Divergence of biochemical function in the HAD superfamily: D-glycero-D-manno-heptose-1,7-bisphosphate phosphatase (GmhB)

Article abstract of DOI:10.1021/bi902018y

D-glycero-D-manno-Heptose-1,7-bisphosphate phosphatase (GmhB) is a member of the histidinol-phosphate phosphatase (HisB) subfamily of the haloalkanoic acid dehalogenase (HAD) enzyme superfamily. GmhB supports two divergent biochemical pathways in bacteria: the D-glycero-D-manno-heptose-1α-GDP pathway (in S-layer glycoprotein biosynthesis) and the L-glycero-D-manno- heptose-1β-ADP pathway (in lipid A biosynthesis). Herein, we report the comparative analysis of substrate recognition in selected GmhB orthologs. The substrate specificity of the L-glycero-D-manno-heptose-1β-ADP pathway GmhB from Escherichia coli K-12 was evaluated using hexose and heptose bisphosphates, histidinol phosphate, and common organophosphate metabolites. Only D-glycero-D-manno-heptose 1β,7-bisphosphate (k_cat/K _m=7 × 10⁶ M^-1 s^-1) and D-glycero-D-manno-heptose 1α,7-bisphosphate (k_cat/K_m = 7 × 10⁴ M^-1 s^-1) displayed physiologically significant substrate activity. ³¹P NMR analysis demonstrated that E. coli GmhB selectively removes the C(7) phosphate. Steady-state kinetic inhibition studies showed that D-glycero-D-manno-heptose 1β-phosphate (K_is = 60 μM, and K_ii = 150 μM) and histidinol phosphate (K_is = 1 mM, and K_ii = 6 mM), while not hydrolyzed, do in fact bind to E. coli GmhB, which leads to the conclusion that nonproductive binding contributes to substrate discrimination. High catalytic efficiency and a narrow substrate range are characteristic of a well-evolved metabolic enzyme, and as such, E. coli GmhB is set apart from most HAD phosphatases (which are typically inefficient and promiscuous). The specialization of the biochemical function of GmhB was examined by measuring the kinetic constants for hydrolysis of the α- and β-anomers of D-glycero-D-manno-heptose 1β,7-bisphosphate catalyzed by the GmhB orthologs of the L-glycero-D-manno-heptose 1β-ADP pathways operative in Bordetella bronchiseptica and Mesorhizobium loti and by the GmhB of the D-glycero-D-manno-heptose 1α-GDP pathway operative in Bacteroides thetaiotaomicron. The results show that although each of these representatives possesses physiologically significant catalytic activity toward both anomers, each displays substantial anomeric specificity. Like E. coli GmhB, B. bronchiseptica GmhB and M. loti GmhB prefer the β-anomer, whereas B. thetaiotaomicron GmhB is selective for the α-anomer. By determining the anomeric configuration of the physiological substrate (D-glycero-D-manno-heptose 1,7-bisphosphate) for each of the four GmhB orthologs, we discovered that the anomeric specificity of GmhB correlates with that of the pathway kinase. The conclusion drawn from this finding is that the evolution of the ancestor to GmhB in the HisB subfamily provided for specialization toward two distinct biochemical functions.

Full text of DOI:10.1021/bi902018y

1072 Biochemistry 2010, 49, 1072–1081
DOI: 10.1021/bi902018y
Divergence of Biochemical Function in the HAD Superfamily:
-glycero-
-manno-Heptose-1,7-bisphosphate Phosphatase (GmhB)†
D
D
Liangbing Wang,‡,^ Hua Huang,‡ Henry H. Nguyen,§ Karen N. Allen,*,§, Patrick S. Mariano,*,‡ and
Debra Dunaway-Mariano*,‡
‡Department of Chemistry and Chemical Biology, University of New Mexico, Albuquerque, New Mexico 87131, §Department of
Physiology and Biophysics, Boston University School of Medicine, Boston, Massachusetts 02118-2394, and Department of Chemistry,
Boston University, Boston, Massachusetts 02215. ^Current address: Chengdu Di’Ao Pharmaceutical Co.,
Institute of Medicine, Sichuan, China.
Received November 24, 2009; Revised Manuscript Received December 31, 2009
ABSTRACT:
D
-glycero-
phosphate phosphatase (HisB) subfamily of the haloalkanoic acid dehalogenase (HAD) enzyme superfamily.
GmhB supports two divergent biochemical pathways in bacteria: the -glycero- -manno-heptose-1R-GDP
pathway (in S-layer glycoprotein biosynthesis) and the -glycero- -manno-heptose-1β-ADP pathway (in lipid
A biosynthesis). Herein, we report the comparative analysis of substrate recognition in selected GmhB
orthologs. The substrate specificity of the -glycero- -manno-heptose-1β-ADP pathway GmhB from
Escherichia coli K-12 was evaluated using hexose and heptose bisphosphates, histidinol phosphate, and
D
-manno-Heptose-1,7-bisphosphate phosphatase (GmhB) is a member of the histidinol-
D
D
L
D
L
D
common organophosphate metabolites. Only
D
-glycero-
D
-manno-heptose 1β,7-bisphosphate (kcat/Km = 7 ꢀ
106 M-1 s-1) and
D-glycero-
D
-manno-heptose 1R,7-bisphosphate (kcat/Km = 7 ꢀ 104 M-1 s-1) displayed
physiologically significant substrate activity. 31P NMR analysis demonstrated that E. coli GmhB selectively
removes the C(7) phosphate. Steady-state kinetic inhibition studies showed that
D
-glycero- -manno-heptose
D
1β-phosphate (Kis=60 μM, and Kii=150 μM) and histidinol phosphate (Kis=1 mM, and Kii=6 mM), while
not hydrolyzed, do in fact bind to E. coli GmhB, which leads to the conclusion that nonproductive binding
contributes to substrate discrimination. High catalytic efficiency and a narrow substrate range are char-
acteristic of a well-evolved metabolic enzyme, and as such, E. coli GmhB is set apart from most HAD
phosphatases (which are typically inefficient and promiscuous). The specialization of the biochemical
function of GmhB was examined by measuring the kinetic constants for hydrolysis of the R- and β-anomers
of
heptose 1β-ADP pathways operative in Bordetella bronchiseptica and Mesorhizobium loti and by the GmhB of
the -glycero- -manno-heptose 1R-GDP pathway operative in Bacteroides thetaiotaomicron. The results show
D
-glycero-
D-manno-heptose 1β,7-bisphosphate catalyzed by the GmhB orthologs of the
L
-glycero- -manno-
D
D
D
that although each of these representatives possesses physiologically significant catalytic activity toward both
anomers, each displays substantial anomeric specificity. Like E. coli GmhB, B. bronchiseptica GmhB and M.
loti GmhB prefer the β-anomer, whereas B. thetaiotaomicron GmhB is selective for the R-anomer.
By determining the anomeric configuration of the physiological substrate (
D
-glycero- -manno-heptose 1,7-
D
bisphosphate) for each of the four GmhB orthologs, we discovered that the anomeric specificity of GmhB
correlates with that of the pathway kinase. The conclusion drawn from this finding is that the evolution of the
ancestor to GmhB in the HisB subfamily provided for specialization toward two distinct biochemical
functions.
D
-glycero-
D
-manno-Heptose-1,7-bisphosphate phosphatase
enzyme commonly known as “HisB”2 (3). HisB is a bifunctional
enzyme whose N-terminal domain catalyzes the dephosphoryla-
tion of histidinol phosphate and whose C-terminal domain
catalyzes the dehydration of imidazole glycerol phosphate. These
two reactions constitute the sixth and eighth steps of histidine
biosynthesis, respectively. HisB has a limited biological range
(viz. the γ-subdivision of proteobacteria wherein it coexists with
GmhB), whereas GmhB is found in a wide variety of archea and
bacteria (1). Fani et al. (1) proposed that the HisB subfamily
ancestor functioned as a promiscuous phosphatase that ultimately
(GmhB)1 is a member of a subfamily of the haloalkanoic acid
dehalogenase (HAD) enzyme superfamily (2) named after the
†This work was supported by National Institutes of Health Grant
GM61099 to K.N.A. and D.D.-M.
*To whom correspondence should be addressed. D.D.-M.: e-mail,
dd39@unm.edu; phone, (505) 277-3383; fax, (505) 277-2609. P.S.M.:
e-mail, mariano@unm.edu; phone, (505) 277-6390; fax, (505) 277-2609.
K.N.A.: e-mail, drkallen@bu.edu; phone, (617) 358-5544; fax, (617)
358-5554.
1Abbreviations: GmhB,
D-glycero-D-manno-heptose-1,7-bisphosphate
specialized to form D-glycero-D-manno-heptose-1,7-bisphosphate
phosphatase; HAD, haloalkanoic acid dehalogenase; HisB, bifunctional
histidinol-phosphate phosphatase/imidazole-glycerol-phosphate dehy-
dratase; IPTG, isopropyl R-D-thiogalactopyranoside; ES-MS, electro-
spray mass spectrometry; NMR, nuclear magnetic resonance; KDOP-
8-P, 2-keto-3-deoxyoctulosonate 8-phosphate.
2Recently, the name HisNB was recommended for use in place of
HisB (1); however, the HisB name is used herein.
r
pubs.acs.org/Biochemistry
Published on Web 01/05/2010
2010 American Chemical Society

Article
Biochemistry, Vol. 49, No. 6, 2010 1073
F
IGURE 1: Chemical steps and enzyme catalysts of the A. thermoaerophilus D-glycero-D-manno-heptose 1R-GDP pathway (4) and the E. coli
L-glycero-D-manno-heptose 1β-ADP pathway (5) .
phosphatase. Duplication of the ancestral gene in γ-proteobac-
teria allowed one copy to evolve to histidinol-phosphate phos-
phatase and eventually fuse with imidazole-glycerol-phosphate
dehydratase to form HisB. In the study described herein, we
examined the divergence of biochemical function in GmhB that
has accompanied its colonization of bacteria.
nucleotidyl transferase (HldE_k/HldE_t) selects the β-anomer
for phosphorylation with ATP to produce D-glycero-D-manno-
heptose 1β,7-bisphosphate (5, 13). The C(7) phosphate group
is subsequently removed by GmhB to form the corresponding
D-glycero-D-manno-heptose 1-phosphate, which is the substrate
for the nucleotidyl transferase. In A. thermoaerophilus, the nucleo-
The GmhB substrate,
phate, is formed as an intermediate in the
heptose 1R-GDP (4) and -glycero- -manno-heptose 1β-ADP
D
-glycero-
D
-manno-heptose 1,7-bisphos-
tidyl transferase catalyzes the transfer of GMP from GTP to the
C(1) phosphate of the D-glycero-D-manno-heptose 1R-phosphate,
whereas in E. coli, the C-terminal domain (HldE_t) of the
D
-glycero- -manno-
D
L
D
pathways (5) illustrated in Figure 1. The pathways provide
activated glycero-manno-heptose units for incorporation into
membrane surface glycoproteins and glycolipids (4-12). The
bifunctional kinase/nucleotidyl transferase catalyzes the transfer
of AMP from ATP to the C(1) phosphate of the
-manno-heptose 1β-phosphate. Finally, in A. thermoaerophilus,
a glycosyl transferase delivers the -glycero- -manno-heptose
unit of the -glycero- -manno-heptose 1R-GDP to the carbohy-
drate core of the S-layer glycoprotein (7), whereas in E. coli, the
-glycero- -manno-heptose 1β-ADP is first converted to the
-glycero epimer by the epimerase GmhD, before glycosyl transfer
D-glycero-
D
two pathways begin with the isomerization of
7-phosphate to -glycero- -manno-heptose 7-phosphate catalyzed
by the heptose isomerase GmhA. -glycero- -manno-Heptose
D-sedoheptulose
D
D
D
D
D
D
D
D
7-phosphate exists as a mixture of C(1) R- and β-anomers. In
Aneurinibacillus thermoaerophilus, the kinase HddA selects the
R-anomer for phosphorylation with ATP, which generates
D
D
L
to the lipid A core (14).
Herein, we describe a comparative analysis of substrate recog-
nition in four GmhB orthologs that function in one pathway or
D
-glycero-
D
-manno-heptose 1R,7-bisphosphate (4). In Escherichia
coli, the N-terminal “kinase” domain of a bifunctional kinase/

1074 Biochemistry, Vol. 49, No. 6, 2010
Wang et al.
the other. Our results show that E. coli GmhB displays high
buffer A. The desired protein fractions were combined, adjusted
to 15% (NH4)2SO4 (w/v), and loaded onto an 18 cm ꢀ 3 cm
Butyl-Sepharose column pre-equilibrated with buffer A contain-
ing 15% (NH4)2SO4. The column was eluted with a 0.5 L linear
gradient of (NH4)2SO4 (from 15 to 0%) in buffer A. The desired
protein fractions were combined, dialyzed against buffer A,
concentrated, and stored at -80 ꢀC: yield, 30 mg of protein/g
of wet cells; mass, calcd 33846 Da, found 33846 Da.
(ii) B. bronchiseptica D-glycero-D-manno-Heptose-1β,7-
bisphosphate Phosphatase (SwissProt accession number
Q7WG29). The cell supernatant was chromatographed on a
40 cm ꢀ 5 cm DEAE-Sepharose column using a 2 L linear
gradient of KCl (from 0 to 0.5 M) in buffer A as the eluant. The
desired protein fractions were combined, adjusted to 20%
(NH4)2SO4 (w/v), and loaded onto an 18 cm ꢀ 3 cm Butyl-
Sepharose column equilibrated with buffer A containing 20%
(NH4)2SO4. The column was eluted with a 0.5 L linear gradient of
(NH4)2SO4 (from 20 to 15%) in buffer A, followed by 0.5 L of
buffer A containing 15% (NH4)2SO4. The desired protein frac-
tions were combined, concentrated, and loaded onto a Sephadex
G-75 column. The column was eluted with buffer A, and the
desired protein fractions were combined, dialyzed against buffer
A, concentrated, and stored at -80 ꢀC: yield, 6 mg of protein/g of
wet cells; mass, calcd 19035 Da, found 19034 Da.
(iii) B. bronchiseptica D-glycero-D-manno-Heptose-1β-
phosphate Adenylyltransferase (SwissProt accession num-
ber Q7WF17). The cell supernatant was chromatographed on a
40 cmꢀ5 cm DEAE-Sepharose 50-120 column using a 2 L linear
gradient of NaCl (from 0 to 0.5 M) in buffer A as the eluant. The
desired protein fractions were combined, adjusted to 15% (NH4)2-
SO4 (w/v), and then loaded onto an 18 cm ꢀ 3 cm Butyl-
Sepharose column equilibrated with buffer A containing 15%
(NH4)2SO4. The column was eluted at 4 ꢀC with a 0.5 L linear
gradient of (NH4)2SO4 (from 15 to 0%) in buffer A. The desired
protein fractions were combined, dialyzed against buffer A,
concentrated, and stored at -80 ꢀC: yield, 30 mg of protein/g
of wet cells; mass, calcd 17440 Da, found 17322 Da. The N-terminal
Met appears to have been removed by posttranslational modi-
fication.
(iv) E. coli K-12 Phosphoheptose Isomerase (NCBI
accession number NP_414757). The supernatant was chro-
matographed on a 40 cm ꢀ 5 cm DEAE-Sepharose 50-120
column using a 2 L linear gradient of KCl (from 0 to 0.5 M) in
buffer A as the eluant. The desired fractions were combined,
dialyzed against buffer A, and concentrated: yield, 10 mg of
protein/g of wet cells; mass, calcd 20815 Da, found 20815 Da.
(v) A. thermoaerophilus D-glycero-D-manno-Heptose-7-
P Kinase (SwissProt accession number Q9AGY8). The
E. coli DH5R clone was a kind gift from P. Messner (Universitaet
fuer Bodenkultur Wien Zentrom fuer NanoBiotechnologie,
Wien, Austria). The gene was subcloned into destination vector
pDEST17 by homologous recombination. The isolated plasmid
was used to transform E. coli BL21 star (DE3) cells. The
transformed cells were grown at 30 ꢀC and 200 rpm in 4 L of
LB medium (containing 50 μg/mL ampicillin) to an OD660 of
0.6-1.0 and then induced with 0.4 mM IPTG. The cells were
harvested by centrifugation at 6500 rpm for 15 min, suspended in
150 mL of ice-cold lysis buffer [50 mM NaH2PO4, 300 mM NaCl,
and 10 mM imidazole (pH 8.0)], and passed through a French
pressure cell press at 1500 psi. The lysate was clarified by
centrifugation and chromatographed on a Ni-NTA column
(equilibrated with lysis buffer). The column was washed with
catalytic efficiency toward its physiological substrate (
D-glycero-
D-manno-heptose 1,7-bisphosphate) and a narrow substrate
range, characteristic of a well-evolved metabolic enzyme.
Furthermore, we demonstrate that an anomeric preference exists
in each of the GmhB orthologs and show a correlation with the
anomeric configuration of the physiological substrate. The
results of this effort establish the foundation for a companion
paper (15) in which we report and analyze the liganded structure
of the GmhB from E. coli K-12 and the liganded structure
of GmhB from Bordetella bronchiseptica. Together, these two
studies define the structural elements of substrate recognition
that underlie the divergence of GmhB function within the HAD
HisB subfamily.
MATERIALS AND METHODS
Commercial Materials, Services, Spectral Determina-
tions, Buffers, and General Procedures. Except where in-
dicated, all chemicals were obtained from Sigma-Aldrich.
Primers, T4 DNA ligase, and restriction enzymes were from
Invitrogen. Pfu polymerase and the pET23 vector kit were from
Stratagene. DEAE-Sepharose was from Amersham Biosciences.
Host cells were purchased from Novagen. Genomic DNA
(10798, 29148D, BAA-588D) was purchased from ATCC. The
nucleotide sequence of each cloned gene was determined by the
Center for Genetics in Medicine, University of New Mexico.
Electrospray mass spectrometry (ES-MS) determinations were
conducted by the University of New Mexico Mass Spectral
Facility. 1H NMR, 13C NMR, and 31P NMR (proton-decoupled)
spectra were recorded on a Bruker Avance 500 NMR spectro-
meter at room temperature using D2O or H2O as the solvent.
1H NMR and 13C NMR data are reported as follows: chemical
shift (parts per million) and multiplicity (s, singlet; d, doublet;
t, triplet; q, quartet; m, multiplet). Protein concentration deter-
minations were conducted with the Bradford assay. SDS-PAGE
chromatography was used to monitor protein purification.
Plasmid DNA was purified using the Qiaprep Spin Miniprep
Kit from Qiagen, and transformations were conducted with
E. coli BL21(DE3) competent cells from Stratagene. The protein
sample concentration was determined using a 10K Amicon Ultra
Centrifugal filter (Millipore). All protein purifications were
conducted at 4 ꢀC using buffer A which consists of 50 mM
KþHEPES (pH 7.0), 5 mM MgCl2, and 1 mM DTT.
Gene Cloning and Cell Lysate Preparation. The gene was
amplified from genomic DNA by PCR using Pfu DNA poly-
merase and commercial oligonucleotide primers containing NdeI
and BamHI cut sites. The linearized pET-23b vector was ligated
to the digested PCR product for cell transformation. The purified
plasmid DNA was used to transform cells which were grown at
25 ꢀC with agitation at 200 rpm in 10 L of Luria broth (containing
50 μg/mL ampicillin) to an OD600 of 0.6-1.0 and then induced
for 4 h at 20 ꢀC with 0.4 mM IPTG. The cells were harvested by
centrifugation (6500 rpm for 15 min at 4 ꢀC) to yield 3 g of cell
paste/L of culture medium. The cell pellet was suspended in ice-
cold buffer A and passed through a French press at 1200 psi.
The lysate was centrifuged at 20000 rpm and 4 ꢀC for 30 min, and
the supernatant was immediately used in protein purification.
Recombinant Protein Purification. (i) B. bronchiseptica
RB50 D-glycero-D-manno-Heptose-7-P Kinase (SwissProt
accession number BB3463). The cell supernatant was loaded
onto a 40 cm ꢀ 5 cm DEAE-Sepharose 50-120 column, which
was eluted with a 2 L linear gradient of KCl (from 0 to 0.5 M) in

Article
Biochemistry, Vol. 49, No. 6, 2010 1075
buffer containing 50 mM NaH2PO4, 300 mM NaCl, and 20 mM
imidazole (pH 8.0) before the protein was eluted with buffer
containing 50 mM NaH2PO4, 300 mM NaCl, and 250 mM
imidazole (pH 8.0). The desired fractions were combined, con-
centrated, and dialyzed against 50 mM KþHEPES and 10 mM
MgCl2 (pH 7.5): yield, 13 mg of protein/g of wet cells.
column. After being washed with 600 mL of deionized water, the
column was eluted with a 2 L linear gradient of NH4HCO3 (from
0 to 1 M). The column fractions were monitored using an acid-
labile phosphate assay (16). The
D-glycero-D-manno-heptose
1β,7-bisphosphate-containing fractions (eluted at 0.4-0.5 M
NH4HCO3) were combined and concentrated to provide 110 mg
(vi) E. coli D-glycero-D-manno-Heptose-1,7-bisphos-
phate Phosphatase (SwissProt accession number P63228).
The cell supernatant was chromatographed on a 40 cm ꢀ 5 cm
DEAE-Sepharose column using a 2 L linear gradient of KCl
(from 0 to 0.5 M) in buffer A as the eluant. The desired protein
fractions were combined, adjusted to 20% (NH4)2SO4 (w/v), and
loaded onto an 18 cm ꢀ 3 cm Butyl-Sepharose column equili-
brated with buffer A containing 20% (NH4)2SO4. The column
was eluted with a 0.5 L linear gradient of (NH4)2SO4 (from 20 to
15%) in buffer A and then with 0.5 L of buffer A containing 15%
(NH4)2SO4. The desired protein fractions were combined and
concentrated to yield 15 mg of protein/g of wet cells: mass, calcd
21294 Da, found 21163 Da. The N-terminal Met appears to have
been removed by posttranslational modification.
of D-glycero-D-manno-heptose 1β,7-bisphosphate (28% yield):
1H NMR δ 4.86 (d, 1H, J = 9.0 Hz), 4.00 (s, 1H), 3.80 (d, 2H,
J=12 Hz), 3.66 (m, 1H), 3.60 (t, 1H, J=9.0 Hz), 3.47 (d, 1H, J=
9.5 Hz), 3.30 (d, 1H, J = 10.0 Hz); 13C NMR δ 94.9 (bs), 77.1,
72.5, 70.9 (d, J=4.4 Hz), 69.7 (d, J=5.9 Hz), 66.2, 63.5 (d, J=
3.9 Hz); 31P NMR δ 4.80 and 3.01; mass, calcd for C7H15O13P2
368.9988 Da, found 368.9965 Da.
(ii) D-glycero-D-manno-Heptose 1R,7-Bisphosphate.
5-phosphate (0.5 g, 1.6 mmol), 0.5 g of -fructose 6-phosphate
(1.6 mmol), 1.70 g of ATP (3.0 mmol), 10 μL (100 mM stock,
1 μmol) of -glyceraldehyde 3-phosphate, and 100 μL (100 mM
D-Ribose
D
D
stock, 10 μmol) of thiamine pyrophosphate were combined with
100 mL of 50 mM Tris containing 10 mM MgCl2. The pH of the
solution was adjusted to 7.5 with aqueous potassium hydroxide,
and 25 mg of commercial E. coli transketolase [0.16 unit/mg
(Sigma)], 120 mg of freshly prepared recombinant (E. coli K-12)
(vii) Bacteroides thetaiotaomicron D-glycero-D-manno-Hep-
tose-1,7-bisphosphate Phosphatase (SwissProt accession
number Q8AAI7). This enzyme was purified from the cell
supernatant by using the purification procedure described for the
E. coli enzyme, with the modification that the protein solution was
adjusted to 15% (NH4)2SO4 before being loaded onto the Butyl-
Sepharose column, and a 15 to 0% (NH4)2SO4 gradient in buffer A
was used as the eluant: yield, 7 mg of protein/g of wet cells.
(viii) Mesorhizobium loti D-glycero-D-manno-Heptose-
1,7-bisphosphate Phosphatase (SwissProt accession num-
ber Q98I56). The E. coli clone was obtained from the PSI
Materials Repository (clone MlCD00090897). The E. coli cells
were grown at 37 ꢀC and 200 rpm in 6 L of LB medium
(containing 50 μg/mL kanamycin) to an OD600 of 0.6-1.0,
D
-sedoheptulose-7-phosphate isomerase, and 120 mg of re-
combinant A. thermoaerophilus [R-C(1)OH specific] -glycero-
-manno-heptose-7-P kinase were added. The solution was incu-
D
D
bated overnight at 25 ꢀC and subjected to the same workup and
purification procedure described above to provide 600 mg of
1
D
-glycero-
D
-manno-heptose 1R,7-bisphosphate (84% yield): H
NMR δ 5.20 (d, 1H, J=8.0 Hz), 4.03 (dd, 1H, J=6.75 Hz, J=
3.0 Hz), 3.90 (td, 1H, J=7.8 Hz, J=3.5 Hz), 3.81 (m, 2H), 3.77
(t, 1H), 3.55 (t, 1H, J=3.0 Hz), 3.49 (td, 1H, J=10.0 Hz, J=3.0
Hz); 13C NMR δ 94.9 (d, J=3.8 Hz), 73.1, 70.6 (d, J=6.9 Hz),
70.4 (d, J=6.6 Hz), 70.1, 66.9, 64.5 (d, J=3.4 Hz); 31P NMR δ
3.42, 1.47; mass, calcd for C7H15O13P2 368.9988 Da, found
368.9981 Da.
induced with 5 mM
L-arabinose at 25 ꢀC, and harvested by
centrifugation upon reaching an OD600 of ∼1.7. The cell pellet
(30 g) was suspended in 300 mL of ice-cold lysis buffer [50 mM
NaH2PO4, 300 mM NaCl, and 10 mM imidazole (pH 8.0)] and
passed through a French pressure cell press at 1500 psi. The lysate
was clarified by centrifugation and chromatographed on a Ni-
NTA column equilibrated with lysis buffer. The column was
washed with 50 mM NaH2PO4, 300 mM NaCl, and 20 mM
imidazole (pH 8.0) before the protein was eluted with 50 mM
NaH2PO4, 300 mM NaCl, and 250 mM imidazole (pH 8.0). The
desired fractions were combined, concentrated, and dialyzed
against 50 mM KþHEPES and 10 mM MgCl2 (pH 7.5): yield,
20 mg of protein/g of wet cells.
(iii) D-glycero-D-manno-Heptose 1β-Phosphate and D-glycero-
D-manno-Heptose 1R-Phosphate.
D
-glycero-
D-manno-Heptose
1β,7-bisphosphate (30 mg) and -glycero-
D
D-manno-heptose 1R,
7-bisphosphate (80 mg) were separately dissolved in 1 mL of a
10 mM MgCl2/10 mM KþHEPES mixture (pH 7.5) to which
an aliquot (1 or 20 μL) of 5 mM E. coli GmhB (vide supra) was
then added. Following a 12 h incubation period at 25 ꢀC,
each reaction mixture was subjected to anion exchange column
chromatography (vide supra) to provide
heptose 1β-phosphate in 83% yield and
heptose 1R-phosphate in 92% yield.
D
-glycero-
D
-manno-
-manno-
D-glycero-
D
(a) D-glycero-D-manno-Heptose 1β-phosphate: 1H NMR
δ 4.90 (d, 1H, J=8.5 Hz), 3.84 (t, 1H, J=3.5 Hz), 3.80 (s, 1H),
3.59 (m, 2H), 3.49 (m, 2H), 3.30 (t, 1H); 13C NMR δ 94.9, 76.6,
72.3, 70.9, 70.4 (d), 66.7, 61.0; 31P NMR δ 0.965; mass, calcd for
C7H14O10P 289.0325 Da, found 289.0300 Da.
Preparation of GmhB Substrates and Substrate Ana-
logues. (i) D-glycero-D-manno-Heptose 1β,7-Bisphosphate.
Ribose 5-phosphate (0.310 g, 1.0 mmol), 0.30 g of fructose
6-phosphate (1.0 mmol), 1.10 g of ATP (2.0 mmol), 10 μL
(100 mM stock, 1 μmol) of
D
-glyceraldehyde 3-phosphate, and
(b) D-glycero-D-manno-Heptose 1R-phosphate: 1H NMR
δ 4.98 (d, 1H, J=8.5 Hz), 3.68 (p, 1H, J=3.0 Hz), 3.58 (m, 3H),
3.42 (m, 3H); 13C NMR δ 94.4 (d, J=4.5 Hz), 71.9, 71.5, 70.3 (d,
J= 7.2 5 Hz), 69.9, 67.5, 61.3; 31P NMR δ 2.36 (d, J=7.3 Hz);
mass, calcd for C7H14O10P 289.0325 Da, found 289.0330 Da.
(iv) Preparation of Glucose 1β,6-Bisphosphate. Maltose
(0.36 g, 1.0 mmol) and 0.4 g (3.0 mmol) of KH2PO4 were
dissolved in 100 mL of a 50 mM Tris/10 mM MgCl2 mixture.
The pH of the solution was adjusted to 7.5, and freshly pre-
pared recombinant Lactococcus lactis maltose phosphorylase
(100 units) (17) was added. Following a 12 h incubation period
100 μL (100 mM stock, 10 μmol) of thiamine pyrophosphate
were combined with 100 mL of 50 mM Tris containing 10 mM
MgCl2. The pH of the solution was adjusted to 7.5 with aqueous
potassium hydroxide, and 10 mg of commercial E. coli transke-
tolase [0.16 unit/mg (Sigma)], 60 mg of freshly prepared recom-
binant (E. coli K-12)
(3 mg/mL), and 4 mg of freshly prepared recombinant B. bron-
chiseptica [β-C(1)OH specific] -glycero- -manno-heptose-7-P
kinase were added. The solution was incubated overnight at
D-sedoheptulose-7-phosphate isomerase
D
D
-
25 ꢀC and chromatographed on a Dowex 1X8-100 (HCO3
)

1076 Biochemistry, Vol. 49, No. 6, 2010
Wang et al.
at 25 ꢀC, ATP (1.0 g, 1.9 mmol) was added. The pH of the
solution was adjusted to 9.0 before addition of commercial
Bacillus stearothermophilus fructose-6-phosphate kinase (100
units). Following a 12 h incubation period at 25 ꢀC, the reaction
mixture was subjected to the same purification procedure de-
scribed above to yield 180 mg of glucose 1β,6-bisphosphate:
1H NMR δ 4.67 (d, 1H, J=14 Hz), 3.82 (dd, 1H, J=11 Hz, J=
5 Hz), 3.76 (s, 1H), 3.31 (s, 3H), 3.10 (t, 1H); 13C NMR δ 94.9 (d,
J = 4.25 Hz), 73.2 (d, J = 6.75 Hz), 72.7, 72.2 (d, J = 5.75 Hz),
66.7, 60.7; 31P NMR δ 4.80, 2.96; mass, calcd for C6H13O12P2
338.9882 Da, found 338.9883 Da.
(viii) D-glycero-D-manno-Heptose 1β-ADP and D-Glu-
cose 1β-ADP.
(0.01 g, 0.038 mmol) and 0.033 g of ATP (0.06 mmol) or
-glucose 1β-phosphate (100 mg, 0.3 mmol) (vide supra) and
D-glycero-D-manno-Heptose 1β-phosphate
D
0.17 g of ATP (0.3 mmol) were dissolved in 5 mL of 50 mM Tris
containing 10 mM MgCl2. The pH of the solution was adjusted to
7.5 with aqueous potassium hydroxide before addition of a 1 mL
aliquot from a 1 mg/mL stock solution of B. bronchiseptica
D
-glycero-D-manno-heptose-1β-phosphate adenylyltransferase (vide
supra). The solution was incubated for 1 h at room temperature
and then chromatographed on a Sephadex G-10 column using
deionized water as the eluant. The desired fractions were
(v) Preparation of Mannose 1β,6-Bisphosphate and
Mannose 1R,6-Bisphosphate. In two separate solutions, man-
nose (0.30 g, 1.67 mmol) and 2.0 g (3.6 mmol) of ATP were
dissolved in 100 mL of 50 mM Tris containing 10 mM MgCl2.
The pH of the solutions was adjusted to 7.5 with aqueous
potassium hydroxide before addition of 1000 units of commercial
hexokinase to each. Following a 12 h incubation period at 25 ꢀC,
combined and lyophilized to give 20 mg of
D-glycero-D-manno-
heptose 1β-ADP or 52 g of -glucose 1β-ADP.
D
(a) D-glycero-D-manno-Heptose 1β-ADP: 1H NMR δ
8.48 (s, 1H), 8.21 (s, 1H), 6.12 (d, 1H, J=5.5 Hz), 5.19 (d, 1H, J=
8.5 Hz), 4.52 (t, 1H, J=4.5 Hz), 4.39 (s, 1H), 4.22 (s, 2H), 4.06 (d,
1H, J=2.5 Hz), 3.99 (s, 1H), 3.74 (d, 2H, J=3.5 Hz), 3.68 (t, 1H,
J=9.5 Hz), 3.62 (dd, 1H, J=9.5 Hz, J=2.5 Hz), 3.43 (dd, 1H,
J = 9.5 Hz, J = 2.5 Hz); 13C NMR δ 154.9, 152.1, 148.4, 139.2,
117.9, 95.2 (d, J=3.6 Hz), 86.2, 83.2 (d, J=8.9 Hz), 76.3, 72.0,
71.1, 69.9 (d, J=7.0 Hz), 69.7, 66.4, 64.6 (d, J=6.1 Hz), 61.0; 31P
NMR δ -10.50 (d, J=17.4 Hz), -12.51 (dd, J=16.97 Hz, J=
6.46 Hz); mass, calcd for C17H26N5O16P2 618.0850 Da, found
618.0845 Da.
5 mL of 2 mM B. bronchiseptica
kinase and A. thermoaerophilus
D
-glycero-
D
-manno-heptose-7-P
-manno-heptose-7-P
D-glycero-
D
kinase (vide supra) were added to the individual solutions. After
36 h, the reaction solutions were chromatographed as described
for D-glycero-D-manno-heptose 1β,7-bisphosphate (vide supra) to
give 170 mg of mannose 1β,6-bisphosphate and 470 mg of
mannose 1R,6-bisphosphate, respectively.
(a) Mannose 1β,6-bisphosphate: 1H NMR δ 4.86 (d, 1H,
J=9 Hz), 3.78 (m, 2H), 3. (dd, 1H, J=11.5 Hz, J=6 Hz), 3.48
(d, 2H, J = 7.5 Hz), 3.24 (s, 1H); 13C NMR δ 94.8, 75.5 (d, J =
7.13 Hz), 72.0, 71.1 (d, J=4.25 Hz), 65.8, 62.7 (d, J=2.75 Hz);
31P NMR δ 4.71, 2.81; mass, calcd for C6H13O12P2 338.9882 Da,
found 338.9870 Da.
(b) D-Glucose 1β-ADP: 1H NMR δ 8.49 (s, 1H), 8.24 (s,
1H), 6.13 (d, 1H, J = 6.0 Hz), 5.00 (t, 1H, J = 7.5 Hz), 4.53 (s,
1H), 4.39 (s, 1H), 4.22 (s, 2H), 3.87 (d, 1H, J = 12.5 Hz), 3.67
(dd, 1H, J= 12.0 Hz, J=6.0 Hz), 3.51 (dd, 2H, J=19 Hz, J=
8.5 Hz), 3.36 (t, 2H, J=8.0 Hz); 13C NMR δ 155.0, 152.3, 148.5,
139.3, 118.0, 97.5 (d, J=5.5 Hz), 86.5, 83.4 (d, J=8.5 Hz), 76.0,
74.7, 73.2 (d, J=8.0 Hz), 70.0, 69.0, 64.9 (d, J=5.3 Hz), 60.3; 31P
NMR δ -10.39 (d, J = 20.0 Hz), -12.14 (dd, J = 20.0 Hz, J =
7.27 Hz); mass, calcd for C16H24N5O15P2 588.0744, found
588.0734.
(b) Mannose 1R,6-bisphosphate: 1H NMR δ 5.06 (d, 1H,
J=8 Hz), 3.72 (m, 4H), 3.64 (d, 1H, J=9.50 Hz), 3.52 (t, 1H, J=
10.0 Hz); 13C NMR δ 94.6 (d, J=3.5 Hz), 71.9 (d, J=7.0 Hz),
70.5 (d, J=6.5 Hz), 69.2, 65.9, 62.6 (d, J=2.75 Hz); 31P NMR δ
4.30, 1.98; mass, calcd for C6H13O12P2 338.9882 Da, found
338.9888 Da.
Substrate Activity Screens. (i) 31P NMR Analysis.
E. coli GmhB-catalyzed hydrolysis reactions of the sugar bisphos-
(vi) Preparation of Sedoheptulose 1,7-Bisphosphate. Ery-
throse 4-phosphate (25 mg, 0.1 mmol) and dihydroxyacetone
phosphate (20 mg, 0.1 mmol) were dissolved in 100 mL of 50 mM
Tris containing 10 mM MgCl2. The pH of the solution was
adjusted to 7.5 with aqueous potassium hydroxide before addi-
tion of 50 μL of 200 μM recombinant Giardia lamblia fructose-
1,6-bisphosphate aldolase (18). After 15 h, the reaction solution
phates
D
-glucose 1R,6-diphosphate,
-mannose 1R,6-bisphosphate,
-sedoheptulose 1,7-bisphosphate,
-glycero-
D
-glucose 1β,6-bisphosphate,
D
D
D
D
-mannose 1β,6-bisphosphate,
-fructose 1R,6-bisphosphate,
-glycero-
D
D
D
-manno-heptose 1R,7-bisphosphate, and
D
-manno-heptose 1β,7-bisphosphate were examined by using
31P NMR in the proton-decoupled mode. The reaction solutions
initially contained 10 mM sugar bisphosphate, 1 or 50 μM
GmhB, 2 mM MgCl2, and 50 mM KþHEPES (pH 7.5, 25 ꢀC)
and were incubated for 5-30 h. 31P NMR spectra of inorganic
phosphate and the anticipated sugar monophosphate products
were measured for use as chemical shift standards.
was chromatographed as described for D-glycero-D-manno-hep-
tose 1β,7-bisphosphate (vide supra) to yield 30 mg of sedoheptu-
lose 1,7-bisphosphate (19): mass, calcd for C7H14O10P 368.9988
Da, found 368.9961 Da.
(vii) L-Histidinol Phosphate. This compound was prepared
(ii) Spectrophotometric Analysis. The rate of E. coli
GmhB-catalyzed hydrolysis of phosphate esters was measured
spectrophotometrically by using the Enzcheck phosphate
assay kit (Invitrogen) to continuously monitor phosphate
formation (at 360 nm, Δε = 11 mM-1 cm-1). The 1 mL
reaction solutions initially contained 50 mM Tris-HCl (pH 7.5,
25 ꢀC), 1 mM MgCl2, 1 mM substrate, 10-20 μM E. coli
GmhB, 0.2 mM MESG, and 1 unit of purine nucleoside
phosphorylase. The background level of phosphate release
was measured using the corresponding control reaction, which
excluded the GmhB. In cases where the catalyzed and control
reactions produced similar absorbance versus time traces, the
phosphate ester was judged not to be a substrate and the
from
L-histidinol and polyphosphoric acid using a modification
of the published procedure (20). A mixture of
L-histidinol
dihydrochloride (1 g, 4.5 mmol) and polyphosphoric acid (4 g)
was heated on a steam bath for 1 h, after which time 50 mL of ice-
cold water was added. The pH of the solution was adjusted to
7.5 with aqueous barium hydroxide, and the resulting precipitant
was removed by centrifugation. The supernatant was chromato-
graphed as described for
D-glycero-D-manno-heptose 1β,7-bi-
sphosphate (vide supra) to yield 300 mg of
L
-histidinol phos-
phate: 1H NMR δ 8.16 (s, 1H), 7.18 (s, 1H), 3.89 (s, 1H), 3.77 (s,
1H), 3.61 (s, 1H), 3.04 (m, 2H); 13C NMR δ 134.7, 129.3, 117.4,
62.4, 51.3, 24.8; 31P NMR δ 5.3; mass, calcd for C6H11N3O4P
220.0487 Da, found 220.0468 Da.
kcat defined as <1 ꢀ 10-4 s-1
.

Article
Steady-State Kinetic Constant Determination. Steady-
Biochemistry, Vol. 49, No. 6, 2010 1077
D
-sedoheptulose 1,7-bisphosphate (Figure 2) displayed a modest
state kinetic constant determinations were conducted at 25 ꢀC
using reaction solutions initially containing GmhB (0.001-
10 μM), varying concentrations of phosphate ester (0.5-5Km),
1 mM MgCl2, 0.2 mM 2-amino-6-mercapto-7-methylpurine
ribonucleoside, 1 unit of purine nucleoside phosphorylase, and
50 mM Tris-HCl (pH 7.5). Data were fitted to the equation V0=
Vmax[S]/(Km þ [S]) with KinetAsyst I, where V0 is the initial
velocity, Vmax is the maximum velocity, [S] is the substrate
concentration, and Km is the Michaelis-Menten constant for
substrate. The kcat value was calculated from Vmax and [E]
according to the equation kcat = Vmax/[E], where [E] is the
level of activity [kcat/Km = 800 M-1 s-1 (Table 1)]. 31P NMR
spectral analysis of the reaction of 10 mM D-sedoheptulose 1,7-
bisphosphate with 50 μM GmhB in 50 mM Tris-HCl and 1 mM
MgCl2 (pH 7.5, 25 ꢀC, 30 h) revealed that one of the two phos-
phate groups (resonances coincide at 5.5 ppm) was removed
(indicated by the appearance of an orthophosphate resonance at
3.3 ppm) but did not identify which phosphate group was
removed. Together, these findings, along with the kinetic con-
stants reported in Table 1, show that E. coli GmhB is not active
with monophosphorylated hexoses and heptoses and that it has
limited activity with ketohexose and ketoheptose bisphosphates.
E. coli GmhB was shown to dephosphorylate its physiological
enzyme concentration. The inhibition constants for
-manno-heptose 1β-phosphate and histidinol phosphate were
determined by measuring the initial velocities of E. coli GmhB-
catalyzed hydrolysis of -glycero- -manno-heptose 1β,7-bispho-
D-glycero-
D
substrate D-glycero-D-manno-heptose 1β,7-bisphosphate with
high efficiency [kcat/Km = 7 ꢀ 106 M-1 s-1 (Table 1)] and the
D
D
corresponding R-anomer,
D-glycero-D-manno-heptose 1R,7-bis-
sphate as a function of substrate and inhibitor concentration [I].
The initial velocity data were fitted to the equation V =
(Vmax[S])/[Km(1 þ [I]/Kis) þ [S](1 þ [I]/Kii)] to define the slope
and intercept inhibition constants, Kis and Kii, respectively.
phosphate, with reduced, yet physiologically significant, effi-
ciency [kcat/Km =7ꢀ 104 M-1 s-1 (Table 1)]. 31P NMR analysis
of the reaction between 10 μM E. coli and 10 mM D-glycero-
D
-manno-heptose 1β,7-bisphosphate [C(1)P δ = 3.0 ppm; C(7)P
δ = 4.8 ppm] or -glycero- -manno-heptose 1R,7-bisphosphate
D
D
RESULTS AND DISCUSSION
[C(1)P δ=2.9 ppm; C(7)P δ=5.2 ppm] in 50 mM Tris-HCl and
1 mM MgCl2 (pH 7.5, 25 ꢀC, 5 h) revealed complete removal of
the substrate C(7) phosphate group and no detectable hydrolysis
GmhB Substrate Specificity Profile Determination. The
participation of the GmhB orthologs from A. thermoaerophilus
and E. coli in the respective biosynthetic pathways leading to
of the corresponding
D-glycero-D-manno-heptose 1β-phosphate
D
-glycero-
D
-manno-heptose 1R-GDP and
L-glycero-D-manno-
[C(1)P δ=3.0 ppm] or
[C(1)P δ = 2.9 ppm] product. Thus, E. coli GmhB is specific
in targeting removal of the C(7)P of
1,7-bisphosphate.
D
-glycero-D
-manno-heptose 1R-phosphate
heptose 1β-ADP (Figure 1) was previously demonstrated by
reconstitution of the pathways in vitro (4, 5). However, the
catalytic efficiencies and anomeric specificities of the two phos-
phatases were not reported. More recently, E. coli GmhB (along
with 22 other E. coli HAD phosphatases) was screened for
activity against an 80-compound library of common organopho-
sphatemetabolitesrepresenting avarietyofstructural classes(21).
The E. coli GmhB was found to display activity toward only
D-glycero-D-manno-heptose
D-glycero-D-manno-Heptose 1β-phosphate and
D-glycero-
D-manno-heptose 1R-phosphate were synthesized (Figure 2)
and shown by using a spectrophotometric assay for orthopho-
sphate to have no detectable substrate activity with E. coli GmhB.
This observation agrees with the findings of the 31P NMR
one library compound, fructose 1,6-bisphosphate (kcat/Km
=
analysis. The
D-glycero-D-manno-heptose 1β-phosphate was
450 M-1 s-1). The first objective of this study was to measure the
catalytic efficiency of E. coli GmhB toward its physiological
substrate and further define its substrate specificity profile. This
was accomplished by preparing and measuring the substrate
activities of natural hexose 1,6-bisphosphates and heptose 1,7-
shown to be a noncompetitive inhibitor. On the basis of the
measured inhibition constants (Kis=60 ( 10 μM, and Kii=150 (
50 μM), it is evident that D-glycero-D-manno-heptose 1β-phos-
phate binds to the GmhB active site tightly, yet nonproductively.
Histidinol phosphate was tested as a substrate for E. coli
GmhB, but no detectable activity was observed. Although
histidinol phosphate bears little resemblance to the physiological
substrate of GmhB, the proposed evolutionary link between
GmhB and the histidinol-phosphate phosphatase (1, 3, 22) made
some amount of cross reactivity a possibility. Histidinol phos-
phate proved to be a noncompetitive inhibitor of E. coli GmhB
with a modest binding affinity (Kis =1 ( 0.1 mM, and Kii =6 (
bisphosphates, including the R- and β-anomers of
-manno-heptose 1β,7-bisphosphate.
With the exceptions of fructose 1,6-bisphosphate and glucose
D-glycero-
D
1R,6-bisphosphate, the hexose 1,6-bisphosphates and heptose
1,7-bisphosphates were prepared by enzymatic synthesis as
illustrated in Figure 2 (see Materials and Methods for details).
Fructose 1,6-bisphosphate proved to be a poor substrate for
E. coli GmhB [kcat/Km = 30 M-1 s-1 (Table 1)]. 31P NMR
spectral analysis of the reaction between 10 mM fructose 1,6-
bisphosphate and 50 μM GmhB in 50 mM Tris-HCl and 1 mM
MgCl2 (pH 7.5, 25 ꢀC, 30 h) revealed that the substrate C(1)P was
selectively removed. Specifically, whereas the spectrum of the
fructose 1,6-bisphosphate contains resonances at 5.0 ppm [C(1)P]
and 4.7 ppm [C(6)P], the spectrum of the product mixture
exhibited resonances at 4.7 ppm (fructose 6-phosphate) and 3.3
ppm (orthophosphate). Fructose 6-phosphate and fructose
1-phosphate were subsequently shown to have no detectable
substrate activity (kcat < 1 ꢀ 10-4 s-1) by using a spectropho-
metric assay for orthophosphate. Glucose 1R,6-bisphosphate,
glucose 1β,6-bisphosphate, mannose 1R,6-bisphosphate, and
mannose 1β,6-phosphate (Figure 2) were also found to have
no detectable substrate activity with E. coli GmhB; however,
1 mM). Therefore, like
D-glycero-D-manno-heptose 1β-phosphate
and -glycero- -manno-heptose 1R-phosphate, histidinol phos-
D
D
phate appears to escape GmhB catalysis by binding in a non-
productive mode.
Together, these results characterize E. coli GmhB as a HAD
phosphatase with a narrow substrate range and a high catalytic
efficiency toward its physiological substrate. Because these two
properties are characteristic of a highly evolved metabolic
enzyme, they indicate that the evolution of GmhB has occurred
under strong selection pressure. As HAD phosphatases primarily
assume the roles of repair enzymes, housekeeping enzymes, or
phosphate scavengers (23), they tend to be inefficient and lax in
their substrate specificities. The 80-compound screen applied to
the 22 HAD phosphatases of E. coli revealed a general tendency
toward substrate promiscuity (21). Focused substrate activity

1078 Biochemistry, Vol. 49, No. 6, 2010
Wang et al.
FIGURE 2: Chemical steps and enzyme catalysts used in the synthesis of the hexose and heptose bisphosphates used in the substrate profile
determination of GmhB.
screening designed to explore structural space closely tied to
the respective physiological substrates of two HAD house-
keeping enzymes, E. coli NagD (a nucleoside monophosphatase)
and BT4131 from B. thetaiotaomicron (a hexose-6-phosphate
phosphatase), showed that although the two enzymes do
not hydrolyze organophosphates from other structural classes,

Article
Biochemistry, Vol. 49, No. 6, 2010 1079
Table 1: Steady-State Kinetic Constants for E. coli GMHB-Catalyzed Hydrolysis of Phosphate Esters in 50 mM Tris-HCl (pH 7.5, 25 ꢀC) Containing 1 mM
a
MgCl2
substrate
Pb
kcat (s-1
)
Km (μM)
kcat/Km (M-1 s-1
)
D
D
-glycero-
-glycero-
D
-manno-heptose 1β,7-bisphosphate
-manno-heptose 1R,7-bisphosphate
C(7)P
35.7 ( 0.2
4.6 ( 0.1
0.51 ( 0.01
5.0 ( 0.1
67 ( 1
610 ( 20
1501 ( 5
7 ꢀ 106
7 ꢀ 104
8 ꢀ 102
3 ꢀ 101
D
C(7)P
sedoheptulose 1,7-bisphosphate
C(7) or C(1)
C(1)P
fructose 1,6-bisphosphate
0.039 ( 0.001
aCompounds that were tested with the E. coli or B. bronchiseptica GmhB but failed to display substrate activity (kcat < 1 ꢀ 10-4 s-1) include fructose
6-phosphate, mannose 6-phosphate, fructose 1-phosphate, glucose 1R,6-bisphosphate, glucose 1β,6-bisphosphate, mannose 1R,6-bisphosphate, mannose 1β,
6-phosphate, histidinol phosphate, phosphoenolpyruvate, uridine 5-diphosphate, adenosine 5-monophosphate, and dimyristoyl-sn-glycero-3-phosphate. bPhos-
phate removed in the catalyzed reaction.
organophosphates that belong to the same structural class as the
Table 2: Steady-State Kinetic Constants for Hydrolysis of
-manno-Heptose 1R,7-Bisphosphate and -glycero- -manno-Heptose
-manno-Heptose-1,7-bisphos-
D-glycero-
physiological substrates are hydrolyzed with modest efficiency
(kcat/Km = 1 ꢀ 102 to 1 ꢀ 104 M-1 s-1) and with little dis-
tinction (24, 25). E. coli GmhB does not fit this profile but instead
mirrors that of E. coli 2-keto-3-deoxyoctulosonate-8-phosphate
(KDO-8-P) phosphatase, a member of the HAD “8KDO”
subfamily (26, 27). Like E. coli GmhB, KDO-8-P phosphatase
functions in the Lipid A core glycan biosynthetic pathway.
Although these two phosphatases support different branches of
the Lipid A core pathway, we propose that their evolution may
have been subject to a common selection pressure.
D
D
D
1β,7-Bisphosphate by the Selected D-glycero-D
phate Phosphatases (GmbBs)a
kcat
Km
k
cat/Km
enzyme source
E. coli
anomer
(s-1
)
(μM)
(M-1 s-1
)
β/R
R
β
R
β
R
β
R
β
4.6 ( 0.1
67 ( 1
6.9 ꢀ 104 100
E. coli
35.7 ( 0.2 5.0 ( 0.1 7.1 ꢀ 106
B. bronchiseptica
B. bronchiseptica
M. loti
5.9 ( 0.2 280 ( 30
2.1 ꢀ 104 150
22 ( 0.5 6.9 ( 0.5 3.2 ꢀ 106
4.4 ( 0.1
18 ( 0.2
29 ( 1
58 ( 1
7.6 ꢀ 104
18
0.17
M. loti
13 ( 0.2 1.4 ꢀ 106
Tuning GmhB Anomeric Specificity for Biochemical
Function. An interesting feature of the specialization of GmhB
is its anomeric specificity. Whereas the physiological substrate for
B. thetaiotaomicron
B. thetaiotaomicron
33 ( 1
86 ( 1
8.8 ꢀ 105
13 ( 1
1.5 ꢀ 105
aThe reactions were conducted in 50 mM Tris-HCl (pH 7.5, 25 ꢀC)
containing 1 mM MgCl2.
the E. coli enzyme is the β-anomer of D-glycero-D-manno-heptose
1,7-bisphosphate, the R-anomer is the physiological substrate of
the A. thermoaerophilus GmhB (Figure 1). A priori, one might
have expected that GmhB would lack the ability to discriminate
the spatial disposition of the substrate C(1) phosphate. However,
this work has shown that E. coli GmhB prefers the β-anomer over
the R-anomer by a factor of 100:1 (Table 1). If this specificity is a
consequence of selection pressure, then the reverse might hold
true for the A. thermoaerophilus GmhB; i.e., it would show a
preference for the R-anomer over the β-anomer. Placed in a
broader context, the important question is “whether the anomeric
specificities of the pathway kinase and phosphatase co-evolved”.
To address this issue of anomeric specificity and physiological
substrate, comparisons were made for four GmhB orthologs
derived from distantly related bacterial species. The selection of
orthologs for study was guided by technical concerns that
included the availability of the GmhB clone (or the genomic
DNA for gene cloning) and protein solubility, as well whether the
physiological substrate could be assigned with confidence using
bioinformatics analysis. In addition, the availability of the X-ray
crystal structure, which would allow anomeric specificity to be
analyzed in the context of active site structure, was also a factor
guiding ortholog selection.
were cloned and the substrate specificity of the purified enzymes
was determined. The genes were located within the B. bronchi-
septica genome via Blast searches using the sequences of the
E. coli and A. thermoaerophilus pathway kinases and nucleotidyl
transferases as queries. The sequence of bifunctional kinase/
adenyl transferase (HldE_k/HldE_t) from E. coli identified the
putative B. bronchiseptica kinase Q7WGU8 (51% identical) and
nucleotidyl transferase Q7WF17 (48% identical). The purified
HldE_k homologue Q7WGU8 was shown to selectively produce
D-glycero-D-manno-heptose 1β,7-bisphosphate when incubated
with ATP and
D
-glycero- -manno-heptose 7-phosphate in buf-
D
fered solution (Figure 2) (see Materials and Methods for details
of product isolation and spectral properties). Likewise, the puri-
fied HldE_t homologue Q7WF17 was shown to form
D
-glycero-
-manno-heptose 1β-ADP upon incubation with ATP and
-glycero- -manno-heptose 7-phosphate in buffered solution
D
D
D
(see Materials and Methods for details of product isolation and
spectral properties). No product was obtained when GTP, UTP,
or CTP was substituted for ATP. Moreover, no product was
obtained when
substituted for
D
-glycero-
D
-manno-heptose 1R-phosphate was
-manno-heptose 1β-phosphate. On
The steady-state kinetic constants were measured for GmhB-
D-glycero-
D
catalyzed hydrolysis of the R- and β-anomers of
D
-glycero-
the other hand, Q7WF17 was shown to catalyze ATP phosphory-
lation of glucose 1β-phosphate to form glucose 1β-ADP, as
demonstrated by spectral analysis of the purified compound.
Thus, Q7WF17 is specific for ATP as the phosphoryl donor, and
for the β-anomer of hexose or heptose 1-phosphate as the adenyl
group acceptor.
D
-manno-heptose 1,7-bisphosphate (Table 2). The orthologs from
E. coli, B. bronchiseptica, and M. loti were shown to prefer the
β-anomer (by 100-, 150-, and 18-fold, respectively), whereas the
GmhB from B. thetaiotaomicron prefers the R-anomer
(6-fold). To correlate the observed GmhB anomeric preference
with the anomeric configuration of the physiological substrate, we
investigated the anomeric specificity of the kinase and nucleotidyl
transferase of the corresponding biosynthetic pathway (Figure 1).
In the case of GmhB from B. bronchiseptica (Q7WG29), the
genes encoding the pathway kinase and nucleotidyl transferase
On the basis of the findings of a Blast search of the B. bronchi-
septica genome, we conclude that only one set of pathway kinase-
phosphatase-transferase activities is present and that the three en-
zymes (Q7WGU8, Q7WG29, and Q7WF17) convert the
D-glycero-
D
-manno-heptose 7-phosphate to -glycero- -manno-heptose 1β-ADP.
D
D

1080 Biochemistry, Vol. 49, No. 6, 2010
Wang et al.
The pathway phosphoheptose isomerase (Q7WEW5) and
D-gly-
(Figure 1). The gene cluster is located in one of the numerous
cero-D-manno-heptose-1β-ADP epimerase (Q7WGU9) were identi-
capsular polysaccharide biosynthetic loci of the B. thetaiotaomi-
fied using the sequences of the known E. coli enzymes as queries
(40 and 54% identical, respectively). Thus, all enzymes of the
cron genome. Thus, the likely function of the
heptose 1R-GDP pathway of B. thetaiotaomicron is to generate
activated -glycero- -manno-heptose units for incorporation into
the extracellular capsular polysaccharide. The physiological
substrate of the B. thetaiotaomicron GmhB is -glycero- -man-
D-glycero-D-manno-
L
-glycero-
D-manno-heptose 1β-ADP pathway are accounted for.
D
D
It can be concluded that the physiological substrate of the
B. bronchiseptica GmhB is the β-anomer and that it, like the
E. coli GmhB, displays significantly higher reactivity with the
β-anomer versus the R-anomer.
The biochemical function of GmhB from M. loti was also
examined. Blast searches of the M. loti genome identified a single
GmhB gene (Q98I56), single matches to the E. coli bifunctional
kinase/adenyl transferase (HldE_k/HldE_t) (M. loti Q98I54,
42% identical sequence), and no matches3 to the A. thermoaer-
D
D
no-heptose 1R,7-bisphosphate, which is shown to have an activity
constant (kcat/Km = 8.8 ꢀ 105 M-1 s-1) 6-fold higher than that
measured for the β-anomer (kcat/Km = 1.5 ꢀ 105 M-1 s-1
)
(Table 2).
SUMMARY AND CONCLUSIONS
The results of the studies described above demonstrate that the
bacterial HAD phosphatase GmhB has a narrow substrate range
and a high catalytic efficiency toward its physiological substrate
ophilus C(1)OH R-anomer specific
phosphate kinase (HddA) or the A. thermoaerophilus
-manno-heptose-1R-phosphate guanidyl transferase HddC. The
D-glycero-D-manno-heptose-7-
D-glycero-
D
D
-glycero-
GmhB is a highly evolved metabolic enzyme that functions in
the -glycero- -manno-heptose 1β-ADP and -glycero- -manno-
D-manno-heptose 1,7-bisphosphate. Consequently,
genes encoding HldE_k/HldE_t and GmhB are located in a gene
cluster, which includes genes that encode heptose-phosphate
isomerase GmhA (Q98I55; 35% identical with E. coli GmhA
L
D
D
D
heptose 1R-GDP pathways. The anomeric specificity of GmhB
correlates with that of the pathway kinase. Thus, the evolution-
ary path leading from an ancient ancestor to modern GmhB
diverged to support two distinct biochemical functions. Notably,
each of the four GmhB orthologs examined has retained a
physiologically significant level of activity toward the anomeric
antipode of the physiological substrate. Anomer selectivity may
have evolved after catalytic efficiency, or alternatively, structural
changes made to enhance the activity toward the physiological
substrate might have also enhanced the activity toward the
glycoside epimer, but to a lesser degree. On a practical level, a
P63224) and the D-glycero-D-manno-heptose-1β-ADP epimerase,
GmhD (Q98I52; 61% identical with E. coli GmhA P67910). The
presence of the epimerase, GmhD, is a strong indicator that the
end product of the GmhB pathway is
1β-ADP and thus that the physiological substrate of the M. loti
GmhB is -glycero- -manno-heptose 1β-phosphate.
L-glycero-D-manno-heptose
D
D
The biochemical functions of the GhmB orthologs in E. coli,
B. bronchiseptica, and M. loti are identical in that they support
the synthesis of
L-glycero-D-manno-heptose 1β-ADP for the
production of the Lipid A glycan core. The three bacterial species
derive from different subdivisions of the kingdom of proteobac-
teria (gamma, beta, and alpha, respectively). Even though the
sequences of three GmhB orthologs have diverged to ∼70%
nonidentity (as deduced from pairwise alignments), the catalytic
GmhB that dephosphorylates both D-glycero-D-manno-heptose
1,7-bisphosphate anomers might be capable of multitasking, i.e.,
supporting both biosynthetic pathways within the same bacter-
ium. Indeed, the expansion of GmhB biological range via
horizontal gene transfer might be facilitated if the GmhB activity
efficiencies toward their physiological substrate
D-glycero-
D
-manno-heptose 1β,7-bisphosphate [kcat/Km = 7.1, 3.3, and
acquired by the gene transfer is applicable to both
D-glycero-
1.4 ꢀ 106 M-1 s-1, respectively (Table 2)] are conserved as are
D-manno-heptose 1,7-bisphosphate anomers, rather than to a single
their lower catalytic efficiencies toward D-glycero-D-manno-heptose
anomer. Nevertheless, a bioinformatics-based search for evi-
dence of GmhB multitasking revealed that the co-occurrence of
the two heptose pathways in a single bacterium is rare. Moreover,
the two bacterial species that are found to contain both pathways
[viz. E. coli 052 (11) and Yersinia pseudotuberculosis (9)] also
contain two GmhB phosphatases, one for each pathway. Thus,
the occurrence of one GmhB specialist or the other (differing in
anomeric specificity) among bacterial species appears to be tied
1β,7-bisphosphate [kcat/Km = 6.6, 2.1, and 7.6 ꢀ 104 M-1 s-1
,
respectively (Table 2)]. However, the anomeric specificity of the
M. loti GmhB is noticeably lower (18:1) than those of the E. coli
GmhB (100:1) and B. bronchiseptica GmhB (150:1). The X-ray
structures of all three enzymes are now known and will be
discussed in the context of the catalytic efficiency and substrate
specificity in the following paper (15).
The final objective of this investigation was to demonstrate the
biochemical function of GmhB from B. thetaiotaomicron. Blast
searches were conducted to show that there is a single copy of a
GmhB encoding gene and that it is contained in a gene cluster
that encodes the homologues of all four of the A. thermoaer-
to the occurrence of the corresponding (
L-glycero-D-manno-
heptose 1β-ADP vs D-glycero-D-manno-heptose 1R-GDP) path-
way.
REFERENCES
ophilus D-glycero-D-manno-heptose 1R-GDP pathway enzymes:
1. Fani, R., Brilli, M., Fondi, M., and Lio, P. (2007) The role of gene
fusions in the evolution of metabolic pathways: The histidine bio-
synthesis case. BMC Evol. Biol. 7 (Suppl. 2), S4.
GmhA (Q8AA19; 58% identical sequence), HddA (Q8AAJ0;
58% identical sequence), GmhB (Q8AAI7; 34% identical
sequence), and HddC (Q8AAI8; 47% identical sequence)
2. Koonin, E. V., and Tatusov, R. L. (1994) Computer Analysis of
Bacterial Haloacid Dehalogenases Defines a Large Superfamily of
Hydrolases with Diverse Specificity: Application of an Iterative
Approach to Database Search. J. Mol. Biol. 244, 125–132.
3. Burroughs, A. M., Allen, K. N., Dunaway-Mariano, D., and
Aravind, L. (2006) Evolutionary genomics of the HAD superfamily:
Understanding the structural adaptations and catalytic diversity in a
superfamily of phosphoesterases and allied enzymes. J. Mol. Biol. 361,
1003–1034.
3The C(1)OH β-anomer specific
phate kinase (HldE_k) and the C(1)OH R-anomer specific
D
-glycero-
D
-manno-heptose-7-phos-
-glycero-
D
D-
manno-heptose-7-phosphate kinase (HddA) derive from different en-
zyme superfamilies (ribokinase and GHMP, respectively) and thus are
unambiguously distinguished at the amino acid sequence level. Like-
wise, the nucleotidyl transferases HldE_t and HddC of the respective
pathways have evolved within separate clades of the cytidyltransferase
superfamily and are thus easily distinguished at the amino acid sequence
level.
4. Kneidinger, B., Marolda, C., Graninger, M., Zamyatina, A.,
McArthur, F., Kosma, P., Valvano, M. A., and Messner, P. (2002)

Article
Biochemistry, Vol. 49, No. 6, 2010 1081
Biosynthesis pathway of ADP-
cherichia coli. J. Bacteriol. 184, 363–369.
5. Kneidinger, B., Graninger, M., Puchberger, M., Kosma, P., and
Messner, P. (2001) Biosynthesis of nucleotide-activated -glycero-
-manno-heptose. J. Biol. Chem. 276, 20935–20944.
6. Valvano, M. A., Messner, P., and Kosma, P. (2002) Novel pathways
for biosynthesis of nucleotide-activated glycero-manno-heptose pre-
cursors of bacterial glycoproteins and cell surface polysaccharides.
Microbiology 148, 1979–1989.
L
-glycero-β-
D
-manno-heptose in Es-
Heptose-1,7-bisphosphate Phosphatase, GmhB. Biochemistry DOI
10.1021/bi902019q.
16. Chen, P. S., Toribara, T. Y., and Warner, H. (1956) Microdetermina-
tion of Phosphorus. Anal. Chem. 28, 1756–1758.
17. Nilsson, U., and Radstrom, P. (2001) Genetic localization and
regulation of the maltose phosphorylase gene, malP, in Lactococcus
lactis. Microbiology 147, 1565–1573.
18. Galkin, A., Kulakova, L., Melamud, E., Li, L., Wu, C., Mariano, P.,
Dunaway-Mariano, D., Nash, T. E., and Herzberg, O. (2007) Char-
acterization, kinetics, and crystal structures of fructose-1,6-bispho-
sphate aldolase from the human parasite, Giardia lamblia. J. Biol.
Chem. 282, 4859–4867.
D
D
7. Kosma, P., Wugeditsch, T., Christian, R., Zayni, S., and Messner, P.
(1995) Glycan structure of a heptose-containing S-layer glycoprotein
of Bacillus thermoaerophilus. Glycobiology 5, 791–796.
8. Schaffer, C., and Messner, P. (2004) Surface-layer glycoproteins: An
example for the diversity of bacterial glycosylation with promising
impacts on nanobiotechnology. Glycobiology 14, 31R–42R.
9. Bruneteau, M., and Minka, S. (2003) Lipopolysaccharides of bacterial
pathogens from the genus Yersinia: A mini-review. Biochimie 85, 145–
152.
10. Kondakova, A. N., Ho, N., Bystrova, O. V., Shashkov, A. S.,
Lindner, B., Creuzenet, C., and Knirel, Y. A. (2008) Structural studies
of the O-antigens of Yersinia pseudotuberculosis O:2a and mutants
19. Smyrniotis, P. Z., and Horecker, B. L. (1956) The preparation of
sedoheptulose diphosphate. J. Biol. Chem. 218, 745–752.
20. Ames, B. N., and Horecker, B. L. (1956) The biosynthesis of histidine:
Imidazoleacetol phosphate transaminase. J. Biol. Chem. 220, 113–128.
21. Kuznetsova, E., Proudfoot, M., Gonzalez, C. F., Brown, G.,
Omelchenko, M. V., Borozan, I., Carmel, L., Wolf, Y. I., Mori, H.,
Savchenko, A. V., Arrowsmith, C. H., Koonin, E. V., Edwards,
A. M., and Yakunin, A. F. (2006) Genome-wide analysis of substrate
specificities of the Escherichia coli haloacid dehalogenase-like phos-
phatase family. J. Biol. Chem. 281, 36149–36161.
thereof with impaired 6-deoxy-D-manno-heptose biosynthesis path-
way. Carbohydr. Res. 343, 1383–1389.
22. Rangarajan, E. S., Proteau, A., Wagner, J., Hung, M. N., Matte, A.,
and Cygler, M. (2006) Structural snapshots of Escherichia coli
histidinol phosphate phosphatase along the reaction pathway. J. Biol.
Chem. 281, 37930–37941.
11. Feng, L., Senchenkova, S. N., Yang, J., Shashkov, A. S., Tao, J., Guo,
H., Cheng, J., Ren, Y., Knirel, Y. A., Reeves, P. R., and Wang, L.
(2004) Synthesis of the heteropolysaccharide O antigen of Escherichia
coli O52 requires an ABC transporter: Structural and genetic evi-
dence. J. Bacteriol. 186, 4510–4519.
12. Masoud, H., Perry, M. B., Brisson, J. R., Uhrin, D., Li, J., and
Richards, J. C. (2009) Structural elucidation of the novel core
oligosaccharide from LPS of Burkholderia cepacia serogroup O4.
Glycobiology 19, 462–471.
13. McArthur, F., Andersson, C. E., Loutet, S., Mowbray, S. L., and
Valvano, M. A. (2005) Functional analysis of the glycero-manno-
heptose 7-phosphate kinase domain from the bifunctional HldE
protein, which is involved in ADP-L-glycero-D-manno-heptose bio-
synthesis. J. Bacteriol. 187, 5292–5300.
14. Raetz, C. R., Guan, Z., Ingram, B. O., Six, D. A., Song, F., Wang, X.,
and Zhao, J. (2009) Discovery of new biosynthetic pathways: The lipid
A story. J. Lipid Res. 50 (Suppl.), S103–S108.
23. Allen, K. N., and Dunaway-Mariano, D. (2009) Markers of fitness in
a successful enzyme superfamily. Curr. Opin. Struct. Biol. 19, 658–665.
24. Tremblay, L., Zhang, G., Dai, J., Dunaway-Mariano, D., and Allen,
K. N. (2006) Structure and Activity Analyses of E. coli K-12 NagD
Provide Insight Into the Evolution of Biochemical Function in the
HAD Enzyme Superfamily. Biochemistry 45, 1183–1193.
25. Lu, Z., Dunaway-Mariano, D., and Allen, K. N. (2005) HAD super-
family phosphotransferase substrate diversification: Structure and
function analysis of HAD subclass IIB sugar phosphatase BT4131.
Biochemistry 44, 8684–8696.
26. Parsons, J. F., Lim, K., Tempczyk, A., Krajewski, W., Eisenstein, E.,
and Herzberg, O. (2002) From structure to function: YrbI from
Haemophilus influenzae (HI1679) is a phosphatase. Proteins 46,
393–404.
15. Nguyen, H., Wang, L., Huang, H., Peisach, E., Dunaway-Mariano,
D., and Allen, K. N. (2010) Structural Determinants of Substrate
27. Wu, J., and Woodard, R. W. (2003) Escherichia coli YrbI is 3-deoxy-
D
-manno-octulosonate 8-phosphate phosphatase. J. Biol. Chem. 278,
18117–18123.
Recognition in the HAD Superfamily Member D-glycero-D-manno-