Inactivation of the phenylpyruvate tautomerase activity of macrophage migration inhibitory factor by 2-oxo-4-phenyl-3-butynoate

Article abstract of DOI:10.1016/j.bioorg.2006.05.001

Macrophage migration inhibitory factor (MIF) is an important immunoregulatory protein that has been implicated in several inflammatory diseases. MIF also has a phenylpyruvate tautomerase (PPT) activity, the role of which remains elusive in these biological activities. The acetylene compound, 2-oxo-4-phenyl-3-butynoate (2-OPB), has been synthesized and tested as a potential irreversible inhibitor of its enzymatic activity. Incubation of the compound with MIF results in the rapid and irreversible loss of the PPT activity. Mass spectral analysis established that the amino-terminal proline, previously implicated as a catalytic base in the PPT-catalyzed reaction, is the site of covalent modification. Inactivation of the PPT activity likely occurs by a Michael addition of Pro-1 to C-4 of the inhibitor. Attempts to crystallize the inactivated complex to confirm the structure of the adduct on the covalently modified Pro-1 by X-ray crystallography were not successful. Nor was it possible to unambiguously interpret electron density observed in the active sites of the native crystals soaked with the inhibitor. This may be due to crystal packing in that the side chain of Glu-16 from an adjacent trimer occupies one active site. However, this crystal contact may be partially responsible for the high-resolution quality of these MIF crystals. Nonetheless, because MIF is a member of the tautomerase superfamily, a group of structurally homologous proteins that share a β-α-β structural motif and a catalytic Pro-1, 2-OPB may find general use as a probe of tautomerase superfamily members that function as PPTs.

Full text of DOI:10.1016/j.bioorg.2006.05.001

Bioorganic Chemistry 34 (2006) 183–199
www.elsevier.com/locate/bioorg
Inactivation of the phenylpyruvate tautomerase
activity of macrophage migration inhibitory factor
q
by 2-oxo-4-phenyl-3-butynoate
a
b
Pavel A. Golubkov , William H. Johnson Jr. ,
b
c
b
Robert M. Czerwinski , Maria D. Person , Susan C. Wang ,
b,
a,
*
Christian P. Whitman *, Marvin L. Hackert
a
Department of Chemistry and Biochemistry, University of Texas at Austin, Austin, TX 78712, USA
Division of Medicinal Chemistry, College of Pharmacy, University of Texas at Austin, Austin, TX 78712, USA
b
c
Division of Pharmacology and Toxicology, College of Pharmacy, University of Texas at Austin,
Austin, TX 78712, USA
Received 23 March 2006
Available online 14 June 2006
Abstract
Macrophage migration inhibitory factor (MIF) is an important immunoregulatory protein that
has been implicated in several inflammatory diseases. MIF also has a phenylpyruvate tautomerase
(PPT) activity, the role of which remains elusive in these biological activities. The acetylene com-
pound, 2-oxo-4-phenyl-3-butynoate (2-OPB), has been synthesized and tested as a potential irrevers-
ible inhibitor of its enzymatic activity. Incubation of the compound with MIF results in the rapid
and irreversible loss of the PPT activity. Mass spectral analysis established that the amino-terminal
proline, previously implicated as a catalytic base in the PPT-catalyzed reaction, is the site of covalent
modification. Inactivation of the PPT activity likely occurs by a Michael addition of Pro-1 to C-4 of
the inhibitor. Attempts to crystallize the inactivated complex to confirm the structure of the adduct
on the covalently modified Pro-1 by X-ray crystallography were not successful. Nor was it possible
to unambiguously interpret electron density observed in the active sites of the native crystals soaked
with the inhibitor. This may be due to crystal packing in that the side chain of Glu-16 from an
q
This research was supported by the National Institutes of Health Grant GM-65324 (C.P.W. and M.L.H.) and
the Robert A. Welch Foundation (F-1219 to M.L.H.). SCW is a Fellow of the American Foundation for
Pharmaceutical Education.
*
Corresponding authors. Fax: +1 512 471 6135 (M.L. Hackert), +1 512 232 2606 (C.P. Whitman).
E-mail addresses: whitman@mail.utexas.edu (C.P. Whitman), m.hackert@mail.utexas.edu (M.L. Hackert).
0045-2068/$ - see front matter ꢀ 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.bioorg.2006.05.001

184
P.A. Golubkov et al. / Bioorganic Chemistry 34 (2006) 183–199
adjacent trimer occupies one active site. However, this crystal contact may be partially responsible
for the high-resolution quality of these MIF crystals. Nonetheless, because MIF is a member of the
tautomerase superfamily, a group of structurally homologous proteins that share a b–a–b structural
motif and a catalytic Pro-1, 2-OPB may find general use as a probe of tautomerase superfamily mem-
bers that function as PPTs.
ꢀ 2006 Elsevier Inc. All rights reserved.
Keywords: Enzyme; Inactivation; MIF; Acetylene compound
1. Introduction
Macrophage migration inhibitory factor (MIF)1 was first identified in the 1960s as a
protein involved in delayed-type hypersensitivity responses [1–3]. Since its initial discov-
ery, MIF has been implicated in several inflammatory and autoimmune diseases, including
arthritis [4–7], glomerulonephritis [8], and Gram-positive and Gram-negative sepsis [9,10].
The pro-inflammatory properties of MIF are coupled with an ability to counter the effect
of corticosteroids [9,11]. Thus, MIF inhibitors are of considerable interest as potential
anti-inflammatory agents. MIF also exhibits a phenylpyruvate tautomerase (PPT) activity,
catalyzing the ketonization of the enol isomers of phenylpyruvate (1, Scheme 1) [12] and
p-hydroxyphenylpyruvate (2) to the respective keto isomers (3 and 4) [13]. Despite intense
study, cellular receptors for MIF and the relevance of the enzymatic activity to the biolog-
ical activities remain unknown.
Three-dimensional X-ray crystallographic studies of human and murine MIF demon-
strate that MIF exists as a homotrimer and is structurally related to the bacterial 4-oxalo-
crotonate tautomerase (4-OT) and 5-carboxymethyl-2-hydroxymuconate isomerase
(CHMI) [14,15] despite low (<20%) sequence identity with these enzymes. Thus, MIF is
a member of the tautomerase superfamily, a group of structurally homologous proteins
characterized by a common b–a–b fold and a catalytic amino-terminal proline [16]. Pro-
1 has been implicated as a critical residue in the PPT activity [13,17].
The acetylene compounds, 2-oxo-3-pentynoate (2-OP, 5, Scheme 2) and 2-oxo-3-
butynoate (2-OB, 6), are active-site-directed, irreversible inhibitors of 4-OT [18,19]. Both
compounds form a covalent bond with the amino-terminal proline, which functions as the
general base in the 4-OT-catalyzed reaction. More significantly, crystallographic analysis
of 4-OT inactivated by 2-OP provided much of the information about the positioning of
active site residues and resulted in a proposed mechanism for the 4-OT-catalyzed reaction.
The structural and functional similarities between 4-OT and MIF, which also uses Pro-1 as
the catalytic base in the PPT-catalyzed reaction, suggested that the acetylene analogue of
1
Abbreviations used: APS, advanced photon source; ARP, automated refinement procedure; CHMI,
5-carboxymethyl-2-hydroxymuconate isomerase; cis-CaaD, cis-3-chloroacrylic acid dehalogenase; ESI-MS,
electrospray ionization mass spectrometry; Fo and Fc, observed and calculated structure factors, respectively;
HPLC, high pressure liquid chromatography; Kn, kanamycin; MIF, macrophage migration inhibitory factor;
MSAD, malonate semialdehyde decarboxylase; MALDI-PSD, matrix assisted laser desorption–ionization post-
source decay; MALDI-TOF, matrix assisted laser desorption–ionization time-of-flight; NCS, non-crystallo-
graphic symmetry; 4-OT, 4-oxalocrotonate tautomerase; 2-OPB, 2-oxo-4-phenyl-3-butynoate; 2-OP, 2-oxo-3-
pentynoate; 2-OB, 2-oxo-3-butynoate; PPT, phenylpyruvate tautomerase; rmsd, root-mean-square deviation;
SDS–PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis.

P.A. Golubkov et al. / Bioorganic Chemistry 34 (2006) 183–199
185
O
OH
-
-
CO2
CO2
R
R
1: R = H
3: R = H
4: R = OH
2: R = OH
Scheme 1.
O
R
-
CO2
5: R = CH3
6: R = H
7: R = Phenyl
Scheme 2.
5, 2-oxo-4-phenyl-3-butynoate (2-OPB, 7, Scheme 2) might be an irreversible inhibitor of
MIF and a comparable diagnostic probe for exploring the enzymatic reactions as well as
the biological activities. Hence, 2-OPB was synthesized and evaluated as a potential irre-
versible inhibitor of MIF. In addition, a structure of MIF crystals soaked with 7 has been
˚
determined to 1.45 A resolution. The results are reported herein.
2. Materials and methods
Materials. All reagents, buffers, and solvents were obtained from Sigma–Aldrich Chem-
ical Co. (St Louis, MO) with the following exceptions. The syntheses of tert-butyl glyoxy-
late (10) and (E)-2-fluoro-p-hydroxycinnamate (13) have been described [18,20].
Recombinant murine MIF was purified as described previously [15]. Isopropyl-b-D-thiog-
alactoside (IPTG) and thin-walled PCR tubes were obtained from Ambion, Inc. (Austin,
TX). Pre-packed PD-10 Sephadex G-25 columns were purchased from Biosciences AB
(Uppsala, Sweden). Bacterial culture media, PCR reagents, molecular biology kits and
reagents, oligonucleotide primers, and the bacterial strains used for the construction of
the expression vector were obtained from the sources listed elsewhere [21].
Methods. Techniques for restriction enzyme digestions, ligation, transformation, and
other standard molecular biology manipulations were based on methods described else-
where [22]. DNA sequencing was done at the University of Texas (Austin) Sequencing
Facility. Kinetic data and UV absorbance readings were obtained on a Hewlett–Packard
8452A Diode Array spectrophotometer. The kinetic data were fitted by nonlinear regres-
sion data analysis using the Grafit program (Erithacus Software Ltd., Staines, UK)
obtained from Sigma–Aldrich Chemical Co. HPLC was performed on a Waters 501/510
system using a TSKgel DEAE-5PW (150 · 21.5 mm) column (Tosoh Bioscience, Mont-
gomeryville, PA). Protein was analyzed by tricine sodium dodecyl sulfate-polyacrylamide
gel electrophoresis (SDS–PAGE) under denaturing conditions on 16% gels on a vertical
gel electrophoresis apparatus obtained from Bio-Rad (Hercules, CA) [23]. Protein concen-
trations were determined using the method of Waddell [24]. NMR spectra were obtained

186
P.A. Golubkov et al. / Bioorganic Chemistry 34 (2006) 183–199
on a Varian UNITY-plus 300 MHz spectrometer. Chemical shifts were referenced as not-
ed below. Mass spectral data were obtained on an LCQ electrospray ion-trap mass spec-
trometer (ThermoFinnigan, San Jose, CA) or a delayed extraction Voyager-DE PRO
matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) instrument
(PerSeptive Biosystems, Framingham, MA), as indicated below. Both instruments are
housed in the Analytical Instrumentation Facility Core in the College of Pharmacy at
the University of Texas at Austin.
2.1. Chemical syntheses
Preparation of tert-butyl 2-hydroxy-4-phenyl-3-butynoate (11). Methylmagnesium bro-
mide (1.05 equiv.) in ether was added dropwise over a 20 min period to a stirring solution
of phenylacetylene (8, 2.4 g, 23.1 mmol) dissolved in 15 mL of anhydrous THF at room
temperature and under argon. The reaction mixture was warmed and gently refluxed.
After stirring overnight, the reaction mixture was cooled to ꢀ78 ꢁC. Subsequently, tert-bu-
tyl glyoxylate (10, 3.0 g, 23.1 mmol) dissolved in 7 mL THF was added rapidly. The reac-
tion mixture was allowed to warm to room temperature over a 2 h period and added to a
10% NaH2PO4 solution (200 mL). The resulting mixture was extracted with a mixture of
hexanes and ethyl acetate (2:1, 3 · 200 mL). The organic layers were pooled, dried over
anhydrous Na2SO4, and concentrated to dryness to give crude product, 11, as a yellow
liquid (ꢁ6.1 g). The crude product was used in the next step without further purification.
Preparation of tert-butyl 2-oxo-4-phenyl-3-butynoate (12). To a solution of 11 in 50 mL
of anhydrous CH2Cl2 was added activated MnO2 (25 g) in small portions. The resulting
reaction mixture was stirred overnight at room temperature and then stirred for another
night at 6 ꢁC. Subsequently, the reaction mixture was filtered through a silica gel plug col-
umn in order to remove the MnO2 and the filtrate was concentrated under reduced pres-
sure. The residue was subjected to flash chromatography (8:1 hexanes, ethyl acetate). The
1
UV active fractions were collected and concentrated to give a yellow solid (3.0 g). H
NMR (CDCl3, 300 MHz): d 1.61 (9H, s), 7.42 (2H, t, m-H), 7.52 (1H, triplet of triplets,
p-H), 7.66 (2H, dd, o-H). 13C NMR (CDCl3, 75.5 MHz): d 28.0 (CH3 of t-butyl), 87.0
(C-4), 85.2 (C of t-butyl), 97.6 (C-3), 119.5 (Ca), 129.0 (Cc), 131.9 (Cd), 133.9 (Cb),
158.4 (C-1), 171.1 (C-2). HRMS m/z calculated for C10H7O3 (MH+) 231.1021, found
231.1021.
Preparation of 2-oxo-4-phenyl-3-butynoate (7). The tert-butyl group was removed by a
literature procedure [18]. Accordingly, 12 (0.33 g, 1.4 mmol) was treated with trifluoroace-
tic acid as described [18] to yield 7 (0.25 g) in 100% yield. 1H NMR (CDCl3, 300 MHz): d
7.47 (2H, t, m-H), 7.55 (1H, triplet of triplets, p-H), 7.69 (2H, dd, o-H). 13C NMR (CDCl3,
75.5 MHz): d 86.8 (C-4), 101.4 (C-3), 119.0 (Ca), 129.1 (Cc), 132.5 (Cd), 134.3 (Cb), 159.4
(C-1), 170.6 (C-2). HRMS m/z calculated for C10H7O3 (MH+) 175.0395, found 175.0400.
2.2. Site-directed mutagenesis and enzymology methods
Construction of the E16A MIF-mutant. The E16A-MIF mutant was constructed by
overlap extension PCR as described elsewhere [25]. The external primers for the construc-
tion of the mutant were oligonucleotides 50-GCG GAT AAC AAT TCC CCT CT-30
(primer A) and 50-CTC AGC TTC CTT TCG GGC TT-30 (primer D). Primer A corre-
sponds to a region 36 bases upstream of the NdeI restriction site in the pET-24a(+)

P.A. Golubkov et al. / Bioorganic Chemistry 34 (2006) 183–199
187
plasmid, while primer D corresponds to the complementary sequence of a region 20 bases
downstream of the His-tag region in the pET-24a(+) plasmid. The internal primers were
50-CAG AAA CCC CGC TGG CAC G-30 (primer B) and 50-C GTG CCA GCG GGG
TTT CTG-30 (primer C). In primers B and C, the mutation is underlined. The remaining
bases correspond to the nucleotide coding sequence (primer C) or the complementary
sequence (primer B) of the gene.
For the construction of the AB and CD fragments, each reaction mixture contained
1 lL of a 20 lM solution of primers A and B or primers C and D, 5 lL 10· high fidelity
buffer, 1 lL 10 mM dNTPs, 2 lL 50 mM MgSO4, 5 lL of a 1:500 dilution of pET-11b
MIF, 0.4 lL of platinum Taq high fidelity, and a sufficient quantity of sterile water to
make a final volume of 50 lL. The PCR protocol consisted of an initial 45-s incubation
period at 94 ꢁC, followed by 35 cycles of the PCR where each cycle contained a denatur-
ation step at 94 ꢁC for 30 s, an annealing step at 50 ꢁC for 45 s, and an elongation step at
68 ꢁC for 2 min, and a final 10-min incubation period at 68 ꢁC. The PCR products were
purified by gel electrophoresis followed by use of the Ultrafree-DA centrifugal units.
The AD fragment was constructed similarly except that the AB and CD fragments
(2 lL of each) were used as template. The AD fragment was electrophoresed, excised,
and purified using the Ultrafree-DA centrifugal unit. Restriction digests of the AD frag-
ment and the pET-24a(+) vector were performed with the NdeI and BamHI enzymes as
described elsewhere [22]. The digested fragments were purified by electrophoresis followed
by use of the Ultrafree-DA units and ligated using the Quick Ligation Kit (New England
Biolabs). The DNA was purified by ethanol precipitation and used to transform Escherich-
ia coli strain BL21-Gold(DE3)pLysS by electroporation according to the manufacturer’s
instructions. Transformed cells were plated onto LB/Kn (50 lg/mL) and incubated over-
night at 37 ꢁC. A randomly selected colony, designated pET-24a-E16A-MIF2, had the
correct sequence for the desired mutation with no other changes.
Expression and purification of the E16A-MIF mutant. The E16A-MIF mutant was
grown and purified under the conditions described elsewhere [15] except the Sephadex
G-75 column was equilibrated and eluted using 20 mM NaH2PO4 buffer, pH 7.3. Under
these conditions, 3 L of culture yields ꢁ5 g of cells and ꢁ95 mg of protein purified to near
homogeneity (ꢁ95% as assessed by SDS–PAGE).
Kinetics of irreversible inhibition of MIF by 2-OPB (7). The inhibition of the enzymatic
activity of MIF by 7 was carried out at 0 ꢁC as follows. Incubation mixtures (total vol-
ume = 1.0 mL) containing different concentrations of 7 (0–100 lM from stock solutions)
and MIF (48 lM based on subunit molecular mass) in 50 mM NaH2PO4 buffer (pH
6.0) were made up in 1.5 mL Eppendorf micro test tubes. Aliquots were removed at var-
ious time intervals (0, 5, 10, 15, 30, 60, and 120 s) and the residual PPT activity was
assayed following the disappearance of the enol isomer of phenylpyruvic acid (1,
2.5 mM) at 288 nm as described [15].
Protection of the PPT activity from inactivation by 13. The competitive inhibitor, (E)-2-
fluoro-p-hydroxycinnamate (13) [15,20], was pre-incubated with MIF in order to protect
the PPT activity from inactivation by 7. Accordingly, various concentrations of 13 (0–
250 lM) were incubated with MIF (50 lM based on subunit molecular mass) in 50 mM
NaH2PO4 buffer (1 mL, pH 6.8) for 5 min. Subsequently, 7 (60 lM) was added to the mix-
ture. Aliquots (5 lL) were removed at various time intervals (0, 5, 10, 15, 30, 60, and 120 s)
and added to the boric acid assay mixture (1 mL) in order to stop the reaction. The com-
ponents of this mixture and the assay are described elsewhere [20].

188
P.A. Golubkov et al. / Bioorganic Chemistry 34 (2006) 183–199
F
-
CO2
HO
13
Irreversibility of the inactivation. MIF (0.4 lmol based on subunit molecular mass) and
7 (4 lmol) were incubated in 50 mM NaH2PO4 buffer (final volume 10 mL, pH 6.8) for
5 min at 4 ꢁC. In a separate control reaction, the same quantity of MIF was incubated
without 7 under otherwise identical conditions. Both samples were dialyzed against
50 mM NaH2PO4 buffer (pH 6.8) for 68 h. Aliquots were removed and assayed for resid-
ual activity using the enol isomer of phenylpyruvic acid (1) as described [15].
Mass spectral analysis of modified MIF. To a quantity of MIF (2 mg, 0.16 lmol based
on subunit molecular mass) in 1 mL of 20 mM NaH2PO4 (pH 7.3) on ice were added 3 5-
lL portions of 7 (5.7 mM solution, 0.03 lmol per aliquot) at 5 min intervals. There was no
activity (as determined following the disappearance of 1) after the addition of the third ali-
quot. The incubation mixtures containing the inactivated MIF or the control sample
(incubated without 7) were loaded onto separate PD-10 Sephadex G-25 gel filtration col-
umns, which had previously been equilibrated with 10 mM NH4HCO3 buffer (pH 8.0) as
described [26]. The mixtures were processed and the resulting proteins analyzed by electro-
spray ionization mass spectrometry (ESI-MS) as described (26). The same samples were
used in the peptide mapping experiments described below.
Peptide mapping. A sample of MIF or MIF modified by 7 (after elution from the gel
filtration column) was treated with sequencing grade trypsin as described [26]. Subsequent-
ly, the trypsin treated samples were analyzed by matrix assisted laser desorption–ioniza-
tion time-of-flight (MALDI-TOF) mass spectrometry [26]. Selected ions in the samples
were subjected to MALDI-post-source decay (PSD) analysis using the protocol described
elsewhere [26,27].
2.3. Crystallography methods
Crystallization. Crystals of murine MIF were grown by vapor diffusion from 24 to 30%
PEG 8000 in 50 mM NaH2PO4 buffer at pH 7.0, 5 lL (20 mg/mL) MIF, and 5 lL of well
solution per droplet. All of the crystals from one hanging drop (ꢁ0.1 mg protein) were
transferred into 200 lL of 30% PEG 8000, 50 mM NaH2PO4 buffer at pH 7.0, and soaked
in a 2-fold excess (ꢁ0.1 mM) of 7 for 40 h.
Data collection. Crystals were cryoprotected with 30% PEG 8000 and flash-cooled in
liquid nitrogen. The crystal used for data collection was maintained at 100 K in a gaseous
nitrogen stream provided by an Oxford Cryo-Jet system. X-ray diffraction intensities were
˚
measured to 1.45 A resolution with a Quantum 4 CCD detector using synchrotron radia-
tion at the advanced photon source (APS) beamline 14-BMD (BioCARS). The energy
˚
used was 12.660 keV (0.97935 A). The crystal-to-detector distance was 120 mm. A total
of 80 frames were collected with a 1ꢁ interval at an exposure of 45 s per frame. The inten-
sity data were collected from a single crystal, scaled, and merged at the APS using the
HKL-2000 program package [28]. The data collection statistics are summarized in Table 1.
Structure determination and refinement. The space group is P63 with a = 94.71 and
˚
c = 87.67 A. Assuming one trimer per asymmetric unit, the Matthews’ coefficient [29] is
3
˚
3.3 A /Da and the implied solvent content is 63% [30]. The model used for molecular

P.A. Golubkov et al. / Bioorganic Chemistry 34 (2006) 183–199
189
Table 1
Data collection and refinement statistics
Data-collection statistics
Space group
P63
Unit-cell parameters
˚
a = b (A)
c (A)
94.71
87.67
100
0.9794
50.0–1.45 (1.49–1.45)
˚
Temperature (K)
Wavelength (A)
Resolution range (A)
˚
˚
Rmerge (%)
4.7
Completeness (%)
ÆI/r(I)æ
97.4 (96.8)
10.3
No. measured reflections
No. unique reflections
No. >3r
676,784
78,898
76,875
0.376
Mosaicity (ꢁ)
Refinement statistics
No. working reflections
No. test set reflections
Rwork
73,003 (5323)
3872 (278)
0.14 (0.22)a
0.18 (0.30)
2606
Rfree
Protein atoms
Water molecules
465
Geometry statistics
˚
rmsd, bond lengths (A)
0.05
3.57
22.1
17.2
26.0
1.11
3.12
0.975
0.897
rmsd, bond angles (ꢁ)
Average B value for all atoms (A )
2
˚
Average B value for main chain
Average B value for side chain and waters
Average rms B for 342 main-chain residues
Average rms B for 288 side-chain residues
Overall correlation coefficient
Overall figure of merit
a
˚
X-ray diffraction data were collected to a low-resolution limit of 50.0 A, but the refinement was carried out
˚
with the data between 23.8 and 1.45 A resolution.
replacement was that of the Y95F mutant of MIF (PDB entry 1MFF) [31]. The program
suite CCP4 (Collaborative Computational Project, Number 4) [32] using the graphic user
interface ccp4i [33] for MOLREP and AMORE [34] was used for obtaining rotational and
translational parameters. The best molecular replacement solution had an R value of 0.354
and a correlation coefficient of 0.675. REFMAC5 was used for maximum-likelihood
refinement [35]. The initial round of refinement, conducted with non-crystallographic sym-
metry (NCS) constraints, geometric restraints, and translation, libration, screw (TLS)
groups [36], lowered R to 0.27 and Rfree to 0.29. The graphics program O [37] was used
to adjust the model to fit the resulting electron density map generated by FFT [38]. The
high quality of the electron density map was immediately apparent as the map clearly
showed the tyrosine at position 95 when using phases derived from the Y95F mutant mod-
el. In later stages of refinement NCS constraints were removed and the B-factors were
refined anisotropically. Solvent molecules were added by a combination of automatic

190
P.A. Golubkov et al. / Bioorganic Chemistry 34 (2006) 183–199
identification by a routine in automated refinement procedure ARP/wARP [39] and man-
ual insertions in the program O using O macro written by Gerard Kleywegt. The refined
structure contains 2606 protein atoms and 465 water molecules with R and Rfree of 0.14
and 0.18, respectively, with excellent geometry as evidenced by the fact that 100% of all
non-glycine, non-proline residues fall within the most-favored (92.4%) or additionally
allowed (7.6%) regions of the Ramachandran plot. The refinement statistics are summa-
rized in Table 1.
3. Results
Synthesis of 2-oxo-4-phenyl-3-butynoate (7). The synthesis of 7 is outlined in Scheme 3
and is analogous to the procedure used for the synthesis of 2-OP [19]. Accordingly, the
magnesium bromide salt of phenylacetylene was generated in situ from the commercially
available compounds, methylmagnesium bromide and phenylacetylene (8). Subsequent
condensation of 9 with the tert-butyl ester of glyoxylate (10) produced 11. Oxidation of
11 by activated MnO2 yielded 12 as the major product after flash chromatography. The
tert-butyl group was removed by anhydrous trifluoroacetic acid to result in the formation
1
of 7 as the free acid. The H and 13C NMR spectra and mass spectral data are consistent
with the structure of 7.
Construction and characterization of the E16A mutant of MIF. The E16A mutant of
MIF was constructed by overlap extension PCR [25], expressed in E. coli strain
BL21(DE3)pLysS, and purified to >95% (as determined by SDS–PAGE) using a modifi-
cation of a previously described procedure [15]. The DNA sequence of the mutant was
confirmed by DNA sequencing. The steady-state kinetic parameters for the PPT activity
of the E16A mutant were measured using 1 and 2 as substrates, and compared to the
parameters for the PPT activity of wild-type MIF [31]. The PPT activity of E16A-MIF
converts 1 to 3 (Scheme 1) with a Km = 160 40 lM, a kcat = 200 30 sꢀ1, and a kcat
Km = 1.25 · 106 Mꢀ1 sꢀ1. These values are comparable to those measured for the PPT
activity of wild-type MIF (Km = 150 40 lM, kcat = 320 80 sꢀ1
and kcat
Km = 2.1 · 106 Mꢀ1 sꢀ1) [31]. The PPT activity of E16A-MIF converts 2 to 4 (Scheme
1) with a Km = 100 10 lM, a kcat = 60 5 sꢀ1, and a kcat/Km = 6.0 · 105 Mꢀ1 sꢀ1
/
,
/
.
These values are also comparable to those measured for the PPT activity of wild type
MIF (Km = 200 70 lM, kcat = 140 20 sꢀ1, and kcat/Km = 7.0 · 105 Mꢀ1 sꢀ1) [31].
Hence, for both substrates, the kinetic parameters for the PPT activity of E16A-MIF
CH3MgBr
HCOCO2But
10
+
H
MgBr
8
9
OH
O
CO2But
O
MnO2
TFA
H
CO2But
CO2H
11
12
Scheme 3.
7

P.A. Golubkov et al. / Bioorganic Chemistry 34 (2006) 183–199
191
are not significantly different from those measured for the wild type. These results are con-
sistent with the observation that Glu-16 is not in the active site region.
Irreversible inhibition of MIF by 2-OPB (7). The inactivation of the PPT activity of
MIF by 7 appears to be a biphasic process where the inactivation occurs primarily within
the first few seconds (data not shown). Moreover, the loss of activity is nearly stoichiom-
etric (based on the subunit molecular mass of MIF). For example, at a ꢁ0.5:1 ratio of
inhibitor to protein, about 40% of the activity is lost within the first 5 s. At a ꢁ1:1 ratio
of inhibitor to protein, nearly 70% of the activity is lost within the first 5 s. At a ꢁ2:1 ratio
of inhibitor to protein, nearly 85% of the activity is lost within the first 5 s. Binding at the
active site is suggested by the observation that 13 partially protects against inactivation. In
the presence of a 250 lM solution of 13, only ꢁ50% of the PPT activity is lost in 120 s even
though MIF was incubated with ꢁ1.2 equiv. of 7. Dialysis (68 h) does not result in signif-
icant reactivation of the PPT activity, consistent with the formation of a covalent bond
between the enzyme and the adduct resulting from 7.
Identification of the residue modified by 7. MIF was titrated with aliquots of 7 in order to
determine the site of modification that results in the loss of PPT activity. After the addition
of the third aliquot, corresponding to a ratio of ꢁ0.6:1 inhibitor to protein, there was no
detectable PPT activity.2 The inactivated MIF and a control sample were purified by gel
filtration chromatography and analyzed by ESI-MS. ESI-MS analysis of the untreated
MIF showed a subunit molecular mass of 12,370.0 ( 6) Da, which is comparable to the
expected subunit molecular mass of 12,373.1 Da (average mass). ESI-MS analysis of the
inactivated MIF showed a subunit molecular mass of 12,544.0 Da. The difference between
these masses (174.4 Da) indicates a single site of modification by one unit of a species with
a mass of 174 6 Da.
The site of the covalent modification was identified by subjecting MIF and the inacti-
vated MIF to proteolytic digestion by trypsin and analyzing the resulting peptide mixtures
by MALDI-TOF mass spectrometry (Fig. 1 and Table 2). The modified residue was deter-
mined by a comparative analysis of the spectra followed by selective fragmentation of the
peptides of interest [27]. In both spectra, a peak is observed at m/z 1288.6, which corre-
sponds to the calculated mass of the N-terminal sequence (PMFIVNTNVPR) (Fig. 1,
Table 2). There is also a smaller peak at m/z 1304.6 attributed to the oxidation of Met-
2 residue in the peptide. In the samples treated with 7, there are additional peaks at m/z
1391.0, 1406.9, 1417.8, 1433.6, 1463.0, and 1479.3. The peaks appear in pairs spaced by
16 Da with the first peak being more intense than the second, suggesting that they are
derived from a partially oxidized methionine-containing peptide. The peptide displaying
a peak at m/z 1463.0 is 174.4 Da higher than the mass of the N-terminal sequence
(PMFIVNTNVPR).3 The increase in mass is due to the covalent attachment of species
with a mass of 174 Da.
2
At a higher ratio of inhibitor to protein (ꢁ70:1), 2 equiv. of inhibitor were covalently bound, as determined by
ESI-MS analysis. ESI-MS analysis of the inactivated MIF showed subunit molecular masses of 12544.0 and
12718.0 Da. The differences between these masses and that of the unmodified MIF subunit are 174.0 and
348.0 Da, indicating modification of MIF by two molecules of 7. The second site of modification was not
identified.
3
Confirmatory results were obtained by subjecting MIF and MIF inactivated by 7 to digestion by
endoproteinase Glu-C (protease V8), which cleaves at the carboxylate side of glutamate and aspartate residues
(data not shown). In these experiments, the N-terminal peptide sequence is PMFIVNTNVPRASVPE with an
average MW of 1772.1 Da.

192
P.A. Golubkov et al. / Bioorganic Chemistry 34 (2006) 183–199
Fig. 1. MALDI MS spectra showing the region from m/z 1250–1500 for the tryptic digest of MIF (upper) and
MIF treated with 7 (lower). In both panels, the peak at m/z 1288 corresponds to the N-terminal peptide
‘‘PMFIVNTNVPR’’, and the peak at 1304 corresponds to the same sequence with an oxidized Met-2. The peaks
at 1442 and 1458/1459 correspond to an unidentified protein that co-purified with MIF. The sample of MIF
treated with 7 has additional peaks at 1391/1406 (+102 Da), 1417/1433 (a mix of +128/130 Da), and 1463/1479
(+174 Da). Assignments are presented in the text.
Table 2
Mass spectral analysis of the tryptic digest of MIF and MIF modified by 7a
Obsd mass (Da) (control)
obsd mass (Da) (treated)
Peptide fragment
PMFIVNTNVPR
Mass difference (Da)
1288.9
1304.9
1288.6
1304.6
1391.0
1406.9
1417.8
1433.6
1442.8
1459.3
1463.0
1479.3
16
102
11
129
145
1443.0
1458.9
Contaminant
174
190
a
Ions between 1200 and 1500 Da that are >5% of the base peak. Error is 0.5 Da.
The ions corresponding to the peaks at m/z 1288, 1391, and 1463 were subjected to
MALDI-PSD fragmentation analysis to determine the amino acid sequence and location
of the modified residue [27]. The MALDI-PSD spectra of the peak at m/z 1288 is consis-
tent with its assignment to the N-terminal sequence of MIF, (PMFIVNTNVPR), by the
matched b and y fragment ions (Fig. 2). The peptide at m/z 1391 was also subject to
MALDI-PSD and matched to the same sequence, with the a and b ions shifted by
102 Da. The modified a2 and b3 ions, which retain charge on the N-terminal peptide frag-
ments, indicate that modification occurs at or near the N-terminal of the peptide. A prom-
inent ion at 172 can be assigned as the Pro-1 immonium ion with a modification of 102 Da.
There is no fragment at 172 in the spectra of the unmodified peptides. The MALDI-PSD
of the peptide at 1463 is dominated by neutral loss to 1390. The modification is observed
only weakly as a 174 Da mass addition, but is more strongly observed as 102 and 128/130
mass addition peptides. These peptides likely form from the loss of COCOO from the
inhibitor to form an adduct with a mass of 102 Da, and from the loss of HCOOH to form
an adduct with a mass of 130 Da. The decomposition/rearrangement likely results from a

P.A. Golubkov et al. / Bioorganic Chemistry 34 (2006) 183–199
193
Fig. 2. MALDI-PSD spectra of the parent ions at m/z 1289 (upper) and 1391 (lower) from the tryptic digest of
MIF treated with 7. The spectra are matched to the N-terminal sequence, ‘‘PMFIVNTNVPR’’ (where ‘‘in’’
designates the inhibitor). The upper panel shows the spectra for unmodified peptide whereas the lower panel
shows the peptide modified by a mass addition of 102 Da. The modification is located on Pro-1, as determined by
the presence of modified P immonium ion, and the a2 ions. In both spectra, the parent ion selection window
*
captured Met-2 with and without the sulfoxide. The sulfoxide-modified peptide has an intense ion for neutral loss
of CH3SOH. In the lower panel, neutral loss of modified Pro-1 is also observed.
MALDI-MS related process. Similar decomposition of peptide-bound inhibitors has been
observed previously [26].
The crystal structure of MIF soaked with 7. The overall structure of MIF determined in
this study did not reveal any significant differences from previously reported MIF struc-
tures [13,14,31]. Superpositioning of this structure with that determined for the structure
of mouse MIF in complex with the competitive inhibitor, E-2-fluoro-p-hydroxycinnamate
˚
˚
(13) (PDB code 1MFI) [15] shows only 0.424 A rmsd for 342 Ca atoms and 0.47 A rsmd
for 1368 backbone atoms. Superpositioning of this structure with that determined for the
˚
human MIF structure (PDB code 1GD0) shows 0.326 A rmsd for 342 Ca atoms and
˚
0.35 A rmsd for 1368 backbone atoms [40].
The active site of MIF soaked with 7. Attempts to grow crystals of MIF that had been
previously inactivated by 7 were not successful. Thus, crystals of native MIF were soaked
in solutions of 7, removed, and frozen before cracking. The trimer of MIF clearly shows
three potential active site cavities, one near each Pro-1. Although excellent data were
˚
obtained (1.45 A resolution, Fig. 3) and the MIF electron density was superb, the inhibitor
density seen in two of the three active sites was ambiguous (Fig. 4). It is assumed that the
inhibitor has decomposed under the crystallization conditions.
The third active site showed very distinct 2Fo ꢀ Fc density protruding into the active
site cavity. The density corresponds to the side chain of Glu-16 from another trimer from

194
P.A. Golubkov et al. / Bioorganic Chemistry 34 (2006) 183–199
˚
Fig. 3. Representative electron density showing the quality of data at 1.45 A resolution. (A) The molecular
replacement solution and the first electron density map were calculated using Y95F mutant coordinates (PDB
˚
code 1MFF, 2.0 A resolution) for phases, whereas the protein contained the Tyr-95 residue, whose hydroxyl
moiety was readily apparent. (B) Trp-108 after several cycles of unrestrained anisotropic refinement. The density
shown for both views is 2Fo ꢀ Fc map contoured at 1.0r. The figure was generated in graphics program O [37].
Fig. 4. Pro-1 adduct resulting from MIF crystals soaked with 7 modeled into the active site. (A) The expected
Michael complex (i.e., 15, Scheme 4) between Pro-1 and 7 does not fit well into the observed electron density. (B)
A Pro-1 adduct resulting from the MIF crystals soaked with 7 via a yet unidentified mechanism provides a much
better fit. The green density is 2Fo ꢀ Fc map contoured at 1.0r. The blue is Fo ꢀ Fc positive density contoured at
2.5r, and the red is negative density contoured at ꢀ2.8r. The figure was generated by the program O [37].
a crystallographic symmetry mate in the P63 space group (Fig. 5A and B). The Glu-16 side
chain appears to help stabilize four water molecules in the active site, but makes no direct
contact with Pro-1. The distances between these four density centers (2.68, 2.77, 2.79, and

P.A. Golubkov et al. / Bioorganic Chemistry 34 (2006) 183–199
195
Fig. 5. Active site residues participating in crystal contact between trimers. (A) Glu-160 from a crystallographic
symmetry mate protruding into the active site cavity. Two of the four bound solvent molecules are also visible. (B)
A different view of the same site showing all four water molecules bound in the occluded active site. (C) A close-
up of the occluded, inhibitor-free active site, showing the clearly defined proline pucker. The density for all three
panels is 2Fo ꢀ Fc map contoured at 1.0r. The figure was produced using the graphics program Pymol [41].
˚
3.09 A) correspond to hydrogen bond distances and not to carbon-carbon bonds. The two
˚
density centers closest to the catalytic proline are 3.13 and 2.80 A from it, also suggesting
hydrogen bonds involving waters and not a covalent adduct. Although this trimer–trimer
contact may have contributed to the stability and high-resolution quality of these MIF
crystals, the protruding Glu-160 side chain leaves little room for the inhibitor in that active
site. Hence, attempts were made to crystallize the E16A mutant of MIF with and without
7. Only crystals with no inhibitor could be grown, but at 30% PEG 8000. Unfortunately,
these crystals cracked immediately upon the addition of 7, suggesting that conformational
changes involving crystal contacts take place upon the binding of 7. The E16A-MIF crys-
tallized in the same space group, P63, but these crystals did not diffract to sufficiently high-
resolution to pursue a structure, suggesting that the trimer–trimer contact involving the
side chain of Glu-16 is important for crystal quality and stability.
Nonetheless, the high-resolution of this active site clearly shows a pronounced pucker
of the Pro-1 ring atoms (Fig. 5C), which have been implicated with the pKa variation of
Pro-1 [42]. This proline displays up (U) puckering, with Cc atom in the exo position.
The values for the torsion angles, as calculated by GEOMCALC [32] are v0 = 5.73ꢁ,
v1 = ꢀ27.39ꢁ, v2 = 38.69ꢁ, v3 = ꢀ35.56ꢁ, v4 = 18.42ꢁ, u = ꢀ53.62ꢁ, as defined previously
[43]. These values are in excellent agreement with previously determined dihedral angles
of puckered prolyl residues [44]. The Cb atom shows virtually no puckering, as indicated
by the small v0 angle, and is essentially in the plane defined by Cd–N–Ca atoms.
4. Discussion
Analogs of 2-oxo acids (such as 2-ketoglutarate and pyruvate) are potential mechanistic
probes of a wide variety of enzymes [45,46]. It is therefore not surprising that several such
compounds have been synthesized and tested including the acetylene derivatives, 2-OP (5,
Scheme 2), 2-OB (6), and 2-OPB (7) [18,19,45,46]. Chiu and Jordan first reported a syn-
thesis for 2-OPB in which a lithium derivative of 8 was coupled with an oxalic ester
[45]. However, we chose to adapt our procedure for the syntheses of 2-OP [18] and
2-OB [19] and generated 12, in which the t-butyl group replaces the ethyl ester. The t-butyl
ester can be removed under anhydrous conditions.

196
P.A. Golubkov et al. / Bioorganic Chemistry 34 (2006) 183–199
The subsequent incubation of 7 with MIF showed that the acetylene compound is a
potent active-site-directed irreversible inhibitor of the PPT activity of MIF. Interestingly,
a substoichiometric concentration of inhibitor (ꢁ0.5 equiv.) results in a complete loss of
activity. Under these conditions, a single site on the protein is covalently modified. The
site of modification is Pro-1, previously implicated as the catalytic base in the PPT activity
of MIF [13,17].
One reasonable mechanism for the inactivation involves the Michael addition of the
amino-terminal proline to the C-4 position of 7 (Scheme 4). Attack at this position produc-
es an enol (or enolate, 14), which tautomerizes to afford 15 [18]. The observation that the
covalent adduct on Pro-1 has a molecular mass of 174 Da is consistent with the formation
of 15. Moreover, it is well known that a,b-unsaturated acetylenic ketones are susceptible
to nucleophilic attack and the low pKa for Pro-1 (ꢁ5.6) suggests that it can function as a
nucleophile and react with 7 [18]. Modification of one of the three active sites in MIF
apparently disrupts the remaining two active sites such that these sites no longer have
PPT activity. The disruption of these active sites may result from conformational changes
that occur as the bound linear acetylene molecule is transformed to a more flexible cova-
lent adduct. It was anticipated that a crystallographic analysis might provide a structural
rationale for these observations. However, the electron density observed in the two active
sites that were not blocked by a neighboring Glu-16 does not appear to fit well when mod-
eled with the expected Michael complex (Fig. 4A). Although it is believed that this might
be due to the decomposition of inhibitor over time under the conditions used for crystal-
lization, an unambiguous explanation for this density is lacking.
MIF is also categorized as a tautomerase superfamily member [16,47]. The tautomerase
superfamily consists of five known families represented by MIF, 5-carboxymethyl-2-
hydroxymuconate isomerase (CHMI), 4-OT [16,47,48], cis-3-chloroacrylic acid dehalogen-
ase (cis-CaaD) [49], and malonate semialdehyde decarboxylase (MSAD) [50]. CHMI, from
E. coli C, and 4-OT, from P. putida mt-2, function as tautomerases in degradation path-
ways for aromatic amino acids and aromatic hydrocarbons, respectively [51]. cis-CaaD
precedes MSAD in the degradation pathway for the nematocide, 1,3-dichloropropene [52].
A battery of mechanistic and structural studies has implicated Pro-1 as a critical cata-
lytic residue in all characterized tautomerase superfamily members. Pro-1 can function as
either a catalytic base {the PPT activity of MIF [13,15], CHMI [51], and 4-OT [53,54]} or
an acid {CaaD [55], cis-CaaD [49,56], and MSAD [40,57]}, depending on the pKa value,
which is governed partly by the active site environment. The pKa has been directly mea-
sured for 4-OT (ꢁ6.4) [54], MIF (ꢁ5.6) [58], CaaD [55] and MSAD (both ꢁ9.2) [57] by
15N NMR titration of the uniformly labeled enzyme.
In the absence of a direct NMR measurement of the pKa value, 2-OP (5) has served as
an effective indicator of the pKa value and the active site environments of tautomerase
Enz
Enz
O-
O
N
MIF
N
O
.
-
-
CO2
-
CO2
CO2
7
H+
15
14
Scheme 4.

P.A. Golubkov et al. / Bioorganic Chemistry 34 (2006) 183–199
197
O
O
O-
O
HO
.
H3C
-
H3C
CO2
-
-
CO2
CO2
H3C
H+
16
5
Scheme 5.
superfamily members. The reaction of 2-OP with superfamily members reflects the pre-
dominant ionization state of the catalytic proline (neutral vs. cationic) and the nature
of the active site (hydrophobic vs hydrophilic). For example, 4-OT is inactivated by
2-OP (and 2-OB) [18,19], whereas CaaD, cis-CaaD, and MSAD process 2-OP to aceto-
pyruvate (16, Scheme 5) although with different efficiencies [26,49,57]. In both enzymes,
the addition of water to 2-OP results in the formation of an enol species. Subsequent keto-
nization the enol produces 16.
Likewise, 2-OPB could function as either an inhibitor or a substrate and may emerge as
a useful mechanistic probe of tautomerase superfamily members. As an inhibitor, 2-OPB
would inactivate MIF homologues with a PPT activity in which Pro-1 has a low pKa value
due to a hydrophobic active site. As a substrate, MIF homologues could add water to the
C-4 position and generate benzoylpyruvate. Efforts are currently underway to identify tau-
tomerase superfamily members capable of accommodating the larger phenyl ring and car-
rying out a conjugate addition of water.
Acknowledgments
The mass spectrometry described in this paper was carried out in the Analytical Instru-
mentation Facility Core housed in the College of Pharmacy at the University of Texas at
Austin and supported by Center Grant ES07784.
References
[1] M. George, J.H. Vaughan, Proc. Soc. Exp. Biol. Med. 111 (1962) 514–521.
[2] B.R. Bloom, B. Bennett, Science 153 (1966) 80–82.
[3] J.R. David, Proc. Natl. Acad. Sci. USA 56 (1966) 72–77.
[4] M. Leech, C. Metz, L. Santos, T. Peng, S.R. Holdsworth, R. Bucala, E.F. Morand, Arthritis Rheum. 41
(1998) 910–917.
[5] M. Leech, C. Metz, P. Hall, P. Hutchinson, K. Gianis, M. Smith, H. Weedon, S.R. Holdsworth, R. Bucala,
E.F. Morand, Arthritis Rheum. 42 (1999) 1601–1608.
[6] T. Mikayama, T. Nakano, H. Gomi, Y. Nakagawa, Y.C. Liu, M. Sato, A. Iwamatsu, Y. Ishii, W.Y. Weiser,
K. Ishizaka, Proc. Natl. Acad. Sci. USA 90 (1993) 10056–10060.
[7] S. Onodera, K. Kaneda, Y. Mizue, Y. Koyama, M. Fujinaga, J. Nishihira, J. Biol. Chem. 275 (2000) 444–
450.
[8] H.Y. Lan, W. Mu, N. Yang, A. Meinhardt, D.J. Nikolic-Paterson, Y.Y. Ng, M. Bacher, R.C. Atkins, R.
Bucala, Am. J. Pathol. 149 (1996) 1119–1127.
[9] J. Bernhagen, T. Calandra, R.A. Mitchell, S.B. Martin, K.J. Tracey, W. Voelter, K.R. Manogue, A. Cerami,
R. Bucala, Nature 365 (1993) 756–759.
[10] M. Bozza, A.R. Satoskar, G. Lin, B. Lu, A.A. Humbles, C. Gerard, J.R. David, J. Exp. Med. 189 (1999)
341–346.
[11] T. Calandra, J. Bernhagen, R.A. Mitchell, R. Bucala, J. Exp. Med. 179 (1994) 1895–1902.
[12] E. Rosengren, P. Aman, S. Thelin, C. Hansson, S. Ahlfors, P. Bjork, L. Jacobsson, H. Rorsman, FEBS Lett.
417 (1997) 85–88.

198
P.A. Golubkov et al. / Bioorganic Chemistry 34 (2006) 183–199
[13] J.B. Lubetsky, M. Swope, C. Dealwis, P. Blake, E. Lolis, Biochemistry 38 (1999) 7346–7354.
[14] M. Suzuki, H. Sugimoto, A. Nakagawa, I. Tanaka, J. Nishihira, M. Sakai, Nat. Struct. Biol. 3 (1996) 259–
266.
[15] A.B. Taylor, W.H. Johnson Jr., R.M. Czerwinski, H.S. Li, M.L. Hackert, C.P. Whitman, Biochemistry 38
(1999) 7444–7452.
[16] C.P. Whitman, Arch. Biochem. Biophys. 402 (2002) 1–13.
[17] S.L. Stamps, M.C. Fitzgerald, C.P. Whitman, Biochemistry 37 (1998) 10195–10202.
[18] W.H. Johnson Jr., R.M. Czerwinski, M.C. Fitzgerald, C.P. Whitman, Biochemistry 36 (1997) 15724–15732.
[19] A.B. Taylor, R.M. Czerwinski, W.H. Johnson Jr., C.P. Whitman, M.L. Hackert, Biochemistry 37 (1998)
14692–14700.
[20] M.C. Pirrung, J. Chen, E.G. Rowley, A.T. McPhail, J. Am. Chem. Soc. 115 (1993) 7103–7110.
[21] S.C. Wang, W.H. Johnson Jr., R.M. Czerwinski, C.P. Whitman, Biochemistry 43 (2004) 748–758.
[22] J. Sambrook, E.F. Fritsch, T. Maniatis, Molecular Cloning: A Laboratory Manual, Cold Spring Harbor
Laboratory, Cold Spring Harbor, New York, 1989.
[23] U.K. Laemmli, Nature 227 (1970) 680–685.
[24] W.J. Waddell, J. Lab. Clin. Med. 48 (1956) 311–314.
[25] S.N. Ho, H.D. Hunt, R.M. Horton, J.K. Pullen, L.R. Pease, Gene 77 (1989) 51–59.
[26] S.C. Wang, M.D. Person, W.H. Johnson Jr., C.P. Whitman, Biochemistry 42 (2003) 8762–8773.
[27] M.D. Person, T.J. Monks, S.S. Lau, Chem. Res. Toxicol. 16 (2003) 598–608.
[28] Z. Otwinowski, W. Minor, in: C.W. CarterJr., R.M. Sweet (Eds.), Methods in Enzymology, vol. 276,
Academic Press, New York, 1997, pp. 307–326.
[29] B.W. Matthews, J. Mol. Biol. 33 (1968) 491–497.
[30] K.A. Kantardjieff, B. Rupp, Protein Sci. 12 (2003) 1865–1871.
[31] S.L. Stamps, A.B. Taylor, S.C. Wang, M.L. Hackert, C.P. Whitman, Biochemistry 39 (2000) 9671–9678.
[32] Collaborative Computational Project, Number 4, Acta Crystallogr. D 50 (1994) 760–763.
[33] E. Potterton, P. Briggs, M. Turkenburg, E. Dodson, Acta Crystallogr. D 59 (2003) 1131–1137.
[34] J. Navaza, Acta Crystallogr. A 50 (1994) 157–163.
[35] G.N. Murshudov, A.A. Vagin, E.J. Dodson, Acta Crystallogr. D 53 (1997) 240–255.
[36] M.D. Winn, M.N. Isupov, G.N. Murshudov, Acta Crystallogr. D 57 (2001) 122–133.
[37] T.A. Jones, J.Y. Zou, S.W. Cowan, M. Kjeldgaard, Acta Crystallogr. A 47 (1991) 110–119.
[38] R.J. Read, A.J. Schierbeek, J. Appl. Crystallogr. 21 (1988) 490–495.
[39] A. Perrakis, R. Morris, V.S. Lamzin, Nat. Struct. Biol. 6 (1999) 458–463.
[40] N. Guex, M.C. Peitsch, Electrophoresis 18 (1997) 2714–2723.
[41] W.L. DeLano, The Pymol Molecular Graphics System, DeLano Scientific, San Carlos, CA, USA, 2002.
[42] N. Panasik Jr., E.S. Eberhardt, A.S. Edison, D.R. Powell, R.T. Raines, Int. J. Pept. Protein Res. 44 (1994)
262–269.
[43] F.A. Momany, R.F. McGuire, A.W. Burgess, H.A. Scheraga, J. Phys. Chem. 79 (1975) 2361–2381.
[44] G. Nemethy, K.D. Gibson, K.A. Palmer, C.N. Yoon, G. Paterlini, A. Zagari, S. Rumsey, H.A. Scheraga, J.
Phys. Chem. 96 (1992) 6472–6484.
[45] C.C. Chiu, F. Jordan, J. Org. Chem. 59 (1994) 5763–5766.
[46] A. Brown, N. Nemeria, J. Yi, D. Zhang, W.B. Jordan, R.S. Machado, J.R. Guest, F. Jordan, Biochemistry
36 (1997) 8071–8081.
[47] A.G. Murzin, Curr. Opin. Struct. Biol. 6 (1996) 386–394.
[48] 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.
[49] G.J. Poelarends, H. Serrano, M.D. Person, W.H. Johnson Jr., A.G. Murzin, C.P. Whitman, Biochemistry 43
(2004) 759–772.
[50] G.J. Poelarends, W.H. Johnson Jr., A.G. Murzin, C.P. Whitman, J. Biol. Chem. 278 (2003) 48674–48683.
[51] 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.
[52] G.J. Poelarends, M. Wilkens, M.J. Larkin, J.D. van Elsas, D.B. Janssen, Appl. Environ. Microbiol. 64 (1998)
2931–2936.
[53] J.T. Stivers, C. Abeygunawardana, A.S. Mildvan, G. Hajipour, C.P. Whitman, L.H. Chen, Biochemistry 35
(1996) 803–813.
[54] J.T. Stivers, C. Abeygunawardana, A.S. Mildvan, G. Hajipour, C.P. Whitman, Biochemistry 35 (1996) 814–
823.

P.A. Golubkov et al. / Bioorganic Chemistry 34 (2006) 183–199
199
[55] H.F. Azurmendi, S.C. Wang, M.A. Massiah, G.J. Poelarends, C.P. Whitman, A.S. Mildvan, Biochemistry 43
(2004) 4082–4091.
[56] G.J. Poelarends, H. Serrano, W.H. Johnson Jr., C.P. Whitman, Biochemistry 43 (2004) 7187–7196.
[57] G.J. Poelarends, H. Serrano, W.H. Johnson Jr., D.W. Hoffman, C.P. Whitman, J. Am. Chem. Soc. 126
(2004) 15658–15659.
[58] M. Swope, H.W. Sun, P.R. Blake, E. Lolis, EMBO J. 17 (1998) 3534–3541.