Substrate specificity determinants of the methanogen homoaconitase enzyme: Structure and function of the small subunit

Article abstract of DOI:10.1021/bi901766z

The aconitase family of hydro-lyase enzymes includes three classes of proteins that catalyze the isomerization of a-hydroxy acids to β-hydroxy acids. Besides aconitase, isopropylmalate isomerase (IPMI) proteins specifically catalyze the isomerization of α,β-dicarboxylates with hydrophobic y-chain groups, and homoaconitase (HACN) proteins catalyze the isomerization of tricarboxylates with variable chain length y-carboxylate groups. These enzymes' stereospecific hydro-lyase activities make them attractive catalysts to produce diastereomers from unsaturated precursors. However, sequence similarity and convergent evolution among these proteins lead to widespread misannotation and uncertainty about gene function. To find the substrate specificity determinants of homologous IPMI and HACN proteins from Methanocaldococcus jannaschii, the small-subunit HACN protein (MJ1271 ) was crystallized for X-ray diffraction. The structural model showed characteristic residues in a flexible loop region between α2 and α3 that distinguish HACN from IPMI and aconitase proteins. Site-directed mutagenesis of MJ1271 produced loop-region variant proteins that were reconstituted with wild-type MJ1003 large-subunit protein. The heteromers formed promiscuous hydro-lyases with reduced activity but broader substrate specificity. Both R26K and R26V variants formed relatively efficient IPMI enzymes, while the T27A variant had uniformly lower specificity constants for both IPMI and HACN substrates. The R26V T27Y variant resembles the MJ1277IPMI small subunit in its flexible loop sequence but demonstrated the broad substrate specificity of the R26V variant. These mutations may reverse the evolution of HACN activity from an ancestral IPMI gene, demonstrating the evolutionary potential for promiscuity in hydro-lyase enzymes. Understanding these specificity determinants enables the functional reannotation of paralogous HACN and IPMI genes in numerous genome sequences. These structural and kinetic results will help to engineer new stereospecific hydro-lyase enzymes for chemoenzymatic syntheses.

Full text of DOI:10.1021/bi901766z

Biochemistry 2010, 49, 2687–2696 2687
DOI: 10.1021/bi901766z
Substrate Specificity Determinants of the Methanogen Homoaconitase Enzyme: Structure
and Function of the Small Subunit†,‡
Jeyaraman Jeyakanthan,§,1 Randy M. Drevland, ,1 Dasara Raju Gayathri,^ Devadasan Velmurugan,^ Akeo Shinkai,#
Seiki Kuramitsu,#,4 Shigeyuki Yokoyama,2,) and David E. Graham*, ,(,3
§Life Science Group, National Synchrotron Radiation Research Center, 101 Hsin-Ann Road, Hsinchu Science Park, Hsinch 30076,
Taiwan, Chemistry and Biochemistry Department, The University of Texas at Austin, Austin, Texas 78712, ^Centre of Advanced
Study in Crystallography and Biophysics, University of Madras, Guindy Campus, Chennai 600 025, India, #RIKEN SPring-8 Center,
Harima Institute, 1-1-1 Kouto, Sayo, Hyogo 679-5148, Japan, 4Graduate School of Science, Osaka University, Toyonaka,
Osaka 560-0043, Japan, 2RIKEN Systems and Structural Biology Center, Yokohama Institute, 1-7-22 Suehiro-cho, Tsurumi,
Yokohama 230-0045, Japan, )Graduate School of Science, the University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan,
(Department of Microbiology, University of Tennessee, Knoxville, Tennessee 37996, and 3Biosciences Division, Oak Ridge National
Laboratory, Oak Ridge, Tennessee 37831. 1These two authors contributed equally to this paper.
Received October 13, 2009; Revised Manuscript Received February 1, 2010
ABSTRACT: The aconitase family of hydro-lyase enzymes includes three classes of proteins that catalyze the
isomerization of R-hydroxy acids to β-hydroxy acids. Besides aconitase, isopropylmalate isomerase (IPMI) proteins
specifically catalyze the isomerization of R,β-dicarboxylates with hydrophobic γ-chain groups, and homoaconitase
(HACN) proteins catalyze the isomerization of tricarboxylates with variable chain length γ-carboxylate groups.
These enzymes’ stereospecific hydro-lyase activities make them attractive catalysts to produce diastereomers from
unsaturated precursors. However, sequence similarity and convergent evolution among these proteins lead to
widespread misannotation and uncertainty about gene function. To find the substrate specificity determinants of
homologous IPMI and HACN proteins from Methanocaldococcus jannaschii, the small-subunit HACN protein
(MJ1271) was crystallized for X-ray diffraction. The structural model showed characteristic residues in a flexible loop
region between R2 and R3 that distinguish HACN from IPMI and aconitase proteins. Site-directed mutagenesis of
MJ1271 produced loop-region variant proteins that were reconstituted with wild-type MJ1003 large-subunit protein.
The heteromers formed promiscuous hydro-lyases with reduced activity but broader substrate specificity. Both
R26K and R26V variants formed relatively efficient IPMI enzymes, while the T27A variant had uniformly lower
specificity constants for both IPMI and HACN substrates. The R26V T27Y variant resembles the MJ1277 IPMI
small subunit in its flexible loop sequence but demonstrated the broad substrate specificity of the R26V variant. These
mutations may reverse the evolution of HACN activity from an ancestral IPMI gene, demonstrating the
evolutionary potential for promiscuity in hydro-lyase enzymes. Understanding these specificity determinants enables
the functional reannotation of paralogous HACN and IPMI genes in numerous genome sequences. These structural
and kinetic results will help to engineer new stereospecific hydro-lyase enzymes for chemoenzymatic syntheses.
The methanogen homoaconitase (HACN)1 enzyme catalyzes
the isomerization of (R)-homocitrate to (2R,3S)-homoisocitrate
in the biosynthesis of coenzyme B (Scheme 1) (1). This enzyme
evolved from an isopropylmalate isomerase (IPMI) ancestor, a
ubiquitous enzyme required for leucine biosynthesis. HACN and
IPMI, together with aconitase (ACN; citric acid cycle), homoaco-
nitate hydratase (R-aminoadipate pathway for lysine biosyn-
thesis), 2,3-dimethylmalate dehydratase (nicotinate catabolism),
and 2-methylcitrate dehydratase, belong to the aconitase family
of [4Fe-4S]-dependent hydro-lyases (2-7). Only three members
of this family, HACN, IPMI, and ACN, catalyze sequential
dehydration and hydration reactions, creating new isomers.
Despite extensive structural and mechanistic characterization
of ACN proteins, the substrate specificity determinants of IPMI
and HACN enzymes were unknown (8, 9). As a result, sequence
databases often list incorrect functional annotations for these
proteins, hindering metabolic reconstruction projects. Efforts to
design new hydro-lyase enzymes for the chemoenzymatic synth-
esis of diastereomeric acids will require understanding the
specificity determinants reported here.
†The present work was supported by the RIKEN Structural Geno-
mic/Proteomics Initiative (RSGI), the National Project on Protein
Structural and Functional Analyses, Ministry of Education, Culture,
Sports, Science, and Technology of Japan, National Science Founda-
tion Grant MCB-0817903, and U.S. Department of Energy, Office of
Science Biological and Environmental Research. R.M.D. was sup-
ported in part by a Faraday Fellowship. Oak Ridge National Labora-
tory is managed by UT-Battelle, LLC, for the U.S. Department of
Energy under Contract DE-AC05-00OR22725.
‡The structural model of the MJ1271 protein has been deposited in the
RCSB Protein Data Bank with identifier 2PKP.
*Address correspondence to this author. Tel: 865-574-0559. Fax: 865-
576-8646. E-mail: grahamde@ornl.gov.
1Abbreviations: HACN, homoaconitase; IPMI, isopropylmalate iso-
merase; ACN, aconitase; mACN, mitochondrial aconitase; SOE-PCR,
splicing overlap extension polymerase chain reaction; DTT, dithiothreitol;
CHES, 2-(cyclohexylamino)ethanesulfonic acid; MES, 2-(N-morpholi-
no)ethanesulfonic acid; PEG, polyethylene glycol; HICDH, homoisoci-
trate dehydrogenase; IPMDH, isopropylmalate dehydrogenase.
In the leucine biosynthetic pathway, IPMI catalyzes the rever-
sible dehydration of R-isopropylmalate to cis-dimethylcitraco-
nate and the subsequent trans addition of water, yielding (2R,3S)-
3-isopropylmalate (β-isopropylmalate) (10). IPMI also catalyzes
r
2010 American Chemical Society
Published on Web 02/19/2010
pubs.acs.org/Biochemistry

2688 Biochemistry, Vol. 49, No. 12, 2010
Jeyakanthan et al.
Scheme 1: Hydro-lyase Enzymes Catalyze the Isomerization of
residues in mACN are highly conserved, implying similar struc-
tural organization and catalytic mechanisms. Structural and
mutational analysis of mACN implicated Arg580 from a flexible
loop of domain 4 as the key residue responsible for recognizing
citrate, cis-aconitate, and isocitrate in the aconitase mechanism
(14, 15). The guanidinium nitrogen atoms of Arg580 hydrogen
bond to the substrate’s γ-carboxylate group. Replacing Arg580
with lysine resulted in a 30-fold increase in KM with isocitrate (15).
However, this arginine residue is not conserved in the archaeal
HACN or IPMI sequences. The crystal structure of a putative
bifunctional HACN/IPMI small subunit from the euryarchaeon
Pyrococcus horikoshii shows that the loop region between R-helices
1 and 2 corresponds to the flexible loop region of mACN (16).
Complementation analysis demonstrated the homoaconitase (or
homoaconitate hydratase) function of the P. horikoshii pro-
tein (17), while metabolic reconstruction and phylogenetic ana-
lysis made the IPMI function doubtful (11). An alignment of the
mACN and P. horikoshii HACN sequences suggested thatArg580
is replaced with a polar threonine residue in the loop region (16),
inconsistent with the mACN substrate-binding model (18). With-
out biochemical characterization, both the function and mecha-
nism of this P. horikoshii protein are unresolved.
The similarity between small subunits of the M. jannaschii
proteins IPMIMj (MJ1277) and HACNMj (MJ1271) underscores
the importance of defining the consensus sequence responsible
for substrate recognition. These two protein sequences are 53%
identical, and both were previously annotated as IPMI subunits
based on primary sequence alone, only to be distinguished by
heterologous expression, purification, reconstitution, and in vitro
kinetic analysis (1, 11). Sequences of the predicted flexible loop
regions of these enzymes, identified by homology models and
sequence alignments, may be the best indicators of substrate
specificity. The HACNMj small subunit has a consensus sequence
of Y24LRT while the homologous IPMIMj has the sequence
Y26LVY. The polar Arg26 and Thr27 residues of HACNMj could
potentially form hydrogen bonds with the γ-carboxylates of cis-
homoaconitate, cis-homo2-aconitate, and cis-homo3-aconitate.
The corresponding hydrophobic Val28 and Tyr29 residues of the
IPMIMj loop region may be structural determinants for recogniz-
ing hydrophobic γ-chains. These consensus sequences are con-
served in most of the Euryarchaeota.
R- and β-Hydroxy Acidsa
aR-group substituents are shown below the reaction. Malease
(EC 4.2.1.31) catalyzes the hydration of maleate (2a) to form
D-malate (3a). The isopropylmalate isomerase (IPMI, EC
4.2.1.33) catalyzes the dehydration of 2-isopropylmalate (1c),
forming isopropylmaleate (2c) that is hydrated to produce 3-iso-
propylmalate (3c). This enzyme can also catalyze the dehydra-
tion of citramalate (1b) and the subsequent hydration of citra-
conate (2b) to produce 2-methylmalate (3b). The mitochondrial
aconitase (mACN, EC 4.2.1.3) catalyzes the dehydration of
citrate (1d) to produce cis-aconitate (2d) that is hydrated to form
isocitrate (3d). The homoaconitate hydratase enzyme (EC
4.2.1.36) catalyzes only the hydration of cis-homoaconitate
(2e)toformhomoisocitrate(3e). Finally, themethanogenhomo-
aconitase (HACN, EC 4.2.1.114) catalyzes the dehydration of
(homo)1-3-citrate (1e,f,g) and the hydration of cis-(homo)1-3
aconitate (2e,f,g) to form (homo)1-3-isocitrate (3e,f,g).
-
the isomerization of (R)-citramalate to β-methylmalate in the
pyruvate pathway for isoleucine biosynthesis that replaces threo-
nine dehydratase in some microbes (11). In the R-aminoadipate
pathways of Thermus thermophilus, Saccharomyces cerevisiae,
and Aspergillus nidulans (12), homoaconitate hydratase catalyzes
only the second half-reaction in the isomerization of (R)-homo-
citrate: the hydration of cis-homoaconitate to (2R,3S)-homo-
isocitrate (3, 4, 13). A second enzyme, predicted to be ACN, was
proposed to catalyze the initial dehydration of (R)-homocitrate
to cis-homoaconitate (4). It is therefore remarkable that the
In this work, we present the first structural study of a
characterized homoaconitase. The HACNMj small subunit pro-
˚
tein, MJ1271, was crystallized and refined to 2.1 A. Site-directed
mutagenesis of the proposed flexible loop region of MJ1271,
combined with wild-type MJ1003 large subunit protein, revealed
residues affecting substrate specificity. Replacing residues Arg26
and Thr27 in the Y24LRT loop region of MJ1271 with corre-
sponding residues from the IPMIMj small subunit created a
promiscuous enzyme with both HACN and IPMI activities.
homoaconitase from Methanocaldococcus jannaschii (HACNMj
)
catalyzes the full conversion of (R)-homocitrate to homoisoci-
trate, as well as the full isomerizations of (R)-homo2-citrate and
(R)-homo3-citrate, in the chain elongation reactions of coenzyme
B biosynthesis (1).
We propose that Arg26 is the equivalent of mACN Arg580
,
The HACN and IPMI homologues are monomeric (fungi) or
heterodimeric (bacteria and archaea) iron-sulfur cluster proteins,
similar to the well-characterized monomeric mitochondrial aconi-
tase (mACN). In M. jannaschii, the MJ1271 and MJ1003 proteins
comprise HACNMj, and the paralogous MJ1277 and MJ0499 pro-
teins comprise IPMIMj. The large subunits of the archaeal proteins
(MJ1003 or MJ0499, ∼48 kDa) are homologous to domains 1, 2,
and 3 of mACN, including the CX63CXXC motif that provides
ligands for the [4Fe-4S] cluster. These large subunits interact with
small subunits (MJ1271 or MJ1277, ∼18 kDa), homologous to
domain 4 of mACN. Sequence alignments of archaeal HACN and
IPMI homologues indicate that the majority of the 21 active site
coordinating the γ-carboxylate of the HACN substrates. The
results provide the first structure-function analysis of HACN
substrate specificity, which will be used to correct the functional
annotations of uncharacterized aconitase-like proteins and de-
sign new hydro-lyase enzymes for biocatalysis.
EXPERIMENTAL PROCEDURES
The Escherichia coli BL21(DE3) strains with plasmids pDG141
(expressing MJ1003) and pDG163 (expressing MJ1277) were
described previously (Table 1) (11). Splicing overlap extension

Article
Biochemistry, Vol. 49, No. 12, 2010 2689
Plasmid pDG160, encoding the wild-type MJ1271 gene, was
mutated using the QuikChange II site-directed mutagenesis kit
(Stratagene) and the mutagenic primers listed in Table 2 (together
with their reverse complements). The resulting mutations in the
plasmids listed in Table 1 were confirmed by DNA sequencing.
Plasmid pRD03 was used as a template to generate the double
mutation in pRD17 by the same strategy. Electroporation trans-
formed E. coli BL21(DE3) pDG141 cells with the new plasmids
encoding small subunit proteins.
To express proteins for kinetic studies, E. coli strains were
grown in LB medium, supplemented with 100 μg mL-1 ampicillin
and 50 μg mL-1 streptomycin, at 37 ꢀC with shaking at 250 rpm
until the culture reached an optical density at 600 nm of 0.6-0.8.
Expression was then induced by the addition of 50 μM isopropyl
Table 1: List of Plasmids and Microorganisms
strain or plasmid
(parent plasmid)
description
source or ref
E. coli
BL21(DE3)
protein expression host
expression host with
additional tRNAs
Novagen
BL21 CodonPlus-RIL
Stratagene
XL-1 Blue
general cloning host
Stratagene
plasmids
pMJ1271 (pET-21a)
pDG141 (pCDF-Duet1)
pDG160 (pET-19b)
pDG163 (pET-19b)
pDG476 (pET-19b)
pDG625 (pET-19b)
pRD03 (pDG160)
pRD06 (pDG160)
pRD09 (pDG160)
pRD17 (pRD03)
MJ1271
this work
11
MJ1003
MJ1271
11
MJ1277
11
MJ1271-MJ1277 chimera
MJ1271-LysU
MJ1271 R26V
MJ1271 R26K
MJ1271 T27A
MJ1271 R26V T27Y
this work
this work
this work
this work
this work
this work
1-β-D-galactopyranoside, and the culture was then incubated for
an additional 3-4 h. The wild-type MJ1003 and MJ1271 variant
proteins were purified by heating the cell lysate, followed by
anion-exchange chromatography, dialysis, and concentration
by centrifugal ultrafiltration (1). Protein purity was determined
by SDS-PAGE with Coomassie Blue dye staining. Interactions
between MJ1003 and MJ1271 mutants were tested by measuring
apparent masses of protein complexes using analytical size
exclusion chromatography (20).
Table 2: Oligonucleotide Primers Used To Construct MJ1271 Mutations
primer name
sequence (50 to 30)
Copurified apoenzymes containing wild-type MJ1003 and
variant MJ1271 proteins (1-2 mg mL-1) were reconstituted in
the presence of dithiothreitol (DTT), Fe(NH4)2(SO4)2, and Na2S
as described previously (1). A mock reconstitution solution
consisted of the above mixture without protein. Enzymatic
activity was measured using previously described continuous
spectrophotometric assays (1), where a decrease in UV absor-
bance by cis-unsaturated substrates was measured under anoxic
conditions. Reactions (1 mL) were conducted in quartz semi-
microcells with screw cap septa (Starna) containing 50 mM
2-(cyclohexylamino)ethanesulfonic acid (CHES)-KOH (pH 9.0),
200 mM KCl, 10-100 μg mL-1 holoenzyme, and various sub-
strate concentrations.
For protein crystallization, E. coli BL21 CodonPlus (DE3)-
RIL (pMJ1271) cells were grown at 37 ꢀC overnight in a medium
containing 1.0% polypeptone, 0.5% yeast extract, 0.5% NaCl,
and 100 μg mL-1 ampicillin (pH 7.0). The cells were lysed by
sonication in 20 mM Tris-HCl (pH 8.0) containing 500 mM
NaCl, 5 mM β-mercaptoethanol, and 1 mM phenylmethanesul-
fonyl fluoride. The lysate was incubated at 70 ꢀC for 10 min and
centrifuged at 15000 rpm for 30 min at 4 ꢀC. After exchanging the
buffer with 20 mM Tris-HCl (pH 8.0), the protein sample was
loaded onto a TOYOPEARL SuperQ-650 M column (Tosoh)
preequilibrated with 20 mM Tris-HCl (pH 8.0). The flow-
through, containing most MJ1271 protein, was subjected to a
buffer exchange with 20 mM 2-(N-morpholino)ethanesulfonic
acid (MES)-NaOH (pH 6.0) and loaded onto a RESOURCE S
column (GE Healthcare Bio-Sciences) preequilibrated with the
same buffer. The protein was bound to the column and eluted in a
linear gradient from 0 to 400 mM NaCl. Fractions containing
MJ1271 protein were subjected to a buffer exchange with 10 mM
potassium phosphate buffer (pH 7.0) and loaded on a Bio-Scale
CHT20-I column (Bio-Rad) preequilibrated with the same
buffer. The protein was bound to the column and eluted in
a linear gradient from 10 to 500 mM potassium phosphate
(pH 7.0). The MJ1271 protein was subjected to gel filtration on
a HiLoad 16/60 Superdex 200 pg column (GE Healthcare Bio-
Sciences) preequilibrated with 20 mM Tris-HCl and 200 mM
NaCl (pH 8.0). The fraction containing purified protein was
T7
TAATACGACTCACTATAGGG
GCTAACTCGTAAGGGTCTGT-
ATAAACTAAATACCTTGCTGG
CCAGCAAGGTATTTAGTTTA-
TACAGACCCTTACGAGTTAGC
GCTAGTTATTGCTCAGCGG
CCAACCATGAACGGAGCGTA-
TTTTCCTGGAATTATTGCGTC
CGCTCCGTTCATGGTTGGTGA-
ATACGAGTTAGCTTCACACTG
CCAGGACCTTACTTAGTGACTA-
CAGACCCTTACGAG
MJ1277Rev-overlap
MJ1277Fwd-overlap
T7-terminator
3MJ1271-LysU1
5MJ1271-LysU1
R26Va
R26Ka
CCAGGACCTTACTTAAAGAC-
TACAGACCCTTACGAG
T27Aa
CCAGGACCTTACTTAAGGGCTA-
CAGACCCTTACGAG
R26V/T27Ya
CCAGGACCTTACTTAGTGTATA-
CAGACCCTTACGAG
aMutations are underlined in these sequences.
polymerase chain reaction (SOE-PCR) was used to construct two
chimeric genes (19). The MJ1271-MJ1277 chimeric gene contained
codons 1-29 from MJ1277 fused to codons 28-170 from MJ1271.
T7 promoter and MJ1271Rev-overlap primers (Table 2) were
used to amplify the 50-fragment of this chimeric gene from
pDG163, and MJ1271Fwd-overlap and T7 terminator primers
were used to amplify the 30-fragment from pDG160. The purified
fragments were joined by SOE-PCR. The chimeric product was
purified and digested with NcoI and BamHI restriction enzymes
and then ligated into the same sites of vector pET-19b to create
vector pDG476. The MJ1271-LysU chimera replaced codons
23-30 of MJ1271 with codons 18-23 from the T. thermophilus
lysU gene (4). T7 promoter and 3MJ1271-LysU1 primers were
used to amplify the 50-fragment from pDG163, and 5MJ1271-
LysU and T7-terminator primers were used to amplify the
30-fragment from pDG163. SOE-PCR produced a full-length chi-
meric gene that was ligated into the NcoI and BamHI sites of pET-
19b to form vector pDG476. For crystallography experiments, the
MJ1271 gene was amplified from M. jannaschii chromosomal
DNA and ligated into vector pET-21a (Novagen) to produce
plasmid pMJ1271.

2690 Biochemistry, Vol. 49, No. 12, 2010
Jeyakanthan et al.
concentrated using a VIVASPIN 10000 molecular weight cutoff
ultrafiltration device (Sartorius). The final preparation was
dissolved in 20 mM Tris-HCl buffer containing 200 mM NaCl
Table 3: Data Collection and Refinement Statistics for the MJ1271
Structure
and 1 mM DTT (pH 8.0), at a concentration of 18 mg mL-1
.
native
Initial crystallization conditions were screened using the
TERA (automatic crystallization) system (21) from 144 condi-
tions. The crystallization of MJ1271 protein was performed using
the micro batch method by setting up drops consisting of 1 μL of
protein solution and 1 μL of reservoir solution at 295 K. The
reservoir solution consisted of 50% (w/v) polyethylene glycol
(PEG) 200 and 0.1 M Tris-HCl, pH 4.6. The crystals were
transferred to mother liquor containing 20% (v/v) glycerol
cryoprotectant.
data collection
X-ray source
BL26B2, SPring-8
1.0
˚
wavelength (A)
detector
Jupiter 210cs CCD
100
temperature (K)
distance (mm)
space group
170
C2221
˚
unit cell parameters (A)
a = 90.9, b = 108.3,
c = 45.4
˚
resolution range (A)
50-2.1 (2.18-2.10)
90990
The crystals were cooled in an N2 gas stream at 100 K using the
SPring-8 precise automatic cryosample exchanger (SPACE),
which was controlled using the beamline-scheduling software
BSS (22, 23). The intensity data were integrated, scaled, and
total reflections
unique reflections
completeness (%)
Rmerge (%)a
13385 (1286)
99.7 (97.7)
6.9 (32.9)
6.8 (5.7)
redundancy
˚
reduced to 2.1 A resolution using DENZO and SCALEPACK
3
-1
˚
VM (A Da
)
2.9
programs (24). The data were indexed in the centered orthor-
solvent content (%)
refinement
58.9
˚
hombic space group C2221, with unit cell parameters a = 90.9 A,
˚
˚
˚
resolution range (A)
20.0-2.1
20.1
b = 108.3 A, and c = 45.4 A. The value of the Matthews
b
coefficient (25) was 2.9 A Da-1, and the solvent content was
3
˚
Rwork
c
Rfree
24.8
58.9% assuming a homodimer in the asymmetric unit.
no. of protein atoms
no. of solvent atoms
zinc ion and PEG di(hydroxyethyl) ether
Ramachandran plotd (%)
most favored
1286
The MJ1271 protein structure was determined by the molec-
ular replacement method, using the orthorhombic form of the
P. horikoshii small subunit protein (PDB id code 1V7L). The
P. horikoshii protein contains 163 amino acids, with 44% sequence
identity to MJ1271. Residues 1-137 (only core regions or
truncations) were used for molecular replacement (phasing) using
149
1 and 1
96.4
3.6
allowed
PDB identifier
2PKP
P
P
aRmerge
the last shell. Rwork
and calculated structure factors, respectively. Rfree is the R factor for a
subset of 5% of the reflections that were omitted from refinement. dAs
calculated by PROCHECK.
=
|Ii - ÆI æ|/ ÆIæ. The values within the parentheses refer to
˚
the AMoRe program (26), and the resolution limit of 10.0-6.0 A
P
P
b
=
|Fo - Fc|/ |Fo|, where Fo and Fc are the observed
was used for molecular replacement. The remaining MJ1271
residues, from 137 to 167, were manually traced based on the
electron density map, and then the residues were placed using the
MJ1271 protein sequence. A unique solution was obtained with
this search model, giving an Rfactor of 54% and a correlation factor
of 66%. The structure was refined using the program CNS
1.1 (27). The programs FRODO (28) and COOT (29) were used
for model building and viewing electron density maps. Initially 60
cycles of rigid body refinement were carried out, followed by 50
cycles of positional refinement. Without the mutated residues, the
Rwork dropped to 44% (Rfree = 50.5%) for all the reflections in the
c
were aligned with the T-COFFEE program (version 7.71) (32)
and edited using the CINEMA5 editor. Sequence logos were
produced using the WebLogo 3 program (33).
A 4-year-old commercial preparation of cis-aconitic acid
(stored dried at room temperature) contained 31% trans-aconitic
acid. For cis-aconitic acid: 1H NMR (400 MHz, D2O þ TSP, δ)
6.40 (s, 1H) and 3.51 (d, 2H, J = 1.2 Hz); 13C NMR (D2O, 100
MHz, δ) 173.1, 169.6, 169.2, 137.2, 129.0, and 34.5. For trans-
aconitic acid: 1H NMR (400 MHz, D2O þ TSP, δ) 7.03 (s, 1H)
and 3.88 (d, 2H, J = 0.6 Hz). A recently purchased batch of cis-
aconitic acid (Sigma) contained 7% trans-aconitic acid. A fresh
sample of the latter batch was dissolved in water, immediately
adjusted to pH 7 using NaOH, stored briefly on ice, and used for
the kinetic experiments reported here. The synthesis and pur-
ification of cis-homoaconitate analogues were described pre-
viously (1).
˚
resolution range 20-2.1 A. The map was clear enough to mutate
the residues into the electron density map, and then the model was
subjected to simulated annealing by employing a slow-cooling
protocol, followed by 100 cycles of positional refinement that
dropped the Rwork to 33.2% (Rfree = 37.5%). After four rounds of
manual rebuilding and subsequent refinement, the model had an
Rfree of 28.5% and Rwork of 32.5% using all observed reflections.
Solvent molecules were gradually included into the structure at
stereochemically sensible positions and with difference density
higher than 3.0σ and 2|Fo| - |Fc| density higher than 0.8σ. The
final Rwork and Rfree are 20.1% and 24.8%, respectively. During
the progress of the refinement, based on the |Fo| - |Fc| density, a
Zn ion and PEG molecule were located. The stereochemistry of
the refined structure was analyzed with the program PRO-
CHECK (30). A summary of structure determination and refine-
ment statistics is given in Table 3.
RESULTS
Previous studies showed that IPMIMj specifically catalyzed the
hydration of dicarboxylate substrates with nonpolar γ-chain
substituents: maleate, citraconate, and isopropylmaleate (11).
While HACNMj also catalyzed the hydration of the minimal
substrate maleate, the enzyme specifically recognized tricarboxy-
late substrates with variable γ-chain lengths: cis-(homo)1-aco-
nitate, cis-(homo)2-aconitate, cis-(homo)3-aconitate, and even
the nonphysiological cis-(homo)4-aconitate (1). Surprisingly, no
activity was observed in mixtures containing HACNMj and cis-
aconitate, whose γ-chain is one methylene group shorter than
The SWISS-MODEL automated comparative protein model-
ing server (version 8.03) constructed a MJ1277 homology model
using the MJ1271 structure as the template (31). Figures were
prepared using the PYMOL program (version 1.2r1; DeLano
Scientific). Archaeal HACN and IPMI small subunit sequences

Article
Biochemistry, Vol. 49, No. 12, 2010 2691
F
IGURE 2: Ribbon diagram showing the MJ1271 subunit in the
asymmetricunit. Positions of the Arg26 and Thr27 residues are labeled
on the flexible loop between R-helices 2 and 3.
67
an rmsd of 0.1 A. The catalytic Ser and Arg69 residues in the
˚
MJ1277 model are equivalent to Ser65 and Arg67 in MJ1271, but
residues Val28-Tyr29 replace the polar Arg26-Thr27 residues in the
flexible loop region between R2 and R3 (Figure 4). An alignment
of numerous small subunit sequences from homologous archaeal
HACN and IPMI proteins confirmed the correlation between
residues in this flexible loop and the protein’s predicted substrate
specificity (Figure 5). Putative HACN subunits usually have a
basic residue (Lys or Arg) and Thr in this loop. IPMI subunits
show more diversity in this loop: Thr/Ile/Val and Ile/Tyr replace
their more polar counterparts in HACN.
F
IGURE 1: Ribbon diagram showing the crystallographic dimer of
two MJ1271 subunits. Each protomer coordinates a Zn2þ ion
(purple).
cis-(homo)1-aconitate. We subsequently discovered that the dry,
commercially produced cis-aconitic acid sample was contami-
nated with 31% trans-aconitic acid, a potent aconitase, and
homoaconitase inhibitor (1, 34). Concentrated cis-aconitic acid
rapidly isomerizes to trans-aconitic acid in aqueous solutions,
although the dried compound and neutral solutions are consi-
dered stable (35). Using a new batch of cis-aconitate (adjusted to
pH 7 with NaOH, containing 7% trans-aconitate), we deter-
mined that HACNMj catalyzed cis-aconitate hydration with a KM
of 300 ( 90 μM and a kcat/KM = 2.5 ꢀ 103 M-1 s-1. This speci-
To determine whether the Y24LRT sequence of MJ1271
confers substrate specificity for HACNMj, six mutations in the
MJ1271 gene were constructed, and the altered proteins were
coexpressed with wild-type MJ1003 in E. coli, as described
previously (1). In aerobic analytical size exclusion chromatogra-
phy experiments, the six purified HACNMj variants eluted as
complexes with apparent molecular masses of 144-155 kDa.
Masses of the large and small subunit monomers are approxi-
mately 46 and 18 kDa, respectively. Therefore, these variant
forms of HACNMj form heterotetramers consisting of two large
ficity constant is 10-fold lower that values for cis-(homo)1-4
-
aconitate due to the high KM value for cis-aconitate. Therefore,
HACNMj is specific for cis-unsaturated tricarboxylates, while
IPMIMj recognizes cis-unsaturated dicarboxylates.
The structure of the HACNMj small subunit, MJ1271, was
solved by molecular replacement using the P. horikoshii small
subunit structure that shares 44% amino acid identity and 73%
similarity. The MJ1271 protein crystallized as a dimer, with Asp13
and Cys63 side chains from each subunit coordinating Zn2þ ions
(Figure 1). Residues 1-167 (out of 170) were modeled using
and two small subunits, consistent with wild-type HACNMj
,
which eluted with a molecular mass of 143 kDa.
Sequence and structural alignments indicated that Arg26 in the
MJ1271 flexible loop could act similarly to Arg580 of mACN (14).
Zheng et al. reported that replacing Arg580 with Lys resulted in a
30-fold increase in KM for mACN with isocitrate, a decrease in
activity, and a loss of tight substrate binding (15). The analogous
R26K substitution in MJ1271 broadened the enzyme’s substrate
specificity: the enzyme catalyzed citraconate hydration and the
dehydration of 3-isopropylmalate, and the substitution caused a
6-fold increase in KM for cis-homo1-aconitate (Tables 4 and 5). In
contrast to Arg, the smaller Lys side chain may not interfere with
the hydrophobic γ-chains of the citraconate and 3-isopropylma-
late. Consistent with this model, the steady-state kinetic para-
meters KM and kcat are almost identical for the R26K-catalyzed
hydration of citraconate (315 μM, 8.4 s-1) and maleate (310 μM,
8.1 s-1) (Figure 6).
˚
X-ray diffraction data at 2.1 A resolution. These proteins have a
“swiveling” β/β/R domain consisting of a β-sheet with parallel
strands (β1-4) and a second β-sheet with antiparallel strands (β1,
β5-7) in an R/β fold (Figure 2). Together, these two β-sheets
form a β-barrel-type structure that is capped by R7 and R8. The
conserved Ser65 residue, corresponding to the alkoxide base
Ser642 in mACN (36), is located in a loop region between β2
and R5 (Figure 3). The main-chain bond angle of this residue is
10.2ꢀ greater than the ideal small molecule value. The adjacent
Arg67 (corresponding to Arg644 in mACN) stabilizes the serine
alkoxide and interacts with substrate carboxylate groups in
˚
mACN (14). This strained loop is approximately 15 A from a
flexible loop between R2 and R3, which includes Arg580 in
mACN. A structural alignment shows that Arg580 has no direct
equivalent in the smaller loop of MJ1271; instead, we predicted
the nearby Arg26-Thr27 residues could interact with the sub-
strates’ γ-chain substituents (Figure 3).
Replacing the MJ1271 Arg26 with Val, the analogous residue in
MJ1277, had a greater effect on substrate recognition. This variant
catalyzed citraconate and maleate hydration with kinetic rate
constants similar to those of the R26K protein; however, 3-iso-
propylmalate dehydratase activity increased substantially and
homoaconitase activity decreased (Table 4 and Figure 6). The
R26V variant catalyzed 3-isopropylmalate dehydration with a KM
less than 2-fold higher than the wild-type IPMIMj and a similar
The IPMIMj small subunit, MJ1277, shares 53% sequence
identity with the HACNMj small subunit, MJ1271. Therefore, a
homology model of MJ1277 was built using the MJ1271 structure
coordinates as a template, resulting in a structural alignment with

2692 Biochemistry, Vol. 49, No. 12, 2010
Jeyakanthan et al.
F
IGURE 3: Amino acid sequence alignment of the MJ1271, MJ1277, P. horikoshii HACN small subunit, and mitochondrial aconitase (domain 4)
proteins. The MATRAS program (47) aligned structural models of MJ1271 (PDB id 2PKP), PhHACN (1V7L), and mACN (7ACN). A pairwise
sequencealignmentofthe MJ1277and MJ1271sequences was manuallyadded tothe alignment. Identical aminoacidpositions are shown inwhite
text on a black background, while similar positions are shown in black on a gray background. β-Strands identified in the MJ1271 structure are
shown as black arrows above the sequence, and helices are shown as gray cylinders. A box highlights the flexible loop region between R2 and R3,
which includes MJ1271 Arg26 and Thr27 residues. An asterisk above MJ1271 Ser65 identifies the alkoxide base catalyst.
hydrophobic methyl or isopropyl γ-chains of citraconate and
3-isopropylmalate.
In the conserved loop regions of HACNMj and IPMIMj
,
residues adjacent to the critical arginine or valine also differ.
Although no other substrate-binding residues have been identi-
fied in the corresponding loop of mACN, any change in the
loop’s orientation or flexibility could affect substrate specificity.
To determine the catalytic role of Thr27 in HACNMj, the MJ1271
Thr27Ala variant was constructed. In a holoenzyme complex, this
T27A variant catalyzed the hydration of citraconate and maleate
substrates with a 10-fold higher KM than wild-type IPMIMj
(Table 4). Also, the KM values for cis-homoaconitate substrates
increased 10-20-fold relative to the wild-type HACNMj, which
was offset by the variant’s increased turnover to produce similar
specificity constants (Figure 6). Unlike the R26V variant, the
T27A variant had no detectable dehydratase activity with
3-isopropylmalate, probably due to steric hindrance between the
Arg26 guanidinium group and the substrate’s isopropyl group.
A double replacement, MJ1271 R26V T27Y, was constructed to
remodel the loop region of HACNMj to resemble the IPMIMj
loop region. The double mutation resulted in a KM of 489 μM for
maleic acid, similar to wild-type HACNMj and IPMIMj, and a
turnover that was close to the Thr27Ala and IPMIMj proteins’
values. However, compared to the single Arg26Val protein, the
KM and kcat values for both citraconate and 3-isopropylmalate
increased slightly. The KM values for cis-(homo)1-aconitate,
cis-(homo)2-aconitate, and cis-(homo)3-aconitate increased 20-
50-fold compared to wild-type HACNMj. Although the double
replacement did result in an increase in turnover for cis-(homo)1-
aconitate and cis-(homo)2-aconitate, all three tricarboxylate
substrates had specificity constants that were an order of
magnitude lower than those of the wild-type enzyme (Table 5
and Figure 6). The difference in rate constants for citraconate and
3-isopropylmalate between the double replacement and Arg26Val
replacement proteins probably results from structural changes in
the loop region caused by the threonine to tyrosine substitution,
as discussed for MJ1271 T27A above.
F
the homology model of the MJ1277 protein (green).
IGURE 4: Stereoview of the MJ1271 structure (cyan) aligned with
F
IGURE 5: Sequence logos show the correlation between the flexible
loop region of archaeal small-subunit proteins and their predicted
specificities. HACN small subunits contain conserved Arg/Lys26 and
Thr27 residues (part A), while IPMI paralogues contain less polar
Val/Thr/Ile28 and Tyr/Ile29 residues at the homologous positions
(part B). Conserved residues from alignments of 26 putative HACN
and 23 putative IPMI proteins are represented by single letter amino
acid abbreviations, where letter heights (bits of information) corre-
spond to their frequency in the aligned column (33).
turnover (1.1 and 1.9 s-1, respectively). In a trade-off, the KM for
cis-(homo)1-aconitate, cis-(homo)2-aconitate, and cis-(homo)3-
aconitate increased between 10- and 30-fold for the R26V pro-
tein. Therefore, Arg26 of MJ1271 plays a key role in homoaco-
nitate substrate recognition, while discriminating against the
A chimeric MJ1271 protein was constructed with the first 28
amino acids of MJ1277 to test whether the holoenzyme’s sub-
strate specificity could be completely altered from HACNMj

Article
Biochemistry, Vol. 49, No. 12, 2010 2693
Table 4: Steady-State Kinetic Parameters for IPMI Substrates
maleate
citraconate
3-isopropylmalate
kcat (s-1
)
enzyme
KM (μM)a
kcat (s-1
)
KM (μM)
kcat (s-1
)
KM (μM)
b
IPMIMj
400 ( 50
230 ( 10
310 ( 32
3400 ( 200
489 ( 86
36
80 ( 20
190 ( 20
315 ( 71
920 ( 230
290 ( 70
14
39 ( 7
68 ( 5
180 ( 30
NDc
1.9
1.1
0.8
ND
1.6
MJ1271 R26V
MJ1271 R26
MJ1271 T27
MJ1271 R26V T27
9.9
5.5
8.36
4.1
9.8
K
8.1
A
24.2
23.1
Y
133 ( 37
aInitial rate data for the first-order reactions were fitted with the Michaelis-Menten-Henri equation by nonlinear regression. The mean values for both kinetic
parameters are shown, along with the standard error for KM values. bValues for wild-type IMPIMj were reported previously (11). The parameters for HACNMj in
maleate hydratase reactions were KM = 330 ( 50 μMand kcat = 6 s-1. HACNMj has no detectable activity with citraconate or isopropylmalate (1). cND, not detected.
Table 5: Steady-State Kinetic Parameters for HACN Substrates
cis-homo1-aconitate
cis-homo2-aconitate
cis-homo3-aconitate
kcat (s-1
)
enzyme
KM (μM)a
kcat (s-1
)
KM (μM)
kcat (s-1
)
KM (μM)
b
HACNMj
22 ( 3
0.75
0.48
1.43
2.5
30 ( 5
0.66
5.8
36 ( 6
660 ( 170
NT
2.5
2.8
NT
4.1
1.9
MJ1271 R26
MJ1271 R26K
MJ1271 T27
MJ1271 R26V T27
V
220 ( 30
135 ( 27
220 ( 20
460 ( 80
870 ( 150
NTc
NT
2.2
A
269 ( 68
1600 ( 500
650 ( 80
640 ( 90
Y
1.7
6.6
aKinetic parameters were determined as described for Table 4. bValues for wild-type HACNMj were reported previously (1). cNT, not tested.
to IPMIMj. The chimera interacted with MJ1003, forming a
heterotetramer in size exclusion chromatography. Although this
association suggested that the chimeric protein folded correctly,
no activity was detected for any of the tested substrates after
reconstitution. No further experiments were performed to ex-
amine the chimeric protein’s structure. This inactive protein is
reminiscent of the stable MJ1003-MJ1277 heterotetramer, which
also lacked catalytic activity (1).
In contrast to HACN, the homoaconitate hydratase protein
catalyzes only the hydration of cis-homoaconitate to (2R,3S)-
homoisocitrate. Sequence alignments and homology models of
the T. thermophilus homoaconitate hydratase (LysU) protein
reveal a flexible loop region with the sequence Y21APFMV but
do not provide insight into the mechanistic discrepancy. A
chimeric form of MJ1271 was created, replacing the loop region
with the corresponding region of the T. thermophilus LysU pro-
tein, in an effort to alter the reactions catalyzed by the HACNMj to
that of the homoaconitate hydratases. Although the chimeric
protein still interacted with MJ1003 and maintained relatively low
activity with maleic acid (kcat/KM = 1.4 ꢀ 103 M-1 s-1), no acti-
vity was detected for cis-(homo)1-aconitate or for any other tested
substrates.
DISCUSSION
F
IGURE 6: Substrate specificities for wild-type MJ1271 (HACN),
The HACNMj small subunit protein, MJ1271, is homologous to
the fourth domain of mACN. Structural alignment of MJ1271 with
the relaxed (without substrate) or tense (substrate bound) struc-
MJ1271 variants, and wild-type IPMI proteins were estimated from
steady-state rate constants using the six substrates shown in the legend
above the graph. Together with the MJ1003 large-subunit protein, wild-
˚
type MJ1271 catalyzes the hydration of maleate and cis-(homo)1-3
-
tures of mACN resulted in an rmsd of 1.9 A (Z scores between 19.0
aconitate substrates with similar specificity constants but has no activity
with 3-isopropylmalate or citraconate (1). The MJ1271 R26K variant
catalyzes 3-isopropylmalate dehydration, as well the hydration of citra-
conate, maleate, and cis-homoaconitate. The R26V variant is promiscu-
ous, with enhanced IPMI activity and reduced HACN activity. The
T26A variant is also promiscuous, with uniform specificity constants for
both IPMI and HACN reactions. The specificity of the R26V T27Y
variant resembles the R26V variant. Wild-type IPMIMj (MJ1277 and
MJ0499) efficiently acts on 3-isopropylmalate, citraconate, and maleate
but has no detectable activity with cis-(homo)-aconitate analogues (11).
and 19.2), as indicated by the DALI server (Figure 7). In both the
relaxed and tense states of mACN, the R-helical region of domain 4
forms the active site cleft at the interface with domains 1, 2, and 3.
In HACN and IPMI proteins, the large subunit contains domains
1-3 and provides cysteine ligands for the [4Fe-4S] cluster. The
catalytically important Arg580 and Ser642 residues, located within
the loop regions of mACN domain 4, are considered essential for
aconitase activity. Therefore, we predict that the corresponding

2694 Biochemistry, Vol. 49, No. 12, 2010
Jeyakanthan et al.
mACN that contains Arg580. Instead, Arg26 of MJ1271 aligns with
Leu577 of mACN, and there is an approximately 120ꢀ diffe-
rence in the orientation of the arginine residues (Figure 8). Arg26
is oriented perpendicular to the GS65SRE sequence and does not
appear to be in line to contact the substrate’s γ-carboxylate, as in
mACN Arg580. Substrate binding at the mACN active site results in
˚
conformational changes 30 A from the active site; similar changes
could reposition MJ1271 Arg26. The orientation of the Y24LRT
loop region may also be influenced by interaction with MJ1003.
Aconitase proteins require a [4Fe-4S] cluster for substrate binding
and orientation; therefore, we did not attempt to cocrystallize the
MJ1271 protein with ligands. Structures of the MJ1003-MJ1271
ternary complex with substrate will be required to accurately assess
the orientation of Arg26 of MJ1271 in HACNMj
.
Site-directed mutagenesis of MJ1271 suggests that Arg26 is a
critical residue for HACNMj substrate specificity. Replacing this
residue with valine or lysine dramatically increased the Michaelis
constants for cis-(homo)1-aconitate, cis-(homo)2-aconitate, and
cis-(homo)3-aconitate compared to the wild-type HACNMj. At
the same time, these variants had broader substrate specificity:
the Arg26Val substitution resulted in KM and kcat values compar-
able to those of IPMIMj, suggesting that Val28 of the MJ1277
Y26LVY sequence is necessary to accommodate the hydrophobic
γ-chains of citraconate and dimethylcitraconate.
F
IGURE 7: Alignment of MJ1271 (cyan and pink) with mACN in the
relaxed conformation (gray, PDB id 1AMJ) (18). The MJ1271
subunit is homologous to domain 4 of mACN. N- and C-termini
are indicated for both proteins.
The residues adjacent to mACN Arg580 have not been shown to
function in the aconitase mechanism. The adjacent residue to
MJ1271 Arg26 is Thr27, which corresponds to Tyr29 of MJ1277.
Replacing Thr27 with alanine disrupted the activity of HACNMj for
all substrates, most notably by increasing the KM for maleate 10-
fold compared to wild-type HACNMj. The MJ1271 structure
shows the Thr27 side chain oriented away from the predicted active
site. This residue is highly conserved in HACN small subunit
sequences, but additional mutagenesis experiments will be required
to distinguish polar interactions with the hydroxyl group from
steric effects on conformation. Corresponding residues in this
position are overwhelmingly aromatic (Tyr) or β-branched amino
acids (Thr or Ile), which favor an extended conformation.
The homoisocitrate dehydrogenase (HICDH) enzyme evolved
from isopropylmalate dehydrogenase through a similar series of
active site amino acid substitutions. HICDH binds products of
HACN or homoaconitate hydratase and catalyzes their oxidative
decarboxylation, forming 2-oxoadipate and longer chain analo-
gues. Members of this family catalyze the NAD(P)þ-dependent
decarboxylation of R-hydroxy acids with different γ-chain substit-
uents (37). Both isocitrate dehydrogenase and homoisocitrate
dehydrogenase recognize R-hydroxytricarboxylates, while isopro-
pylmalate dehydrogenase (IPMDH) recognizes R-hydroxydicar-
boxylates. Random mutagenesis identified isocitrate dehydro-
genase mutants with low levels of IPMDH activity (38). Site-
directed mutagenesis of the T. thermophilus HICDH identified
Arg85 as the key specificity determinant: an R85V variant was
transformed into an IPMDH (39). Subsequently, a crystal struc-
ture model of the T. thermophilus HICDH identified the Arg85
guanidinium group near the putative substrate-binding site, poised
to form an ion pair with the γ-carboxylate of isocitrate or
homoisocitrate (40). The P. horikoshii HICDH sequence, however,
lacks a basic residue homologous to Arg85 although that enzyme’s
substrate specificity closely resembles that of T. thermophilus
HICDH (41). HACNMj and HICDH enzymes from modern,
currently living species are highly specific for tricarboxylate
substrates. Yet a small number of amino acid substitutions
converts them into promiscuous enzymes (42, 43). Homologous
F
IGURE 8: Side chains of serine and arginine residues form hydrogen
bonds with isocitrateinthe porcine mACNactivesite(partA, PDB id
7ACN). An alignment of the MJ1271 subunit structure with mACN
in the relaxed state (part B, PDB id 1AMJ), made by the DaliLite
server, shows a substantial shift in the corresponding arginine
residues (Arg26/Arg580 and Arg67/Arg644). Carbon atoms for the
MJ1271 residues are shown in green, while those from mACN
are shown in light blue.
regions of MJ1271, including Arg26 and Ser65, are positioned
at the interface with the MJ1003 protein in the HACNMj
holoprotein.
The CR backbones of the conserved GSSRE sequences in
relaxed mACN and MJ1271 overlap in a structural alignment
(Figures 3 and 7), and the homologous serine side chains have
similar orientations. However, Arg644 in mACN, which stabilizes
the Ser642 alkoxide, differs in orientation from Arg67 in MJ1271
(Figure 8). Also, the GSSRE sequence of mACN resides com-
pletely in a loop region, while the GSSRE sequence of MJ1271
extends into R6. Aligning MJ1271 with tense, substrate-bound
mACN (PDB id 1AMI) shows a similar alignment of CR back-
bones and side chain residues of the GSSRE sequence. Ser642 of
tense mACN is rotated almost 180ꢀ from the relaxed mACN and
MJ1271 serine side chains. Therefore, the MJ1271 structure most
resembles the relaxed state of mACN.
The Y24LRT residues are close to the GS65SRE loop in
MJ1271, indicating that this sequence could interact with the
γ-carboxylates of the HACN substrates. Still, the YLRT se-
quence aligns poorly with the corresponding loop region of

Article
Biochemistry, Vol. 49, No. 12, 2010 2695
proteins in ancestral organisms could have had broader specificies.
From these examples, we infer that the evolutionary potential for
catalytic promiscuity fostered the diversification of both the
isomerase and oxidative decarboxylase families (44, 45), enabling
the evolution of the diverse 2-oxo acid elongation pathways (46).
Although the substrate specificity of the MJ1271 protein was
altered to accommodate the substrates of MJ1277, activity was
not completely lost for tricarboxylate substrates, forming a
promiscuous enzyme. The interacting large subunits could also
contribute to substrate specificity by modulating conformational
changes in the small subunits and by interacting with the
substrates’ R- and β-carboxylate groups. However, the steady-
state kinetic analysis of HACNMj and IPMIMj, the crystal-
lography and homology modeling of MJ1271 and MJ1277,
and the site-directed mutagenesis of MJ1271 all indicate that
the conserved Y24LRT and Y26LVY sequences are consistent
indicators of protein specificity in the Archaea. Therefore, uncha-
racterized proteins with the Y24LRT sequence are likely HACNs,
while proteins with the YLV(Y/I/M) sequence are IPMIs,
validating the motifs deduced in Figure 5. These studies should
facilitate annotation of these uncharacterized proteins.
The HACNMj, IPMIMj, and mACN proteins catalyze the
isomerization of R-hydroxy acids to form β-hydroxy acids. Given
the similarity of the homoaconitate hydratases to these enzymes,
it is surprising the initial dehydration of (R)-homocitrate has not
been observed. We replaced the region of MJ1271 containing the
Y24LRT sequence with the corresponding sequence from the
T. thermophilus HACN in an attempt to alter the catalytic mecha-
nism to accommodate only the hydration of cis-homoaconitate
to (2R,3S)-homoisocitrate. Although the MJ1003 and MJ1271/
LysU variant subunits interacted to form a heterotetramer with
malease activity, no activity was observed with any of the other
HACNMj or IPMIMj substrates. Preserving the position and
orientation of residues in this flexible loop may be essential for
reproducing activity with cis-homoaconitate. Therefore, a crystal
structure of HACN with bound substrate, as well as mutagenesis
of homoaconitate hydratase active site amino acids, will be
required to resolve this difference in reactivity.
2. Gruer, M. J., Artymiuk, P. J., and Guest, J. R. (1997) The aconitase
family: three structural variations on a common theme. Trends
Biochem. Sci. 22, 3–6.
3. Weidner, G., Steffan, B., and Brakhage, A. A. (1997) The Aspergillus
nidulans lysF gene encodes homoaconitase, an enzyme involved in the
fungus-specific lysine biosynthesis pathway. Mol. Gen. Genet. 255,
237–247.
4. Jia, Y., Tomita, T., Yamauchi, K., Nishiyama, M., and Palmer, D. R.
J. (2006) Kinetics and product analysis of the reaction catalysed by
recombinant homoaconitase from Thermus thermophilus. Biochem. J.
396, 479–485.
5. Grimek, T. L., and Escalante-Semerena, J. C. (2004) The acnD genes
of Shewenella oneidensis and Vibrio cholerae encode a new Fe/S-
dependent 2-methylcitrate dehydratase enzyme that requires prpF
function in vivo. J. Bacteriol. 186, 454–462.
6. Alhapel, A., Darley, D. J., Wagener, N., Eckel, E., Elsner, N., and
Pierik, A. J. (2006) Molecular and functional analysis of nicotinate
catabolism in Eubacterium barkeri. Proc. Natl. Acad. Sci. U.S.A. 103,
12341–12346.
7. Blank, L., Green, J., and Guest, J. R. (2002) AcnC of Escherichia coli
is a 2-methylcitrate dehydratase (PrpD) that can use citrate and
isocitrate as substrates. Microbiology 148, 133–146.
8. Beinert, H., Kennedy, M. C., and Stout, C. D. (1996) Aconitase as
iron-sulfur protein, enzyme, and iron-regulatory protein. Chem. Rev.
96, 2335–2374.
9. Kennedy, M. C., and Beinert, H. (2007) Aconitase, in Biological
Inorganic Chemistry: Structure and Reactivity (Bertini, I., Gray, H.
B., Stiefel, E. I., and Valentine, J. S., Eds.) pp 209-215, University
Science Books, Sausalito, CA.
10. Gross, S. R., Burns, R. O., and Umbarger, H. E. (1963) The
biosynthesis of leucine. II. The enzymic isomerization of β-carboxy-
β-hydroxyisocaproate and R-hydroxy-β-carboxyisocaproate. Bio-
chemistry 2, 1046–1052.
11. Drevland, R. M., Waheed, A., and Graham, D. E. (2007) Enzymology
and evolution of the pyruvate pathway to 2-oxobutyrate in Metha-
nocaldococcus jannaschii. J. Bacteriol. 189, 4391–4400.
12. Xu, H., Andi, B., Qian, J., West, A. H., and Cook, P. F. (2006) The
R-aminoadipate pathway for lysine biosynthesis in fungi. Cell Bio-
chem. Biophys. 46, 43–64.
13. Strassman, M., and Ceci, L. N. (1966) Enzymatic formation of cis-
homoaconitic acid, an intermediate in lysine biosynthesis in yeast.
J. Biol. Chem. 241, 5401–5407.
14. Lauble, H., Kennedy, M. C., Beinert, H., and Stout, C. D. (1992)
Crystal structures of aconitase with isocitrate and nitroisocitrate
bound. Biochemistry 31, 2735–2748.
15. Zheng, L., Kennedy, M. C., Beinert, H., and Zalkin, H. (1992)
Mutational analysis of active site residues in pig heart aconitase. J.
Biol. Chem. 267, 7895–7903.
16. Yasutake, Y., Yao, M., Sakai, N., Kirita, T., and Tanaka, I. (2004)
Crystal structure of the Pyrococcus horikoshii isopropylmalate iso-
merase small subunit provides insight into the dual substrate specifi-
city of the enzyme. J. Mol. Biol. 344, 325–333.
17. Lombo, T., Takaya, N., Miyazaki, J., Gotoh, K., Nishiyama, M.,
Kosuge, T., Nakamura, A., and Hoshino, T. (2004) Functional
analysis of the small subunit of the putative homoaconitase from
Pyrococcus horikoshii in the Thermus lysine biosynthetic pathway.
FEMS Microbiol. Lett. 233, 315–324.
18. Lauble, H., and Stout, C. D. (1995) Steric and conformational
features of the aconitase mechanism. Proteins 22, 1–11.
19. Horton, R. M., Cai, Z., Ho, S. N., and Pease, L. R. (1990) Gene
splicing by overlap extension: tailor-made genes using the polymerase
chain reaction. BioTechniques 8, 528–535.
Although most crenarchaea use the R-aminoadipate and
isopropylmalate pathways for lysine and leucine biosyntheses,
respectively, they appear to only have one HACN/IPMI homo-
logue with a conserved YL(K/V)Y sequence in the flexible loop
region. These putative promiscuous enzymes are expected to
accept a broad pool of hydroxy acid substrates. No crenarchaeal
homologues have been purified and characterized, although the
results of the MJ1271 mutagenesis show that mutating Arg26 of
YLRT to either lysine or valine allows for the recognition of both
HACNMJ and IPMIMJ substrates. Therefore, future studies will
involve the characterization, crystallization, and mutagenesis of
these crenarchaeal proteins to determine how ancestral hydro-
lyases recognize their substrates.
ꢀ
€
20. Helgadottir, S., Rosas-Sandoval, G., Soll, D., and Graham, D. E.
(2007) Biosynthesis of phosphoserine in the Methanococcales. J.
Bacteriol. 189, 575–582.
21. Sugahara, M., and Miyano, M. (2002) Development of high-through-
put automatic protein crystallization and observation system. Tanpa-
kushitsu Kakusan Koso 47, 1026–1032.
ACKNOWLEDGMENT
22. Ueno, G., Hirose, R., Ida, K., Kumasaka, T., and Yamamoto, M.
(2004) Sample management system for a vast amount of frozen
crystals at SPring-8. J. Appl. Crystallogr. 37, 867–873.
23. Ueno, G., Kanda, H., Kumasaka, T., and Yamamoto, M. (2005)
Beamline Scheduling Software: administration software for auto-
matic operation of the RIKEN structural genomics beamlines at
SPring-8. J. Synchrotron Radiat. 12, 380–384.
24. Otwinowski, Z., and Minor, W. (1997) Processing of X-ray diffraction
data collected in oscillation mode. Methods Enzymol. 276, 307–326.
25. Matthews, B. W. (1968) Solvent content of protein crystals. J. Mol.
Biol. 33, 491–497.
We thank the staff at the SPring-8 beamlines BL12B2 and
BL26B2 for use of their excellent facilities and assistance with
X-ray data collection.
REFERENCES
1. Drevland, R. M., Jia, Y., Palmer, D. R. J., and Graham, D. E. (2008)
Methanogen homoaconitase catalyzes both hydrolyase reactions in
coenzyme B biosynthesis. J. Biol. Chem. 283, 28888–28896.

2696 Biochemistry, Vol. 49, No. 12, 2010
Jeyakanthan et al.
26. Navaza, J. (1994) AMoRe: an automated package for molecular
replacement. Acta Crystallogr. A 50, 157–163.
linked R-hydroxyacid oxidative decarboxylases. Biochemistry 48,
3565–3577.
€
38. Doyle, S. A., Fung, S.-Y. F., and Koshland, D. E. (2000) Redesigning
the substrate specificity of an enzyme: isocitrate dehydrogenase.
Biochemistry 39, 14348–14355.
39. Miyazaki, J., Kobashi, N., Nishiyama, M., and Yamane, H. (2003)
Characterization of homoisocitrate dehydrogenase involved in lysine
biosynthesis of an extremely thermophilic bacterium, Thermus ther-
mophilus HB27, and evolutionary implication of beta-decarboxylat-
ing dehydrogenase. J. Biol. Chem. 278, 1864–1871.
27. Brunger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P.,
Grosse-Kunstleve, R. W., Jiang, J.-S., Kuszewski, J., Nilges, M.,
Pannu, N. S., Read, R. J., Rice, L. M., Simonson, T., and Warren,
G. L. (1998) Crystallography & NMR system: a new software suite
for macromolecular structure determination. Acta Crystallogr., Sect.
D: Biol. Crystallogr. 54, 905–921.
28. Jones, T. A. (1985) Interactive computer graphics: FRODO. Methods
Enzymol. 115, 157–171.
29. Emsley, P., and Cowtan, K. (2004) Coot: model-building tools for
molecular graphics. Acta Crystallogr., Sect. D: Biol. Crystallogr. 60,
2126–2132.
30. Laskowski, R. A., MacArthur, M. W., Moss, D. S., and Thornton, J.
M. (1993) PROCHECK: a program to check the stereochemical
quality of protein structures. J. Appl. Crystallogr. 26, 283–291.
31. Schwede, T., Kopp, J., Guex, N., and Peitsch, M. C. (2003) SWISS-
MODEL: an automated protein homology-modeling server. Nucleic
Acids Res. 31, 3381–3385.
32. Notredame, C., Higgins, D. G., and Heringa, J. (2000) T-Coffee: a
novel method for fast and accurate multiple sequence alignment. J.
Mol. Biol. 302, 205–217.
33. Crooks, G. E., Hon, G., Chandonia, J.-M., and Brenner, S. E. (2004)
WebLogo: a sequence logo generator. Genome Res. 14, 1188–1190.
34. Lauble, H., Kennedy, M. C., Beinert, H., and Stout, C. D. (1994)
Crystal structures of aconitase with trans-aconitate and nitrocitrate
bound. J. Mol. Biol. 237, 437–451.
40. Miyazaki, J., Asada, K., Fushinobu, S., Kuzuyama, T., and Nishiya-
ma, M. (2005) Crystal structure of tetrameric homoisocitrate dehy-
drogenase from an extreme thermophile, Thermus thermophilus:
involvement of hydrophobic dimer-dimer interaction in extremely
high thermotolerance. J. Bacteriol. 187, 6779–6788.
41. Miyazaki, K. (2005) Bifunctional isocitrate-homoisocitrate dehydro-
genase: a missing link in the evolution of beta-decarboxylating
dehydrogenase. Biochem. Biophys. Res. Commun. 331, 341–346.
42. O’Brien, P. J., and Herschlag, D. (1999) Catalytic promiscuity and the
evolution of new enzymatic activities. Chem. Biol. 6, R91–R105.
43. Copley, S. D. (2003) Enzymes with extra talents: moonlighting func-
tions and catalytic promiscuity. Curr. Opin. Chem. Biol. 7, 265–272.
44. Schmidt, D. M. Z., Mundorff, E. C., Dojka, M., Bermudez, E., Ness,
J. E., Govindarajan, S., Babbitt, P. C., Minshull, J., and Gerlt, J. A.
(2003) Evolutionary potential of (β/R)8-barrels: functional promiscu-
ity produced by single substitutions in the enolase superfamily.
Biochemistry 42, 8387–8393.
35. Krebs, H. A., and Eggleston, L. V. (1944) Micro-determination of
isocitric and cis-aconitic acids in biological material. Biochem. J. 38,
426–437.
45. Khersonsky, O., Roodveldt, C., and Tawfik, D. S. (2006) Enzyme
promiscuity: evolutionary and mechanistic aspects. Curr. Opin. Chem.
Biol. 10, 498–508.
36. Lloyd, S. J., Lauble, H., Prasad, G. S., and Stout, C. D. (1999) The
˚
mechanism of aconitase: 1.8 A resolution crystal structure of the
S642A:citrate complex. Protein Sci. 8, 2655–2662.
37. Aktas, D. F., and Cook, P. F. (2009) A lysine-tyrosine pair carries out
acid-base chemistry in the metal ion-dependent pyridine dinucleotide-
46. Graham, D. E., and White, R. H. (2002) Elucidation of methanogenic
coenzyme biosyntheses: from spectroscopy to genomics. Nat. Prod.
Rep. 19, 133–147.
47. Kawabata, T. (2003) MATRAS: a program for protein 3D structure
comparison. Nucleic Acids Res. 31, 3367–3369.