals, Protein Sci. 12 (2003) 1865–1871.
[34] P.J. Briggs, Further improvements to AREAIMOL code. CCP4 Newsletter on
Protein Crystallography 38 (2000).
[35] L. Holm, S. Kaariainen, P. Rosenstrom, A. Schenkel, Searching protein structure
databases with DaliLite v.3, Bioinformatics 24 (2008) 2780–2781.
[36] J.M. Elkins, M.J. Ryle, I.J. Clifton, J.C. Dunning Hotopp, J.S. Lloyd, N.I. Burzlaff,
J.E. Baldwin, R.P. Hausinger, P.L. Roach, X-ray crystal structure of Escherichia
coli taurine/alpha-ketoglutarate dioxygenase complexed to ferrous iron and
substrates, Biochemistry 41 (2002) 5185–5192.
[37] W.L. DeLano, The PyMOL Molecular Graphics System, DeLano Scientific, San
Carlos, CA, USA, 2002.
[38] G.J. Kleywegt, T.A. Jones, Phi/psi-chology: Ramachandran revisited, Structure 4
(1996) 1395–1400.