Kinetic and structural characterization of DmpI from Helicobacter pylori and Archaeoglobus fulgidus, two 4-oxalocrotonate tautomerase family members

Article abstract of DOI:10.1016/j.bioorg.2010.07.002

The tautomerase superfamily consists of structurally homologous proteins that are characterized by a β-α-β fold and a catalytic amino-terminal proline. 4-Oxalocrotonate tautomerase (4-OT) family members have been identified and categorized into five subfa

Full text of DOI:10.1016/j.bioorg.2010.07.002

Bioorganic Chemistry 38 (2010) 252–259
Contents lists available at ScienceDirect
Bioorganic Chemistry
journal homepage: www.elsevier.com/locate/bioorg
Kinetic and structural characterization of DmpI from Helicobacter pylori and
Archaeoglobus fulgidus, two 4-oxalocrotonate tautomerase family members q,qq
Jeffrey J. Almrud a, Rakhi Dasgupta b, Robert M. Czerwinski a, Andrew D. Kern b, Marvin L. Hackert b,
**
,
*
Christian P. Whitman a,
a Division of Medicinal Chemistry, College of Pharmacy, The University of Texas, Austin, TX 78712-1071, United States
b Department of Chemistry and Biochemistry, The University of Texas, Austin, TX 78712-1071, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 14 June 2010
Available online 18 July 2010
The tautomerase superfamily consists of structurally homologous proteins that are characterized by a b-
-b fold and a catalytic amino-terminal proline. 4-Oxalocrotonate tautomerase (4-OT) family members
a
have been identified and categorized into five subfamilies on the basis of multiple sequence alignments
and the conservation of key catalytic and structural residues. Representative members from two subfam-
ilies have been cloned, expressed, purified, and subjected to kinetic and structural characterization. The
crystal structure of DmpI from Helicobacter pylori (HpDmpI), a 4-OT homolog in subfamily 3, has been
determined to high resolution (1.8 Å and 2.1 Å) in two different space groups. HpDmpI is a homohexamer
with an active site cavity that includes Pro-1, but lacks the equivalent of Arg-11 and Arg-39 found in 4-
OT. Instead, the side chain of Lys-36 replaces that of Arg-11 in a manner similar to that observed in the
trimeric macrophage migration inhibitory factor (MIF), which is the title protein of another family in the
superfamily. The electrostatic surface of the active site is also quite different and suggests that HpDmpI
might prefer small, monoacid substrates. A kinetic analysis of the enzyme is consistent with the struc-
tural analysis, but a biological role for the enzyme remains elusive. The crystal structure of DmpI from
Archaeoglobus fulgidus (AfDmpI), a 4-OT homolog in subfamily-4, has been determined to 2.4 Å resolution.
AfDmpI is also a homohexamer, with a proposed active site cavity that includes Pro-1, but lacks any other
residues that are readily identified as catalytic ones related to 4-OT activity. Indeed, the electrostatic
potential of the active site differs significantly in that it is mostly neutral, in contrast to the usual elec-
tropositive features found in other 4-OT family members, suggesting that AfDmpI might accommodate
hydrophobic substrates. A kinetic analysis has been carried out, but does not provide any clues about
the type of reaction the enzyme might catalyze.
Keywords:
4-oxalocrotonate tautomerase
Catalytic proline
Hexamer
Ó 2010 Elsevier Inc. All rights reserved.
Abbreviations: AfDmpI, DmpI from Archaeoglobus fulgidus; CaaD and cis-CaaD,
trans- and cis-3-chloroacrylic acid dehalogenase respectively; CHMI, 5-(carboxy-
methyl)-2-hydroxymuconate isomerase; Fo and Fc, observed and calculated
structure factors respectively; HEPES, N-(2-hydroxyethyl)piperazine-N0-2-ethane-
sulfonic acid; HpDmpI, DmpI from Helicobacter pylori; IPTG, isopropyl-b-D-thiog-
alactoside; Kn, kanamycin; LLG, log-likelihood gain; LB, Luria–Bertani; MIF,
macrophage migration inhibitory factor; MSAD, malonate semialdehyde decarbox-
ylase; MPD, 2-methyl-2,4-pentanediol; NCS, non-crystallographic symmetry; 4-OT,
4-oxalocrotonate tautomerase; PPT, phenylpyruvate tautomerase; PEG, polyethyl-
ene glycol; PCR, polymerase chain reaction; PDB, Protein Data Bank; rmsd, root-
mean-square deviation; SDS–PAGE, sodium dodecyl sulfate–polyacrylamide gel
electrophoresis.
1. Introduction
4-Oxalocrotonate tautomerase (4-OT) from Pseudomonas putida
mt-2 catalyzes the conversion of 2-hydroxy-2,4-hexadienedioate
(1, Scheme 1), also known as 2-hydroxymuconate, to 2-oxo-3-hex-
enedioate (2) [1–4]. This enzyme-catalyzed reaction is one step in a
catabolic pathway encoded by the TOL plasmid [1]. Organisms hav-
ing this plasmid, such as P. putida mt-2, can use benzene, toluene,
or xylene isomers as their sole sources of carbon and energy [1].
4-OT has two distinguishing features: a catalytic amino-terminal
proline and a short monomeric subunit (62 amino acids) that codes a
q
This research was supported by the National Institutes of Health Grant GM-
41239 (CPW) and the Robert A. Welch Foundation grants F-1219 (MLH) and F-1334
(CPW).
b-a-b structural motif [3,5–7]. These features are also the defining
qq
characteristics of the tautomerase superfamily [8–10]. The tautom-
erase superfamily is a structurally and mechanistically diverse
group of proteins whose members exist as dimers, trimers, and
homo- and heterohexamers, and catalyze isomerization, hydration,
dehalogenation, and decarboxylation reactions [12–17]. In each of
these transformations, Pro-1 functions as a general base or a general
The atomic coordinates and structure factors of DmpI from Helicobacter pylori
(P21 and P41 space groups) and Archaeoglobus fulgidus have been deposited in the
Protein Data Bank as entries 2ORM, 3M21, and 3M20, respectively.
* Corresponding author. Fax: +1 512 471 8696.
** Corresponding authors. Fax: +1 512 232 2606.
E-mail addresses: m.hackert@mail.utexas.edu (M.L. Hackert), whitman@mail.ute-
xas.edu (C.P. Whitman).
0045-2068/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.bioorg.2010.07.002

J.J. Almrud et al. / Bioorganic Chemistry 38 (2010) 252–259
253
-
-
ter understanding of how diverse functions arose in the tautomer-
ase superfamily.
CO2
OH
CO2
O
2. Materials and methods
-
-
CO2
CO2
2.1. Materials
1
2
All reagents, buffers, and solvents were obtained from Sigma–
Aldrich Chemical Co. (St. Louis, MO) unless noted otherwise. The
syntheses of 2-hydroxymuconate (1, Scheme 1) and 2-hydroxy-
2,4-pentadienoate (3, Scheme 2) are described elsewhere [2,22].
Scheme 1. The 4-OT-catalyzed reaction.
-
-
-
Isopropyl-b-D-thiogalactoside (IPTG) and thin-walled polymerase
CO2
CO2
CO2
chain reaction (PCR) tubes were obtained from Ambion, Inc. (Aus-
tin, TX). Bacterial culture media, PCR reagents, molecular biology
kits and reagents, oligonucleotide primers, and the bacterial strains
used for the construction of the DmpI expression vectors were ob-
tained from the sources listed elsewhere [7,18,23]. The genomic
DNA of H. pylori strain J99 (ATCC 700824D) and the A. fulgidus
strain DSM 4304 (ATCC 49558D) was obtained from the American
Type Culture Collection (Manassas, VA).
O
OH
O
R
R
R
-
-
-
5: R = CO2
6: R = H
1: R = CO2
2: R = CO2
4: R = H
3: R = H
2.2. General methods
OH
O
-
-
Techniques for restriction enzyme digestions, ligation, transfor-
mation, and other standard molecular biology manipulations were
based on methods described elsewhere [24]. DNA sequencing was
done at the University of Texas (Austin) Sequencing Facility. Ki-
netic data and UV absorbance readings were obtained on a Hewlett
Packard 8452A Diode Array spectrophotometer. The kinetic data
were fitted by nonlinear regression data analysis using the Grafit
program (Erithacus Software Ltd., Staines, UK) obtained from Sig-
ma–Aldrich Chemical Co. High-pressure liquid chromatography
was performed on a Waters system using a Bio-Gel Phenyl 5-PW
hydrophobic column or a Pharmacia Superose 12 (HR 10/30) gel
filtration column. Protein was analyzed by tricine sodium dodecyl
sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) under
denaturing conditions on 16% gels using a vertical gel electropho-
resis apparatus obtained from Bio-Rad (Hercules, CA) [25]. Protein
concentrations were determined using the method of Waddell
[26]. The electrostatic surfaces of 4-OT, HpDmpI, and AfDmpI were
analyzed using the Adaptive Poisson–Boltzmann Solver (APBS)
software described elsewhere [27]. The protocol described else-
where was followed in order to use the APBS software within Py-
Mol [28,29].
CO2
CO2
7
8
Scheme 2. Substrates processed by HpDmpI and AfDmpI.
acid catalyst [18–21]. In addition, the monomeric subunit consists of
a single b-a-b unit or two covalently linked b-a-b units.
Sequence analysis identified several 4-OT homologs in a variety
of organisms including the pathogenic bacterium Helicobacter py-
lori and the thermophilic bacterium Archaeoglobus fulgidus [12].
In contrast to 4-OT, these homologs are chromosomally encoded,
and do not appear to be part of metabolic pathways. Moreover,
the host organisms are generally not known to degrade aromatic
hydrocarbons.1 As part of an effort to assign functions for these 4-
OT homologs, they were classified into five different subfamilies
based on the conservation of key mechanistic and structural residues
[12]. DmpI from H. pylori and A. fulgidus (designated HpDmpI and
AfDmpI, respectively) are members of subfamily 3 and 4, respec-
tively. Interestingly, subfamily three members are 4-OT homologs
found in bacteria noted for their disease-causing potential, such as
H. pylori, Yersinia pestis, and Neisseria meningitidis. Subfamily 4 in-
cludes thermophilic bacteria such as A. fulgidus and Methanobacteri-
um thermoautotrophicum. The DmpI homologs are representative of
these two subfamilies and were selected for further characterization.
In this paper, the cloning, overexpression, purification, kinetic
characterization, and crystal structures of HpDmpI and AfDmpI
are reported. Both enzymes have modest 1,3-keto-enol tautomer-
ase activities converting 1 and 2-hydroxy-2,4-pentadienoate (3,
2.3. Construction of expression vectors and enzymology methods
2.3.1. Construction of the HpDmpI and AfDmpI expression vectors
The H. pylori dmpI gene was amplified from genomic DNA by the
PCR using a modification of a procedure described elsewhere
[7,23]. Two oligonucleotides, 50-ACACACACACACCATATGCCGTTTA
TCAATATCAAA-30 (primer-1) and 50-ACACACACACACGGATCCCTAG
TTTTTTTGCCTCAAATG-30 (primer-2), were synthesized as primers
for the PCR. The first primer contains an NdeI restriction site
(underlined) followed by 18 bases that correspond to the coding
sequence of the H. pylori dmpI gene. The second primer contains
a BamHI restriction site (underlined) followed by 21 bases that cor-
respond to the complementary sequence of the H. pylori dmpI gene.
For the PCR amplification of the A. fulgidus dmpI gene, two oligonu-
cleotide primers, 50-AAGAAGAAGGGATCCCATATGCC AGTCCTGATT
GTTTAC-30 (primer-3) and 50-TCCAAGCTTGTCGACTTATTCTCGCTCT
CTATCCGCTAT-30 (primer-4), were used. Primer-3 contains an NdeI
restriction site (underlined) followed by 18 bases corresponding to
the coding sequence of the A. fulgidus dmpI gene. Primer-4 contains
a SalI restriction site (underlined) followed by 24 bases that
Scheme 2) to their b,c-isomers, 5, and 2-oxo-4-pentenoate (6),
respectively, and phenylenolpyruvate (7, Scheme 2) to its keto iso-
mer (8). The crystal structures show that both enzymes are hexa-
mers where the b-a-b fold is conserved in the monomer.
However, the active site pockets differ significantly from those of
the other characterized 4-OT family members [4,5,8,12]. The re-
sults provide a framework for future studies that might identify
biological functions for both proteins and ultimately lead to a bet-
1
A recent hypothesis suggests that some 4-OT family members have a regulatory
rather than a catalytic biological function. The lack of a metabolic context could be
supportive of this hypothesis [Youzhong Guo, PhD Dissertation, The University of
Texas at Austin, 2010].

254
J.J. Almrud et al. / Bioorganic Chemistry 38 (2010) 252–259
correspond to the complementary sequence of the A. fulgidus dmpI
gene. The additional bases in both sets of primers facilitate DNA
cleavage by the restriction enzymes. The PCR was carried out in
a Perkin Elmer DNA thermocycler 480 using template DNA, syn-
thetic primers, and the PCR reagents supplied in the PCR Reagent
system or the Perkin Elmer Cetus GeneAMP kit following a protocol
described elsewhere [7,23]. The resulting gel-purified DNA frag-
ment and the pET24a(+) vector were digested with the appropriate
restriction enzymes, purified, and ligated using T4 DNA following a
previously described protocol [7,23]. Aliquots of the resulting mix-
ture were transformed into competent E. coli JM109 cells and
Crystals of AfDmpI were grown from 20% dioxane and 50% 2-
methyl-2,4-pentanediol (MPD), pH 6.0. This crystallization condi-
tion was prepared by diluting individual stock solutions of 100%
dioxane and 100% MPD directly with distilled H2O. Protein (5
of a 15 mg/mL solution in 50 mM HEPES buffer, pH 7.3) was com-
bined with 5 L of well solution on a Hampton microbridge, and
then equilibrated by vapor diffusion against 500 L of well solution
at 4 °C. Crystals belong to the space group P3221, and have cell con-
lL
l
l
stants a = b = 49.1 Å, c = 118.8 Å and
c = 120.0°. Assuming a trimer
in the asymmetric unit, the Matthews coefficient is 2.5 Å3/Da, and
the solvent content is estimated to be 50% [30].
grown on Luria–Bertani plates with kanamycin (Kn) (100
at 37 °C. Single colonies were chosen at random and grown in li-
quid LB/Kn media (50–100 g/mL). The newly constructed plas-
mids containing the H. pylori or A. fulgidus dmpI genes were
isolated and sequenced. Subsequently, the plasmids were trans-
formed as described elsewhere into E. coli strain BL21(DE3) pLysS
for protein expression [7,23].
lg/mL)
2.4.2. Data collection
l
Diffraction data of crystals of HpDmpI and AfDmpI were col-
lected using a RAXIS-IV image plate detector installed on a Rigaku
RU200H rotating anode X-ray generator. The crystals were
mounted in cryo-loops directly from their mother liquor, and flash
cooled to ꢁ175 °C by immersing the loop directly in liquid nitro-
gen. Crystals were kept at ꢁ165 °C during data collection using a
cryo-cooling device (Molecular Structures Corp., The Woodlands,
TX). Diffraction data were then processed using the programs DEN-
ZO and SCALEPACK [31]. Intensities were converted to structure
factors using the program TRUNCATE of the CCP4 suite [32] and
Rfree reflections assigned using the CCP4 program FREERFLAG.
The crystal and data collection statistics are summarized in
Table 1.
2.3.2. Overexpression and purification of the recombinant HpDmpI and
AfDmpI
The products of the H. pylori and A. fulgidus dmpI genes were
overexpressed using a previously described protocol [7,23]. Typi-
cally, 2 L of culture grown under these conditions yielded 5–7 g
of cells. The enzymes were purified to near homogeneity (>95%
as assessed by SDS–PAGE) using a combination of the Bio-Gel Phe-
nyl 5-PW hydrophobic column and a Sephadex G-75 gel filtration
column (2.5 cm ꢀ 100 cm) as described [7,18,23]. Typically, the
yield of purified enzyme was 11–16 mg per liter of culture.
2.4.3. Structure determination and refinement of HpDmpI
For the P21 data set, the structure of HpDmpI was solved by
molecular replacement using a polyalanine search model of 4-
oxalocrotonate tautomerase (PDB Accession Code 1-BJP) with B-
values set to 40 for all atoms. The cross rotation and translation
functions were calculated with the molecular replacement pro-
gram AMORE [33] using the 30–3 Å resolution data. The correct
rotation and translation function solution had a correlation coeffi-
cient of 60.3% and an R-value of 47.7%. The HpDmpI structure for
the P41 data set was solved using the HpDmpI structure, previously
solved in space group P21 (above) as the search model (B-factors
set to 40 for all atoms). The correlation coefficient and R-factor
for the correct structure were 70.4% and 40.8% respectively.
In both the space groups, the HpDmpI models were refined by
iterations of computer-based minimization, followed by manual
fitting of the models to electron density maps using the program
O [34]. The structure was first refined using REFMAC version 5.0.
[35] with a bulk-solvent correction for all rounds of refinement
and then refined further using CNSsolve version 1.1 [36]. The de-
crease in the Rfree value was used to direct the refinement proce-
dure. Manual rebuilding was carried out using 2Fo–Fc electron
density maps, with Fo–Fc maps being used to monitor model
rebuilding. The HpDmpI model was initially refined with non-crys-
tallographic symmetry (NCS) restraints that were released during
later stages of structural refinement. CNSsolve’s water-picking rou-
tine was used to identify ordered solvent molecules in Fo–Fc elec-
tron density maps. After refinement of the protein and solvent
model using CNSsolve, the R and Rfree values are 21.7% and 25.3%,
respectively for the P21 space group and 23.6% and 27.7% for the
P41 space group. The refinement statistics are listed in Table 2
and the refined coordinates and structure factors deposited with
the Protein Data Bank (PDB) (P21 space group, 2ORM; P41 space
group, 3M21).
2.3.3. Kinetic studies of HpDmpI and AfDmpI
The kinetic studies of HpDmpI and AfDmpI were carried out in
20 mM NaH2PO4 buffer, pH 7.3, at 23 °C. The assays were initiated
by the addition of varying amounts of 1, 3, or 7. The conversion of
1–5, 1–2, 3–6, and 7–8 were monitored as described [12]. For
HpDmpI, the substrate and enzyme concentrations were as fol-
lows: 1 (5–100
150 M, 3.4 nM). For AfDmpI, the substrate and enzyme concen-
trations were as follows: 1 (5–100 M following 1–5 and 30–
M following 1–2, 16 nM); 3 (10–200 M, 8 nM); and 7 (10–
M, 8 nM).
lM, 14 nM); 3 (20–200 lM, 11 nM); and 7 (10–
l
l
500
150
l
l
l
2.4. Crystallography methods
2.4.1. Crystallization of HpDmpI and AfDmpI
Crystals of HpDmpI were grown from 28% polyethylene glycol
(PEG) 400, 200 mM CaCl2, 0.1 M HEPES buffer (pH 7.5), using
NuncÒ microplates (Fisher Scientific, Pittsburgh, PA). Protein
(3
was combined with 3
ibrated by vapor diffusion against 50
l
L of a 20 mg/mL solution in 50 mM NaH2PO4 buffer, pH 7.3)
L of crystallization solution and then equil-
L of well solution at 4 °C.
l
l
Crystals grew to maturity within 4 days and belong to space group
P21 with cell constants a = 41.83 Å, b = 50.77 Å, c = 89.31 Å and
b = 99.1°. Assuming a hexamer per asymmetric unit, the Matthews
coefficient is 2.1 Å3/Da and corresponds to an approximate solvent
content of 42% [30].
Crystals of HpDmpI were also grown from 25% t-butanol in
0.1 M sodium citrate buffer, pH 5.5. Protein (5
solution in 50 mM NaH2PO4 buffer, pH 7.3) was combined with
L of well solution in a Hampton microbridge (Hampton Re-
search, Aliso Viejo, CA) and equilibrated by vapor diffusion against
500 L of well solution at 4 °C. Crystals grew to maturity within
lL of a 20 mg/mL
5
l
l
2.4.4. Structure determination and refinement of AfDmpI
6 weeks, and belong to space group P41 with cell constants
a = b = 53.04 Å and c = 130.88 Å. Assuming a hexamer per asym-
metric unit, the Matthews coefficient is 2.08 Å3/Da, corresponding
to an approximate solvent content of 40% [30].
The crystal structure of AfDmpI was solved by molecular
replacement using a polyalanine search model of 4-OT (PDB Acces-
sion Code 1-BJP) with B-values set to 40 for all atoms (as described
above). The cross rotation and translation functions were calcu-

J.J. Almrud et al. / Bioorganic Chemistry 38 (2010) 252–259
255
Table 1
Crystal and data collection statistics.
HpDmpI
P41
HpDmpI
P21
AfDmpI
P3221
Space group
Cell constants
a (Å)
b (Å)
c (Å)
53.0
53.0
130.88
90.0
90.0
90.0
130
2
41.8
50.8
89.3
90.0
99.1
90.0
130
1.5
49.1
49.1
118.8
90.0
90.0
120.0
160
2
a
(°)
b (°)
(°)
c
Camera length (mm)
Oscillation angle (°)
Exposure time (min)
Resolution range (Å)
Overall
0.5
20
30
30.0–1.85
1.92–1.85
86
25.0–2.10
2.18–2.10
76
24.53–2.29
2.28–2.20
39
Highest-resolution bin
No. of frames
No. of observations
No. of unique reflections
Rmerge (%)
168,340
30,855
120,991
21,852
52,778
8623
Overall
0.044
0.321
0.109
0.490
0.064
0.407
Highest-resolution bin
Completeness (%)
Overall
Highest-resolution bin
I/r(I)
89.3
79.9
10.3
94.3
90.2
5.8
96.4
85.1
14.2
Table 2
Refinement statistics.
Table 3
Kinetic parameters for HpDmpI using 1, 3, and 7a.
HpDmpI
HpDmpI
AfDmpI
Reaction
kcat (sꢁ1
)
Km
(l
M)
kcat/Km (Mꢁ1 sꢁ1
)
Space group
P41
P21
P3221
24.53–2.37
1270
1 ? 5
3 ? 6
7 ? 8
1 ? 2
0.6 0.1
2.5 0.3
2.6 0.7
0.6 0.03
110 20
190 40
230 90
135 15
5.4 ꢀ 103
1.3 ꢀ 104
1.1 ꢀ 104
4.4 ꢀ 103
Refinement resolution range
No. of protein atoms
No. of water molecules
Ramachandran plot
Residues in core (%)
Residues in allowed region (%)
30.0–1.8
2923
126
25.0–2.1
2969
273
20
a
The steady-state kinetic parameters were determined in 20 mM sodium phos-
phate buffer (pH 7.3) at 23 °C. Errors are standard deviations.
94.9
5.1
0.005
0.98
96.7
3.3
0.007
1.13
83.8
16.2
0.008
1.34
Deviation in bond distances (Å)
Deviation in bond angles (deg)
B-factor (Å2)
Overall
Protein
Solvent
Table 4
Kinetic parameters for AfDmpI using 1, 3, and 7a.
36.5
32.6
40.3
23.6
27.7
32.8
28.9
36.6
21.7
25.3
46.8
47.0
48.4
29.6
31.5
Reaction
kcat (sꢁ1
)
Km
(l
M)
kcat/Km (Mꢁ1 sꢁ1
)
R-factor (%)
R-free (%)
1 ? 5
3 ? 6
7 ? 8
1 ? 2
0.2 0.03
0.1 0.02
2.4 0.3
170 40
90 30
90 20
110 10
1.2 ꢀ 103
1.1 ꢀ 103
2.7 ꢀ 104
4.5 ꢀ 102
0.05 0.001
a
lated with the molecular replacement program BEAST [37] using
the 30–3 Å resolution data. The correct rotation and translation
function solution had a log-likelihood gain (LLG) score of 164.37.
The structure of AfDmpI was refined using the procedures de-
scribed above for the HpDmpI structure. After refinement, the R
and Rfree values were 29.6% and 31.5% respectively. The refinement
statistics are listed in Table 2 and the refined coordinates and
structure factors deposited with the PDB (3M20).
The steady-state kinetic parameters were determined in 20 mM sodium phos-
phate buffer (pH 7.3) at 23 °C. Errors are standard deviations.
1–2 by HpDmpI is about 5800-fold less than that reported for 4-OT
[38,39].
AfDmpI also has a modest tautomerase activity with both 1 and
3, and converts them to their b,
respectively). AfDmpI will process 1 to the
c-unsaturated ketones (5 and 6,
a
,b-isomer (2), but
again, the kcat/Km value is much lower (42,000-fold) than that re-
ported for 4-OT [38,39]. The Km values are comparable, but the kcat
for catalyzed-conversion of 1–2 by AfDmpI is ꢂ70,000-fold less
than that reported for 4-OT [38,39]. The most significant AfDmpI
activity was observed for the conversion of 7–8, with 22- and 24-
fold increases in kcat/Km relative to those measured respectively
for 1 and 3.
3. Results
3.1. Kinetic studies of HpDmpI and AfDmpI
Three compounds (1, 3, and 7, Scheme 2) were examined as po-
tential substrates for HpDmpI and AfDmpI and the steady-state
parameters are reported in Tables 3 and 4. HpDmpI has a modest
tautomerase activity with these compounds preferring the mono-
carboxylated substrates (3 and 7) to the dicarboxylated substrate
1, (as indicated by the ꢂ2-fold higher kcat/Km values) (Table 3).
3.2. The crystal structures of HpDmpI
HpDmpI converts 1 to the
a
,b-isomer (2), although the kcat/Km va-
The crystal structure of HpDmpI was determined to 1.8 Å and
2.1 Å resolution in two different space groups (P21 and P41). The
crystallographic asymmetric unit consists of a hexamer in both
lue is 4300-fold less than that reported for 4-OT [38,39]. While the
Km values are comparable, the kcat for the catalyzed-conversion of

256
J.J. Almrud et al. / Bioorganic Chemistry 38 (2010) 252–259
space groups. Electron density for the polypeptide backbone was
well ordered for most of the 67 amino acid residues, except for a
few C-terminal residues. Residues with missing or poorly defined
side chain density were modeled as alanines and reported as such
in their respective PDB files. Pro-14 is in the cis conformation in all
six monomers, and in both space groups. Ramachandran plots for
all non-glycine and non-proline residues for the HpDmpI models
in the P21 and P41 space groups show that 96.7% and 94.9% of the res-
idues lie within the most-favored regions and 3.3% and 5.1% of the
residues lie within the additionally allowed regions. For the P21
space group, the Ca atoms of residues 1–64 in monomer A superim-
pose with the other five monomers of the same hexamer with an
average rmsd value of 0.39 Å ( 0.05 Å). The total of twelve, indepen-
dent HpDmpI monomers for the two structures solved in space
groups P21 and P41 superimpose with an rmsd value of 0.74 Å.
Like the P. putida mt-2 4-OT, the HpDmpI monomer shows the
ences in the C positions are found in loops. The rmsd value
between 4-OT and HpDmpI decreases to 0.76 Å if residues corre-
a
sponding only to the
a-helices and b-sheets of a monomer from
the two proteins are used in the rmsd calculation.
Although the folds of HpDmpI and 4-OT are highly conserved,
the architecture of their active sites is quite different (Figs. 1A,
1C and 2). The key residues in 4-OT have been identified as Pro-
1, Arg-110, Arg-3900, and Phe-500 [7,8,18,38,39]. Sequence align-
ments show that Pro-1 is the only conserved residue. Notable dif-
ferences in the active sites of 4-OT and HpDmpI are observed with
the side-chain placements of Arg-110/Lys-36, Arg-3900/Val-4200, and
Phe-500/Tyr-530 for these key residues (Fig. 2). Differences are also
observed with the side-chain placements of Ile-2/Phe-2, Leu-80/
Val-80, Ile-520/Leu-550, and Ala-570/Val-600 (Fig. 2). These differ-
ences have an impact on the molecular surface topology and elec-
trostatic potential surface of the HpDmpI active site (Fig. 1B).
The 4-OT active site is bound on opposite ends by residues Arg-
110 and Arg-3900. By the molecular 2-fold axis, Arg-39 has its guan-
idinium group positioned next to Arg-3900 from an adjacent active
site (Fig. 1C). This positioning has two identifiable consequences:
the formation of two active sites with clearly defined partitions
from its neighboring active site and strong electropositive charac-
ter at each end of the active site (Fig. 1D). In HpDmpI there is also a
clearly defined partition between the proposed two active sites
(Fig. 1B). Lys-36 in HpDmpI aligns with Lys-32 of macrophage
migration inhibitory factor (MIF) [40], which interacts with a car-
boxylate oxygen of a substrate (p-hydroxyphenylenolpyruvate)
for the phenylpyruvate tautomerase (PPT) activity of MIF and an
inhibitor [(E)-2-fluoro-p-hydroxycinnamate)] of the activity. How-
ever, HpDmpI lacks Arg-3900, and instead contains a valine (i.e., Val-
4200) at the equivalent position. This change results in a loss of elec-
tropositive character at one end and an alteration of the surface
topology of the active site (Fig. 1B and D).
signature b-
1 (b-1), residues 9–15 form a loop connecting b-1 to an
1, residues 16–34), residues 35–37 form a loop that connects
a
-b fold. Accordingly, residues 1–8 comprise a b-strand
-helix (
-1 to
a
a-
a
a 310 helix and turn (38–41 residues). The turn is followed by a sec-
ond b-strand (b-2, residues 42–47), a loop (residues 48–52), a
Type-II b-hairpin (residues 53–59), and ends with a coil conforma-
tion (residues 60–67). Within a single monomer, b-1 and b-2 asso-
ciate in a parallel fashion. The two monomers form a homodimer in
such a way that the b-1 and b-2 strands of one monomer run anti-
parallel with respect to b-1 and b-2 of the second monomer and
form a four-stranded b-sheet. HpDmpI hexamer formation is stabi-
lized by dimer-dimer interactions between a strand from the b-
hairpin turn from one dimer and an edge of the four-stranded
sheet of a second dimer, the result of which is formation of an eight
stranded b-sheet involving four subunits.
3.3. The active site of HpDmpI
3.5. The crystal structure of AfDmpI
The active site cavity is presumed to contain Pro-1. The se-
quence identity with 4-OT and a 4-OT homolog from Bacillus sub-
tilis designated YwhB (34% and 28%, respectively) [4,8] was used
to identify features and boundaries of the active site cavity.
Accordingly, it is proposed that the HpDmpI active site is com-
prised of residues contributed by the convergence of three mono-
mers (monomer, monomer’, and monomer’’)2 at a dimer-dimer
interface within the hexamer. One side of the active site is defined
by Pro-1, Phe-2, Val-42, and the side chain of Lys-36, whereas the
opposite side is defined by Val-4200, Leu-550, Tyr-530, and Val-600
(Fig. 1A). In contrast to 4-OT and YwhB, HpDmpI lacks arginine at
position 11. Instead, Pro-140 (in HpDmpI) diverts the mainchain such
The crystallographic asymmetric unit of AfDmpI consists of a
trimer. The electron density for the polypeptide backbone is well
ordered except for residues 60–62 (monomer-A), 58–62 (mono-
mer-B), and 58–62 (monomer-C). In the Ramachandran plot,
83.8% of all non-glycine and non-proline residues lie within the
most-favored regions and 16.2% lie within the additionally allowed
region. The C atoms for monomer A (1–58) and monomer B super-
a
impose with an rmsd of ꢂ0.55 Å and the C atoms for monomer A
a
and monomer C superimpose with an rmsd of ꢂ0.59 Å.
Like 4-OT and HpDmpI, the AfDmpI monomer shows the b-a-b
fold, but the identities of the residues making up the elements of
secondary structure and their lengths are very different. The b-
strand 1 (b-1) is made up of residues 1–6 and is followed by a loop
(residues 7–12) that connects b-1 to an
30). A loop (residues 31–37) then connects
strand (b-2, residues 38–44), which is followed by a loop (residues
45–48), a Type-II b-hairpin (residues 49–51), and ends with a coil
conformation (residues 52–58). b-1 and b-2 associate in a parallel
fashion within a single monomer. They then associate to form the
hexamer by the antiparallel interaction of b-1 and b-2 of one
monomer with b-1 and b-2 of adjacent monomers to form a
four-stranded b-sheet. This hexamer is stabilized by dimer-dimer
interactions between a strand from the b-hairpin turn from one di-
mer and an edge of the four-stranded sheet contributed by a sec-
ond dimer.
that the side chain of Lys-36 is placed with its e-amino group in the
active site (in a nearly equivalent position). Lys-36 is conserved in
subfamily-3, and may play a role in binding and/or catalysis (vide in-
fra). The architecture of the HpDmpI active site is further defined by
the side chain of Tyr-530, where the phenolic oxygen points into the
active site. The floor of the active site floor is formed, in part, by the
side chain cluster of Val-80, Val-480 (not shown), and Val-600 on one
side of Tyr-530, and Val-4200 and Ile-4400 (not shown) on the other side
of Tyr-530. The 2-fold molecular symmetry axis relating the two di-
mers of the hexamer (e.g., Val-4200 to Val-42) generates an adjacent
active site (Fig. 1A).
a
-helix (
a-1, residues 13–
-1 to a second b-
a
3.4. Structural comparison of HpDmpI with 4-OT
The C atoms of residues 1–60 of the 4-OT and HpDmpI mono-
a
mers superimpose with an rmsd value of ꢂ6.0 Å. The largest differ-
3.6. The active site of AfDmpI
2
Like HpDmpI and all tautomerase superfamily members charac-
terized to date [11], the area around Pro-1 is presumed to be the
The unprimed, primed, and doubly primed residues refer to the different
subunits.

J.J. Almrud et al. / Bioorganic Chemistry 38 (2010) 252–259
257
Fig. 1. The proposed active sites and electrostatic surface potentials respectively of A and B) HpDmpI, showing the loss of electropositive character at one end (where Val-42
replaces Arg-39 found in 4-OT, C and D) 4-OT, showing the strong electropositive character at each end of the active site, and E and F) AfDmpI, showing the lack of strong
positive or negative electrostatic character. The Adaptive Poisson–Boltzmann Solver (APBS) software was used for evaluating the electrostatic properties [27]. The Figures
were prepared with PyMOL using PDB2PQR [28,29].
superimpose with an rmsd of ꢂ3.66 Å where the largest positional
differences are again observed in the various loops. The rmsd be-
tween 4-OT and AfDmpI decreases to 0.60 Å if the C atom posi-
a
tions corresponding to only the monomer’s b-sheets are
calculated from the superposition.
The architecture of the 4-OT and AfDmpI active sites is strik-
ingly different. AfDmpI retains the critical structural residues
(e.g., Lys-16, Glu-44 or Asp-44, Gly-54, and an aromatic residue,
i.e., Tyr, in the Phe-50 position, using the 4-OT numbering system).
Pro-1 is conserved, but AfDmpI does not conserve any other active
site residues (Fig. 3). Differences in the architecture of the active
sites of 4-OT and AfDmpI are especially evident in the side-chain
placements of Arg-110/(side chain of) Met-32, Arg-3900/Thr-3800,
Ile-520/Val-510, Ala-570/Ile-560, and Phe-500/Val-490 (Fig. 3). These
Fig. 2. Superposition of the active site residues of HpDmpI (Pro-1, Lys-36, Val-4200,
and Tyr-530) and the corresponding key catalytic ones in 4-OT (Pro-1, Arg-110, Arg-
3900, and Phe-500). Arg-640 is at the entrance of the active site, but has no 4-OT
equivalent. The Figure was prepared using PyMol [29].
active site cavity for catalysis, if indeed catalysis is the biological
role for this protein.2 This observation along with sequence align-
ment between 4-OT and AfDmpI suggest that the active site of
AfDmpI could also consist of residues contributed by the conver-
gence of three monomers at a dimer–dimer interface within the
hexamer. Hence, it is proposed that one side of the active site is de-
fined by Pro-1, Val-2, Met-32, and Thr-38 whereas the opposite
side is defined by Thr-3800, Leu-4000 (not shown), Val-490, Val-510,
and Ile-560 (Fig. 1E). The 2-fold molecular symmetry axis relating
two dimers of the hexamer (e.g., Thr-3800 to Thr-38) generates an
adjacent active site (Fig. 1E).
3.7. Structural comparison of AfDmpI with 4-OT
Fig. 3. Superposition of the proposed active site residues of AfDmpI (Pro-1, Met-32,
Thr-3800, and Val-490) and the corresponding key catalytic ones in 4-OT (Pro-1, Arg-
110, Arg-3900, and Phe-500). The figure was prepared using PyMol [29].
AfDmpI shares 14% sequence identity with the P. putida 4-OT.
The C atoms (1–59 of 4-OT and 1–59 of the AfDmpI monomer)
a

258
J.J. Almrud et al. / Bioorganic Chemistry 38 (2010) 252–259
differences have an impact on the molecular surface topology and
electrostatic potential surface of the AfDmpI active site (Fig. 1D and
F). The lack of both Arg-110 and Arg-3900, with methionine (i.e., Met-
32) and threonine (i.e., Thr-3800) at the equivalent positions results
in a calculated surface topology for the AfDmpI active site as an
elongated and hydrophobic cavity with an overall lack of positive
(or negative) electrostatic character (Fig. 1F).
[12]. The b-hairpin sequence and the eight following residues in
4-OT are replaced with 26 residues in YdcE. These residues (form-
ing an a-helix, a p-helix, and another a-helix) take the place of the
adjacent dimers in 4-OT in that the helical region fills in the space
occupied by the adjacent dimers in 4-OT. The aromatic amino acid,
Phe-50 in 4-OT, plays an additional role: it creates a local hydro-
phobic region near Pro-1 that may play a role in lowering the
pKa of Pro-1 to 6.4 [18].
In addition to these structural elements, the members of sub-
family-1 conserve three catalytic residues, Pro-1, Arg-110, and
Arg-3900. In 4-OT, Pro-1 has a pKa of ꢂ6.4, and functions as a base
that transfers the 2-hydroxy proton of 1 to C-5 of 2 with a high de-
gree of stereoselectivity [4,6,7]. Arg-110 binds the C-6 carboxylate
group of 1 and may also function as an electron sink to draw elec-
tron density to C-5, thereby facilitating protonation [38,39]. Arg-
3900 interacts with a carboxylate oxygen of C-1, serving as one bind-
ing determinant, and may stabilize the developing carbanionic
character of the intermediate after deprotonation of the 2-hydroxy
group [11,38,39].
In subfamily 2, Arg-39 is replaced with a histidine except in
YwhB, where it is replaced with a valine. YwhB is an efficient
1,3-keto-enol tautomerase (converting 1, 3, and 7 respectively to
5, 6, and 8) in contrast to 4-OT, which is a highly efficient 1,5-
keto-enol tautomerase (using 1, for example) [4]. YwhB also pro-
cesses 1–2, but the kcat/Km value is down ꢂ680-fold. Both the
YwhB-catalyzed 1,3- and 1,5-keto-enol tautomerase activities are
dependent on Pro-1 and Arg-11. However, changing Arg-11 to an
alanine is much more detrimental to these activities than changing
Pro-1 to an alanine.
4. Discussion
Thus far, the tautomerase superfamily consists of five families
[8–10]. The title enzymes of the three founding families of the
superfamily are 4-OT, a bacterial isomerase known as 5-(carboxy-
methyl)-2-hydroxymuconate isomerase (CHMI), and a mammalian
cytokine, macrophage migration inhibitory factor (MIF), which has
a PPT activity [8,9]. Two additional families were uncovered in the
course of the characterization of a bacterial catabolic pathway for
1,3-dichloropropene [11]. The isomeric mixture of 1,3-dichloropro-
pene is converted (in three enzymatic steps) to the cis- and trans-
isomers of 3-chloroacrylate, which, in turn, are processed by iso-
mer-specific dehalogenases (cis- and trans-3-chloroacrylate dehal-
ogenase designated cis-CaaD and CaaD, respectively) to malonate
semialdehyde. Decarboxylation of malonate semialdehyde by mal-
onate semialdehyde decarboxylase (MSAD) yields acetaldehyde,
which is likely routed to the Krebs cycle. MSAD and cis-CaaD rep-
resent separate families in the tautomerase superfamily. CaaD, and
the 4-OT homologs designated YwhB and YdcE, are 4-OT family
members. All of these enzymes have the characteristic b-a-b build-
ing block and Pro-1, which functions as a catalytic base (4-OT,
CHMI, PPT activity of MIF, YdcE, YwhB) or acid (CaaD, cis-CaaD,
MSAD) [4,8–10,12,23].
In 2002, thirty-seven 4-OT homologs (including CaaD) were
identified and categorized into five subfamilies, based on the con-
servation of key structural and catalytic residues [12]. With the re-
sults of this study, members from all five subfamilies have now
been expressed and their properties characterized. 4-OT and YwhB
are members of subfamilies 1 and 2, respectively [12]. HpDmpI is a
These observations and the stereochemical findings (using
mono- and dicarboxylated substrates) suggest that the mechanism
of YwhB parallels that of 4-OT [4,23]. Accordingly, Pro-1 likely has
a low pKa (ꢂ6.4) and functions as a base [4,23]. The mono- and dia-
cids bind in different orientations: the C-1 carboxylate group of a
monoacid substrate interacts with Arg-110 whereas the C-6 carbox-
ylate group of the diacid substrate interacts with Arg-110. In this
orientation, the C-1 carboxylate group interacts with an unknown
residue. As noted above, YwhB has valine in place of the 4-OT Arg-
39 whereas the other subfamily-2 members have a histidine. It is
not yet known how the presence of this charged residue will affect
the 1,3- and 1,5-keto-enol tautomerase activities (assuming these
are the major activities identified for the other subfamily
members).
Subfamily-3 members, such as HpDmpI, lack both Arg-11 and
Arg-39 (or equivalent residues). However, this group contains a
conserved lysine (Lys-36 in HpDmpI) that aligns with Lys-32 of
MIF. Crystallographic studies implicated Lys-32 in the binding of
a substrate (i.e., p-hydroxyphenylenolpyruvate) for the PPT activity
of MIF as well as in the binding of an inhibitor, (E)-2-fluoro-p-
hydroxycinnamate [40,42]. In both cases, an interaction is ob-
served between the oxygen of the C-1 carboxylate group and
Lys-32 [40,42]. However, changing Lys-32 to an alanine has little
effect on the PPT activity of MIF [43]. Moreover, the K32A mutant
is highly stereoselective, (comparable to that of wild type), when
the reaction is carried out in D2O [43]. There is a change in the Ki
value for (E)-2-fluoro-p-hydroxycinnamate and the K32A mutant,
but only an 15-fold increase (2.6–40 mM) [43]. Based on these
observations, Lys-32 is not a significant determinant in the binding
of this substrate or inhibitor.
member of subfamily 3, and AfDmpI and both the
a- and b-sub-
units of CaaD are members of subfamily 4. YdcE, a homolog from
E. coli, is found in subfamily 5. All five homologs function as tau-
tomerases with varying efficiencies.3 Since 2002, the number of se-
quences annotated as 4-OT homologs has increased considerably,
but biological functions for these as well as many of the thirty-seven
original 4-OT homologs have yet to be determined.
This sequence analysis coupled with structural and biochemical
characterization provide significant insight into the structures and
mechanisms of the 4-OT homologs, and the roles played by the
conserved residues. The proteins in subfamilies 1–4 conserve four
structural residues, (Lys-16, Glu-44, Phe-50, and Gly-54, using the
4-OT numbering system), whereas subfamily 5 members do not
[12]. Consequently, members of subfamilies 1–4 form hexamers
(or a heterohexamer, in the case of CaaD) and those in subfamily
5 form dimers. In 4-OT, hexamer formation is governed by pres-
ence of Phe-50 and Gly-54. Gly-54 is at the end of a b-hairpin se-
quence (GXGG) that follows the aromatic amino acid and
produces a tight turn [5,12]. The hairpin interacts with, and ex-
tends the b-sheet of an adjacent 4-OT dimer. In this way, the b-
hairpin stabilizes adjacent dimers in the hexamer. The same motif
is found in HpDmpI (i.e., a tyrosine residue followed by GLGG) and
AfDmpI (a valine residue followed by GVGG), and structural anal-
ysis confirms that HpDmpI and AfDmpI are both hexamers. YdcE
from E. coli and XF1725 from Xylella fastidiosa (both in subfamily
5) lack the b-hairpin sequence following the aromatic residue
Lys-32 in MIF may also be one factor responsible for the low pKa
of Pro-1 (ꢂ5.6) [42]. Pro-1 is located in a hydrophobic cleft where it
is proximal to Lys-66 and Lys-32 [42]. Mutation of Lys-32 to an ala-
nine significantly raises the pKa of Pro-1 (ꢂ1.3 units) in a pH rate
profile, suggesting that the primary function of Lys-32 is to lower
the pKa of Pro-1 [40]. Based on these observations, the combination
of the aromatic amino acid (tyrosine) preceding the b-hairpin se-
3
CaaD, a member of subfamily 4, has a robust PPT activity, converting 7–8 [41].

J.J. Almrud et al. / Bioorganic Chemistry 38 (2010) 252–259
259
[6] J.T. Stivers, C. Abeygunawardana, A.S. Mildvan, G. Hajipour, C.P. Whitman, L.H.
Chen, Biochemistry 35 (1996) 803–813.
quence (GLGG) and Lys-36 (in HpDmpI) may contribute to the pre-
sumed low pKa of Pro-1. The substrate for HpDmpI may be a mono-
acid and a binding role for Lys-36 cannot be excluded.
[7] R.M. Czerwinski, W.H. Johnson Jr., C.P. Whitman, T.K. Harris, C.
Abeygunawardana, A.S. Mildvan, Biochemistry 36 (1997) 14551–14560.
[8] C.P. Whitman, Arch. Biochem. Biophys. 402 (2002) 1–13.
[9] A.G. Murzin, Curr. Opin. Struct. Biol. 6 (1996) 386–394.
[10] G.J. Poelarends, C.P. Whitman, Bioorg. Chem. 32 (2004) 376–392.
[11] G.J. Poelarends, V.P. Veetil, C.P. Whitman, Cell Mol. Life Sci. 65 (2008) 3606–
3618.
[12] J.J. Almrud, A.D. Kern, S.C. Wang, R.M. Czerwinski, W.H. Johnson Jr., A.G.
Murzin, M.L. Hackert, C.P. Whitman, Biochemistry 41 (2002) 12010–12024.
[13] G.J. Poelarends, R. Saunier, D.B. Janssen, J. Bacteriol. 183 (2001) 4269–4277.
[14] S.C. Wang, M.D. Person, W.H. Johnson Jr., C.P. Whitman, Biochemistry 42
(2003) 8762–8773.
[15] G.J. Poelarends, H. Serrano, M.D. Person, W.H. Johnson Jr., A.G. Murzin, C.P.
Whitman, Biochemistry 43 (2004) 759–772.
[16] G.J. Poelarends, W.H. Johnson Jr., A.G. Murzin, C.P. Whitman, J. Biol. Chem. 278
(2003) 48674–48683.
[17] J.J. Almrud, G.J. Poelarends, W.H. Johnson Jr., H. Serrano, M.L. Hackert, C.P.
Whitman, Biochemistry 44 (2005) 14818–14827.
[18] R.M. Czerwinski, T.K. Harris, M.A. Massiah, A.S. Mildvan, C.P. Whitman,
Biochemistry 40 (2001) 1984–1995.
[19] R.M. de Jong, W. Brugman, G.J. Poelarends, C.P. Whitman, B.W. Dijkstra, J. Biol.
Chem. 279 (2004) 11546–11552.
[20] H.F. Azurmendi, S.C. Wang, M.A. Massiah, G.J. Poelarends, C.P. Whitman, A.S.
Mildvan, Biochemistry 43 (2004) 4082–4091.
[21] G.J. Poelarends, H. Serrano, W.H. Johnson Jr., D.W. Hoffman, C.P. Whitman, J.
Am. Chem. Soc. 126 (2004) 15658–15659.
[22] H. Lian, C.P. Whitman, J. Am. Chem. Soc. 115 (1993) 7978–7984.
[23] S.C. Wang, W.H. Johnson Jr., R.M. Czerwinski, C.P. Whitman, Biochemistry 43
(2004) 748–758.
[24] J. Sambrook, E.F. Fritsch, T. Maniatis, Molecular Cloning: A Laboratory Manual,
Cold Spring Harbor Laboratory, Cold Spring Harbor, New York, 1989.
[25] U.K. Laemmli, Nature 227 (1970) 680–685.
[26] W.J. Waddell, J. Lab. Clin. Med. 48 (1956) 311–314.
[27] N.A. Baker, D. Sept, S. Joseph, M.J. Holst, J.A. McCammon, Proc. Natl. Acad. Sci.
USA 98 (2001) 10037–10041.
[28] T.J. Dolinsky, J.E. Nielsen, J.A. McCammon, N.A. Baker, Nucleic Acids Res. 32
(2004) W665–W667.
[29] W.L. DeLano, The PyMOL molecular Graphics System, DeLano Scientific, San
Carlos, CA, 2002. <http://www.pymol.org>.
Interestingly, 4-OT and YwhB show a low level CaaD activity [8].
In both cases, Pro-1 is critical for the activity. Arg-11 is critical for
the CaaD activity of YwhB but has no detrimental effect on that of
4-OT. (In fact, the activity increases slightly in the R11A mutant of
4-OT.) It is not known if subfamilies 3–5 members have low level
CaaD activities (or other low level activities). The characterization
of YwhB provided significant insight into the reaction and mecha-
nism of 4-OT. Likewise, characterization of the activities of these
other 4-OT homologs may provide clues about other activities
and chart an evolutionary trail.
Subfamily-4 is the least understood of the five subfamilies be-
cause the characteristics are not well defined. Other than the N-
terminal proline, the amino acid sequence comparisons do not
identify any common active site residues (except for CaaD)
[13,19]. The catalytic arginine residues found in 4-OT are replaced
by valines. The calculated surface topology for AfDmpI suggests
that the active site is a deep and well-defined cavity devoid of po-
sitive (or negative) electrostatic patches, and varies significantly
from that of 4-OT and HpDmpI.
Initial enzymatic characterization of AfDmpI appears to support
our interpretation of the chemical dissimilarity between AfDmpI
and 4-OT, and the 4-OT homologs characterized to date. AfDmpI
prefers monocarboxylate substrates with the non-carboxylate
end of the molecule being hydrophobic in nature, which is consis-
tent with the identification of 7 as the best substrate screened in
our studies. This observation is also in agreement with our crystal-
lographic observation that the ends of the AfDmpI active site are
hydrophobic in character where the arginine residues are replaced
with valines. It is unclear which group, if any, interacts with the C-
1 carboxylate. The true substrate may not be a mono- or diacid
substrate, but an enol wedged between hydrophobic groups. The
future characterization of members of this subfamily should pro-
vide better understanding of substrate preferences and architec-
tural functionality of the proteins’ active sites.
[30] B.W. Matthews, J. Mol. Biol. 33 (1968) 491–497.
[31] Z. Otwinowski, W. Minor, in: C.W. Carter Jr., R.M. Sweet (Eds.), Methods in
Enzymology, vol. 276, Academic Press, New York, 1997, pp. 307–326.
[32] Collaborative Computational Project, Number 4. (1994) Acta Crystallogr. D 50
(1994) 760-763.
[33] J. Navaza, Acta Crystallogr. A 50 (1994) 157–163.
[34] T.A. Jones, J.-Y. Zou, S.W. Cowan, M. Kjeldgaard, Acta Crystallogr. A 47 (1991)
110–119.
[35] G.N. Murshudov, A.A. Vagin, E.J. Dodson, Acta Crystallogr. D 53 (1997) 240–
255.
Acknowledgment
[36] A.T. Brünger, P.D. Adams, G.M. Clore, W.L. DeLano, P. Gros, R.W. Grosse-
Kunstleve, J.-S. Jiang, J. Kuszewski, M. Nilges, N.S. Pannu, R.J. Read, L.M. Rice, T.
Simonson, G.L. Warren, Acta Crystallogr. D 54 (1998) 905–921.
[37] R.J. Read, Acta Crystallogr. Biol Crystallogr. D 57 (2001) 1373–1382.
[38] R.M. Czerwinski, T.K. Harris, W.H. Johnson Jr., P.M. Legler, J.T. Stivers, A.S.
Mildvan, C.P. Whitman, Biochemistry 38 (1999) 12358–12372.
[39] T.K. Harris, R.M. Czerwinski, W.H. Johnson Jr., P.M. Legler, C.
Abeygunawardana, M.A. Massiah, J.T. Stivers, C.P. Whitman, A.S. Mildvan,
Biochemistry 38 (1999) 12343–12357.
The authors thank Dr. Youzhong Guo (The University of Texas at
Austin) for helpful comments and technical support of this project.
References
[1] S. Harayama, P.R. Lehrbach, K.N. Timmis, J. Bacteriol. 160 (1984) 251–255.
[2] C.P. Whitman, B.A. Aird, W.R. Gillespie, N.J. Stolowich, J. Am. Chem. Soc. 113
(1991) 3154–3162.
[3] L.H. Chen, G.L. Kenyon, F. Curtin, S. Harayama, M.E. Bembenek, G. Hajipour, C.P.
Whitman, J. Biol. Chem. 267 (1992) 17716–17721.
[4] S.C. Wang, W.H. Johnson Jr., R.M. Czerwinski, S.L. Stamps, C.P. Whitman,
Biochemistry 46 (2007) 11919–11929.
[5] H.S. Subramanya, D.I. Roper, Z. Dauter, E.J. Dodson, G.J. Davies, K.S. Wilson, D.B.
Wigley, Biochemistry 35 (1996) 792–802.
[40] A.B. Taylor, W.H. Johnson Jr., R.M. Czerwinski, H.S. Li, M.L. Hackert, C.P.
Whitman, Biochemistry 23 (1999) 7444–7452.
[41] G.J. Poelarends, W.H. Johnson Jr., H. Serrano, C.P. Whitman, Biochemistry 46
(2007) 9596–9604.
[42] J.B. Lubetsky, M. Swope, C. Dealwis, P. Blake, E. Lolis, Biochemistry 38 (1999)
7346–7354.
[43] W.H. Johnson Jr., R.M. Czerwinski, S.L. Stamps, C.P. Whitman, Biochemistry 38
(1999) 16024–16033.