Structure 14, 601–609, March 2006 ª2006 Elsevier Ltd All rights reservedDOI 10.1016/j.str.2005.12.012
Serendipitous Discovery and X-Ray Structure
of a Human Phosphate Binding Apolipoprotein
Renaud Morales,1Anne Berna,2Philippe Carpentier,1
Carlos Contreras-Martel,1Fre ´de ´rique Renault,3
Murielle Nicodeme,4Marie-Laure Chesne-Seck,6
Franc ¸ois Bernier,2Je ´ro ˆme Dupuy,1
Christine Schaeffer,7He ´le `ne Diemer,7
Alain Van-Dorsselaer,7Juan C. Fontecilla-Camps,1
Patrick Masson,3Daniel Rochu,3
and Eric Chabriere3,5,*
1Laboratoire de Cristallogene `se et Cristallographie
des Prote ´ines
Institut de Biologie Structurale JP EBEL
2Institut de Biologie Mole ´culaire des Plantes du CNRS
Universite ´ Louis Pasteur
67083 Strasbourg Cedex
3Unite ´ d’Enzymologie
De ´partement de Toxicologie
Centre de Recherches du Service de Sante ´ des Arme ´es
38702 La Tronche
4Laboratoire des Biosciences de l’Aliment
5Laboratoire de Cristallographie et Mode ´lisation des
Mate ´riaux Mine ´raux et Biologiques
Universite ´ Henri Poincare ´-Nancy 1
54506 Vandoeuvre-le `s-Nancy
6Laboratoire de Cristallographie Macromole ´culaire
Institut de Biologie Structurale JP EBEL
We report the serendipitous discovery of a human
plasma phosphate binding protein (HPBP). This 38
kDa protein is copurified with the enzyme paraoxo-
nase. Its X-ray structure is similar to the prokaryotic
phosphate solute binding proteins (SBPs) associated
with ATP binding cassette transmembrane trans-
porters, though phosphate-SBPs have never been
characterized or predicted from nucleic acid data-
bases in eukaryotes. However, HPBP belongs to the
family of ubiquitous eukaryotic proteins named
DING, meaning that phosphate-SBPs are also wide-
spread in eukaryotes. The systematic absence of com-
plete genes for eukaryotic phosphate-SBP from data-
sequence conservation between genes belonging to
evolutionary distant species suggests that the corre-
sponding proteins play an important function. HPBP
phate ions in human plasma and may become a new
predictor of or a potential therapeutic agent for phos-
phate-related diseases such as atherosclerosis.
Both cholesterol and HDL (high-density lipoprotein)
levels are commonly used as risk predictors of cardio-
vascular disease (Gordon and Rifkind, 1989; Amann
et al., 2003). HDL protects against atherosclerosis
mainly through the reverse cholesterol transport pro-
cess in which excess cholesterol is transferred from
peripheral tissues to the liver by the ATP binding cas-
sette ABCA1 transporter, which is defective in Tangier
disease (Brooks-Wilson et al., 1999). Although HDLs
have been extensively studied, their metabolic role
has not been completely elucidated yet. For instance,
while the involvement of the calcium-dependent, HDL-
associated paraoxonase 1 (PON1; aryldialkylphospha-
tase; EC 22.214.171.124) in atherosclerosis prevention is well
established (Watson et al., 1995; Shih et al., 1998), its
physiological function remains unknown. PON1, named
for its ability to hydrolyze the insecticide paraoxon, is
the subject of intensive research owing to its capacity
to inactivate various organophosphorous compounds,
including nerve gases and pesticides, representing both
a terrorist threat and an environmental hazard. Albeit its
native activity is suggested to be lactonase (Khersonsky
and Tawfik, 2005), PON1 is currently considered as an
enzyme with promiscuous functions. Accordingly, some
of the activities attributed to PON1 are for the moment
In preparing PON1 for functional studies, we have iso-
lated another protein that copurifies with it. Here, we re-
port the discovery and X-ray structure of this protein,
which is a human plasma solute binding protein (SBP)
that binds phosphate and belongs to the sixth family
of SBPs (Tam and Saier, 1993). In prokaryotes, SBPs
tous molecules that use ATP as an energy source by
which to carry a wide variety of molecules across mem-
branes (Higgins, 1992). ABC transporters may also play
a central role in pathologies such as cystic fibrosis and
are involved in resistance to chemotherapeutic drugs
(Jones and George, 2002). These proteins are com-
posed of two cytoplasmic nucleotide binding domains
and two channel-forming transmembrane domains
(Chang and Roth, 2001). In both archaea and bacteria,
interactions with extracytoplasmic SBPs largely con-
tribute to their efficiency and specificity. Prokaryotic
phosphate-SBPs are 38 kDa proteins, encoded by
PstS genes (Surin et al., 1984), that bind inorganic phos-
phate ions with submicromolar affinity (Medveczky and
Rosenberg, 1971) and are mostly expressed during
phosphate starvation (Gerdes and Rosenberg, 1974).
In gram-negative bacteria, these proteins are soluble
and periplasmic, whereas in gram-positive bacteria,
they are lipoproteins anchored to the outer face of the
membrane (Young and Garbe, 1991).
Interestingly, eukaryotic ABC transporters do not rely
on SBPs. Proteins related to SBPs nevertheless exist in
eukaryotes (Felder et al., 1999), the more relevant cases
being the glutamate receptors, the atrial natriuretic pep-
tide receptors involved in blood pressure regulation,
and the calcium-sensing G protein-coupled receptors,
located in parathyroid and kidney cells. Anchored to
the membrane with an extramembranar domain, they
act in a multimeric state as receptors.
This article describes the discovery, characterization,
sequencing, and crystal structure at 1.9 A˚resolution of
HPBP, a novel, to our knowledge, human HDL-associ-
ated apolipoprotein, not predicted by any genomic da-
tabase.Thisprotein,structurally andfunctionally related
to bacterial phosphate-SBPs, may be the first phos-
phate transporter characterized in human plasma. A
possible use of the protein in diagnosis or therapy of
atherosclerosis emerges from its capability of binding
phosphate, as well as a role for stabilizing functionally
active conformation of PON1.
X-Ray Structure of HPBP
Crystals were obtained from an apparently pure PON1
solution (Contreras-Martel et al., 2006; see the Supple-
mental Data, available with this article online). Due toini-
tial difficulties in obtaining heavy atom derivatives, the
protein envelope was initially determined ab initio at
25 A˚resolution (Fokine et al., 2003). Eventually, a uranyl
at 1.9 A˚resolution and a polyalanine model was traced
automatically into this map (Figures 1A and 1B) (see Ex-
perimental Procedures). As the electron density did not
match the amino acid sequence of PON1, we carried out
structural comparisons with DALI (Holm and Sander,
1994). This analysis revealed that our three-dimensional
model (PDB ID code 2cap) displayed a fold very similar
to those of a phosphate binding protein from Escheri-
chia coli (Luecke and Quiocho, 1990) and the PstS1 pro-
tein from Mycobacterium tuberculosis, a major BCG
vaccine antigen (Vyas et al., 2003). Based on these ob-
servations, the protein was named human phosphate
binding protein (HPBP). The complete amino acid se-
quence ofHPBP was assigned from the electron density
map and later confirmed for about 75% of the protein by
direct sequencing (Figure 2 and Experimental Proce-
dures). HPBP consists of 376 residues with a predicted
molecular mass of 38.4 kDa.
Superposition of the three-dimensional structures
gave rms deviations for a carbon atom positions be-
tween HPBP and E. coli phosphate-SBP (216 atoms)
Figure 1. HPBP Structure
(A) Ribbon representation of HPBP showing
the two domains (blue and green) connected
by a hinge (yellow), the phosphate molecule
(red balls), and the two disulfide bridges
(C113–C158 and C306–C359, orange sticks).
The elongated protein structure exhibits two
similar globular domains, each constituted
by a five-stranded b sheet core (three parallel
b strands) flanked by a helices. A hydropathy
profile computed with the program DAS
(Cserzo et al., 1997) indicates that HPBP is
strongly hydrophobic between residues 168
and 194. This motif, located on the protein
surface (H.S.), is involved in the packing of
the crystallographicdimer. N and C represent
the protein termini.
(B) Stereoview of the Ca trace of HPBP. The
orientation is the same as in (A). Every tenth
a-carbon is labeled with its residue number.
Rainbow coloring—from the N terminus to
the C terminus—has been used.
(C) Structural comparison of different known
phophate-SBPs: HPBP is shown in blue, E.
coli PstS protein is shown in yellow, and M.
tuberculosis PstS1 is shown in green. The
four HPBP-specific loops are indicated.
by two antiparallel
and HPBP and M. tuberculosis PstS1 protein (202
atoms) of 1.53 A˚and 1.52 A˚, respectively. Four loops
protrudring from HPBP globular domains represent its
only major structural difference with bacterial SBPs
An HPBP electron density map displays a peak that is
completely buried in a cleft formed at the intersection of
the two domains and that could be assigned to either
a phosphate (HPO422) or a sulfate (SO422) ion (Fig-
This active site is very similar to other phosphate-SBPs.
Over the 8 residues involved in the binding of the sub-
strate, F11 in E. coli PBP is replaced by L9 in HPBP
(Figure 3B). Moreover, this change has only a minor ef-
fect because this residue binds the substrate with its
NH backbone. The dibasic phosphate is bound by
a rich network comprising 13 hydrogen bonds. Twelve
of them involve dipolar donor groups with six backbone
NHs (T8, L9, S32, S144, G145, T146), mostly located at
the ends of three helices; two NH side chains from
R114; and four from side chains OHs (T8, S32, S144,
T146). R114 makes two hydrogen bonds with both neg-
ative oxygen atoms of the dibasic phosphate. The pos-
itive charge of R114 is neutralized by a salt bridge with
E142. This interaction, by diminishing the charge cou-
pling interaction between the guanidium and phos-
phate, has previously been described as essential for
the fast release of the ionic substrate bound with
charged residues (Vyas et al., 2003). D61, the only dipo-
lar acceptor group, makes a hydrogen bond with the
only proton available on dibasic phosphate. This key
residue is responsible for the discrimination by five or-
ders of magnitude between HPO422and the closely re-
lated sulfate (Luecke and Quiocho, 1990). Sulfate at
physiological pH does not possess any proton and will
be repelled by the aspartic residue. This hydrogen
Figure 2. Sequence Conservation between HPBP and SBPs or DING Proteins
Alignment of the amino acid sequence of HPBP deduced from the crystal structure (residues confirmed by direct amino acid sequencing are in
bold characters) with various SBPs and DING proteins by using the program ALIGN (Myers and Miller, 1988) Potato, potato DING protein (A.B.
et al., unpublished data); L. major, L. major Friedlin almost complete DING protein sequence translated from contigs 2206 and 2072 from the L.
major Friedlin shotgun sequencing program; CAI, human CAI N-terminal and internal peptide fragments (Kumar et al., 2004); E. coli PstS (Gen-
Bank AAA24378.1); and M. tuberculosis PstS1 (GenBank M30046). Boxes indicate conserved residues; arrows indicate residues involved in
phosphate binding in E. coli and M. tuberculosis PstS proteins. The position of protruding loops is indicated. * and # indicate cysteines involved
in disulfide bridges.
A Human Phosphate Binding Apolipoprotein
bond in our structure is short (2.44 A˚) (Figure 3B), as de-
scribed in other phosphate-SBPs at ultra high 0.98 A˚
resolution. Despite this short distance, this hydrogen
bond would not be stronger than other hydrogen bonds
(Wang et al., 1997). Finally, the terminal oxygen atom is
bound with hydrogen donors. According to the struc-
ture, the binding of the monobasic phosphate could be
done similarly to other phosphate-SBPs without any
drastic rearrangement. The g-OHof S32is the only other
groupfavorably positioned toacceptthrough its oxygen
lonepair the second phosphate proton (on O3) while do-
nating its proton to the D61. The negative charge of the
phosphate will be concentrated on the two oxygen
atoms (O1 and O2) (Luecke and Quiocho, 1990). In the
case of HPBP, the surface of the binding pocket is neg-
ative, as described in other phosphate-SBP and sulfate-
SBPs (Ledvina et al., 1996; Vyas et al., 2003). It has been
proposed that this peculiar feature of an anion bound
within a negative environment would increase the dis-
crimination between negative substrates. In this way,
only anions perfectly matching the hydrogen bond net-
vorable electrostatic interaction of the anion.
The binding of dibasic phosphate has been experi-
mentally determinedby32P-orthophosphate bindingas-
says (Figures 3C and 3D). Dialyses experiments (see Ex-
phosphate at pH 8 with a better than submicromolar af-
finity, as described for other phosphate-SBP and DING
proteins (Scott and Wu, 2005). Although, monobasic
phosphate binding has been observed for bacterial
phosphate-SBPs, no phosphate binding has been ob-
served at pH 4.5 for HPBP. Small differences at the
active site between HPBP and these of bacterial
Figure 3. Binding of the Phosphate Ion
(A) Stereoview of the binding site. Residues involved in the binding of the phosphate molecule are labeled.
(B) Schematic representation of the hydrogen bond network involved in phosphate binding. Amino acids involved in phosphate binding in both
HPBP and E. coli PstS (in italics) are indicated. Numbers represent hydrogen bonds lengths in A˚.
(C) In vitro HPBP phosphate binding. PON1/HPBP were purified from six plasma samples (lanes a–f) and assayed for phosphate binding with
radiolabeled phosphate (see Experimental Procedures). Lanes g and h correspond to pure b-lactoglobulin and lysosyme, respectively, used
as negative controls.
(D) Two different PON1/HPBP-containing fractions purified from two different plasma bags (lanes a and b) separated by SDS-PAGE. Proteins
were revealed by Coomassie staining (left) and phosphate binding by autoradiography (right). PON1 and HPBP have been identified by N-ter-
minal amino acid sequencing.
phosphate-SBPs could not explain this fact. Moreover,
no binding has been observed after redialysis at pH 8.
Therefore, it is likely that that HPBP is denatured at pH
4.5, a nonphysiological environment for a plasma pro-
Association of HPBP with PON1
Initial efforts to separate PON1 from HPBP by DEAE
chromatography failed even though the two proteins
have different pIs (Figure 4). The very harsh conditions
required to separate HPBP from PON1 by using two-
dimensional PAGE (see Experimental Procedures) sug-
gest that PON1 and HPBP copurified as a heteroo-
ligomer with two subunits of almost identical molecular
weights. Indeed, using analytical ultracentrifugation,
Josseetal.(2002)showed thatPON1(consideredas ho-
mogeneous at this moment) was oligomeric. Recent
studies (data not shown) obtained by gel filtration chro-
matography indicate that the oligomeric composition of
PON1, HPBP, and PON1/HPBP complex(es) are depen-
dent on the calcium, phosphate, and detergent concen-
HPBP crystallizes as a homodimer that we assume to
be physiologically relevant because its interface, which
is mostly hydrophobic, has an area of about 790 A˚2,
a typical value for transient protein complexes (Jones
and Thornton, 1998). The conformational changes in-
duced by the composition of the crystallization liquor
could have favored the dissociation of the complex
and privileged crystallization of HPBP alone. Indeed, it
has been shown that, upon substrate binding, SBPs un-
dergo a conformational change concerning the two do-
mains that are connected by a flexible hinge (Felder
et al., 1999). This movement, described as a ‘‘Venus fly
trap motion’’ (Mao et al., 1982; Yao et al., 1996), was
also predicted for HPBP by using the program ElNemo
(Suhre and Sanejouand, 2004) (see Movie S1 available
with the Supplemental Data online).
Eukaryotic Phosphate-SBPs Are Related to DING
Surprisingly, a gene sequence corresponding to HPBP
has not been found in either the human genome or
EST databases. Nevertheless, we have been able to re-
late HPBP to a family of still poorly described proteins
that have been named DING according to the sequence
of their 4 conserved N-terminal residues (Berna et al.,
2002; A.B. et al., unpublished data). Proteins belonging
to this family are ubiquitous, since they have been iden-
tified in animals (man, monkey, rat, turkey), plants (Ara-
bidopsis thaliana, potato, tobacco), and fungi (Candida
albicans) (Riah et al., 2000; Berna et al., 2002; Belenky
et al., 2003; Blass et al., 1999; Kumar et al., 2004; Adams
et al., 2002; Weebadda et al., 2001; Scott and Wu, 2005).
DING proteins are extracellular with a mass of w40 kDa.
As in the case of HPBP, which is associated with PON1,
several of the DING proteins are often purified as homo-
or heterooligomers. Closely related proteinsalsoexist in
some Pseudomonas strains (A.B. et al., unpublished
data). In addition, hundreds of other prokayotic proteins
display a lower, but significant, homology with eukary-
otic DING proteins, although they do not start with the
DING peptide. These proteins are the SBPs involved in
phosphate uptake together with ABC transporters (see
The 8residues involved in phosphate binding are con-
served in all prokaryotic SBPs and also in all eukaryotic
DING proteins (Figure 2 and A.B. et al., unpublished
data). Figure 2 also shows that the cysteines that
make two disulfide bridges in HPBP are probably also
conserved in all eukaryotic DING proteins as well as in
most prokaryotic proteins related to them.
HPBP displays about 64% sequence identity with
DING proteins from potato and Leishmania major,
whereas HPBP sequence identities with E. coli and M.
tuberculosis phosphate-SBPs display only 25% and
23%, respectively (Figure 2).
HPBP Biochemical Properties
HPBP was found to be tightly associated with PON1.
Besides, subsequently to our description of a copurified
protein in supposedly pure PON1 preparations (Fokine
et al., 2003), recent studies indicate that the presence
of additional proteins in purified PON1 fractions is not
uncommon and could lead to confusion in the interpre-
tation of PON1 function (Teiber et al., 2004; Connelly
et al., 2005; Draganov et al., 2005). In at least one case,
a protein copurified with PON1 had a molecular weight
and a pI corresponding to HPBP (Smolen et al., 1991).
PON1 is an apolipoprotein largely located on HDL but
also present on chylomicrons and VLDL (Fuhrman
et al., 2005; Deakin et al., 2005). Thus, we assume that
the previously unknown hydrophobic HPBP is an apoli-
poprotein associated in vivo with PON1.
Moreover, experiments assessing the characteriza-
tion of the oligomeric organization of the PON1/HPBP
as well as its impact on the functional activity of the
two proteins, which both appear as dependent to cal-
cium, phosphate, and detergent (in vitro) or lipoproteins
(in vivo) amounts, are in progress (M. Elias et al., per-
Figure 4. Two-Dimensional PAGE of the Purified PON1 Fraction
Identification by two-dimensional PAGE of proteins in a purified
PON1 fraction obtained according to Gan et al. (1991). The first di-
mension corresponds to isoelectric focusing in a 3–10 pH gradient;
the second dimension is 10% SDS-PAGE. PON1 and HPBP, identi-
fied by N-terminal sequencing, have similar masses (w40 kDa), but
very different pIs (w5 and w8, respectively). The right lane shows
molecular weight markers.
A Human Phosphate Binding Apolipoprotein
HPBP structure clearly relates it to bacterial phos-
phate-SBPs associated with ABC transporters. We
have shown that HPBP is capable of binding inorganic
phosphate ions at physiological conditions with submi-
cromolar affinity. The normal concentration of free inor-
ganic phosphate in human plasma is about 0.5 mM; ow-
ing to its high specificity, HPBP is always saturated in
phosphate.For now,wecanonlyspeculate on thephys-
iological function of this first, to our knowledge, phos-
phate binding protein reported so far in human plasma
Eukaryotic Phosphate-SBPs Are Missing from
knowledge, member of the DING family of proteins.
Even though DING proteins are likely ubiquitous in eu-
karyotes, they remain poorly characterized since very
few coding sequences are present in DNA or RNA data-
bases and since none of them is complete (A.B. et al.,
unpublished data). Thus, HPBP may provide the first
complete sequence of a DING protein. To our knowl-
edge, recent cloning of a large part of the potato DING
gene and discovery of two partial DNA sequences of
DING proteins in unannotated parts of the Leischmania
major and Rattus norvegicus genomes allowed for ex-
tensive comparison between different DING genes for
the first time (A.B. et al., unpublished data). This re-
vealed an astonishing conservation among several of
the DING proteins and genes, with pairwise identities
above 95% and 90% at the amino acid and nucleotide
The very high conservation of some phosphate-SBP
gene sequences between distant species such as po-
tato (a higher plant) and Leishmania major (a protozoan)
cannot be explained by a constraint based only on
a physiological function requiring the integrity of the
protein. These observations should stimulate studies
to determine unknown features that could prevent iden-
tification of some specific genes in systematic sequenc-
In humans, DING proteins have previously been iso-
vial stimulatory protein that could be involved in the de-
velopment of rheumatoid arthritis (Hain et al., 1996;
Blass et al., 1999). Subsequently, a related protein se-
creted from fibroblasts and cervical carcinoma cells
was purified by virtue of its affinity for hirudin (Adams
et al., 2002). A protein displaying high affinity for genis-
tein, an estrogen analog, was identified in breast cells
(Belenky et al., 2003). Finally, a protein secreted by kid-
ney epithelial cells was named crystal adhesion inhibitor
(CAI) (Kumar et al., 2004). With a total of 261 amino acids
obtained by N-terminal and internal peptide sequenc-
ing, CAI was, up until now, the longest DING protein se-
quence known. At 59% amino acid identity with CAI,
HPBP clearly belongs to a different DING protein sub-
family. Thus, at least two DING genes are missing from
the sequenced human genome.
By possibly providing the first structure and the first
lated DING proteins and bacterial phosphate-SBPs.
This conclusively demonstrates that this last class of
proteins is likely ubiquitous in living organisms, though
it has never been characterized or predicted from ge-
nome analyses in eukaryotes. In addition, since the 8
residues that constitute the phosphate binding site in
prokaryotic SBPs are conserved in DING proteins, it is
reasonable to assume that functional phosphate-SBPs
are present in all eukaryotes. However, their association
with ABC transporters remains to be characterized.
Physiological Function of Human DING Protein
The presence of an inorganic phosphate scavenger or
receptor in human plasma associated with HDL may
be explained by the need to prevent the interaction of
this anion with Ca2+. Indeed, hyperphosphatemia is
a cardiovascular disease risk factor (Amann et al.,
2003) because the insoluble calcium phosphate, along
with cholesterol, increases the formation of atheroscle-
rotic plaques that block blood vessels (Dorozhkin and
Epple, 2002). Interestingly, the related CAI protein has
been described to prevent the formation of kidney
stone, by inhibiting calcium phosphate precipitation
(Kumar et al., 2004). In plasma, HPBP should be always
saturated with phosphate; therefore, we could suppose
that HPBP could sense or transport phosphate in a low-
phosphate toward a specific acceptor, or, similarly to
the CAI, prevent the formation of calcium phosphate
Alloftheseconcomitant factssuggestthatdisorder of
HPBP might favor hyperphostemia and/or atherosclero-
sis plaque development. Thus, HPBP, serendipitously
discovered as tightly bound to the HDL lipoprotein
PON1, may contribute, with PON1, to the antiathero-
genicity of a category of HDL particles. Accordingly,
HPBP could be tentatively regarded as a new predictor
disorders, including atherosclerosis.
Crystallization and Heavy Atom Derivative Soaks
The HPBP/PON1-containing fractions were obtained as described
in the purification protocol (see the Supplemental Data). Fractions
were dialyzed and concentrated in the presence of C-12 maltoside
(0.64 mM) by using a centrifugation technique (Centriprep Amicon
cutoff: 10 kDa, Millipore, St. Quentin-en-Yvelines, France) to a final
O.D. of 2.3. Crystallization was performed by using the hanging
drop vapor diffusion method. A total of 3 ml protein solution was
mixed with 2 ml of a reservoir solution containing cacodylate buffer
(50 mM, pH 6), ammonium sulfate (2 M), and sodium chloride (1
mM). Smallcrystalsappeared after1weekat4ºC.Heavy atomderiv-
atives were obtained by soaking crystals during 3 days in a reservoir
solution containing a uranium salt at 0.1 mM (Uranium-Plasmocor-
inth B [Chesne, 2002], a generous gift from Olivier Bertrand).
Data Collection and Structure Determination
X-ray diffraction data were collected at 100 K by using synchrotron
radiation at the FIP BM30 beamline (ESRF, Grenoble, France) with
a MarCCD (165 mm) detector. A single data set was recorded at
a wavelength of 0.934 A˚from a native crystal, and two other data
sets were recorded at the uranium LIII edge from one derivatized
crystal (Table 1). X-ray diffraction data were integrated, scaled,
and merged with the XDS2000 program (Kabsch, 1993) and the
CCP4 program suite (CCP4, 1994). CNS (Brunger et al., 1998) and
SnB programs (Weeks and Miller, 1999) were used to find the posi-
tions of the three uranium atoms. Initial experimental phases were
obtained with the program SHARP (De la Fortelle and Bricogne,
1997) by using the single isomorphous replacement with anomalous
scattering (SIRAS) method with a figure of merit of 0.46. A solvent
flattening procedure was applied to the electron density map by us-
ing SOLOMON (Abrahams and Leslie, 1996) and DM (Cowtan, 1994).
The polypeptide chain was partially built in an automatic mode by
using ARP/wARP (Perrakis et al., 2001). CNS was employed for the
structure refinement, and TURBO-FRODO (Roussel et al., 1990)
was used for manual model building. Statistics for refinement are
given in Table 1.
The complete amino acid sequence was first determined from X-ray
data. This was helpful for the alignment of sequences subsequently
obtained by N-terminal sequencing, mass spectrometry analysis,
and internal peptide digestion (Supplemental Data). About 75% of
the sequence has been confirmed, showing an error rate of only
11% in the crystallographically predicted sequence. So, the HPBP
sequence provided herein is complete and exact at 97%.
In Vitro Phosphate Binding Assays
For the dot-blot binding assays, HPBP/PON1-containing fractions
from different plasma samples and negative control proteins (lyso-
membrane. After incubation in 50 mM Tris (pH 8), 2 mM32P-ortho-
phosphate (10 mCi/ml) (Amersham) for 150 min, the membrane
wasrinsedtwice inthesamebufferwithout phosphate andwassub-
film. For the assays in gels, separation of HPBP and PON1 was ac-
complished by using SDS-PAGE under nonreducing conditions
(8-25 Phast gel Pharmacia) after preincubation of the HPBP/PON1-
containing fractions with 0.1% SDS without heating. The gel was
run and then rinsed four times for 15 min in 0.1% Triton X-100-con-
taining buffer to remove SDS, incubated with32P-orthophosphate,
and revealed as mentioned above or stained with Coomassie blue.
An equilibrium dialysis method was employed for studying bind-
ing of Pi to HPBP. Dialysis tubings with membrane with a molecular
weight cutoff of 8 kDa were filled with 400 ml of a solution containing
1 mM HPBP/PON1, 0.1% Triton (v/v), 150 mM NaCl, in 25 mM Tris-
HCl buffer (pH 8.0), or 25 mM acetate buffer (pH 4.5). Tubings were
immersed into vials containing the same solutions without protein,
supplemented with 0.01 mM, 0.1 mM, 1 mM, or 10 mM orthophosphate
labeled with32P-orthophosphate (10 mCi ml21) (Amersham). Dialy-
sis were performed during 1 week to allow achievement of equilib-
rium. Phosphate concentration in vials and dialysis tubings was cal-
culated from scintillation measurement.
Two-Dimensional Gel Electrophoresis
HPBP/PON1-containing fractions were prepared in 9.8 M urea, 4%
(v/v) Triton X-100, 2 mM tributyl phosphine, 0.2% of ampholytes
3–10 (Biolytes, Bio-Rad, Marnes-la-Coquette, France), and 0.001%
(m/v) bromophenol blue. A total of 40 mg protein was separated in
the first dimension by focusing in IPG-Strips (Bio-Rad) on a Protean
IEF cell (Bio-Rad) at 250 V for 15 min, voltage ramping to 4000 V for
2 hr, followed by focusing at 4000 V for 4 hr. Separated proteins of
the first dimension were resolved in the second dimension by stan-
dard SDS-PAGE. Gels were silver stained.
Supplemental Data including the protein purification protocol, ta-
bles of data collection and of refinement statistics, sequencing,
mass spectrometry analysis, and a movie showing the Venus fly
trap motion are available at http://www.structure.org/cgi/content/
This research was supported by grants to P.M., J.C.F.-C., and E.C.
by De ´le ´gation Ge ´ne ´rale pour l’Armement (CO nº010807/03-10); by
a grant to D.R. by Sanita ¨tsakademie der Bundeswehr, and by the
Centre National de la Recherche Scientifique.
Received: July 11, 2005
Revised: November 9, 2005
Accepted: December 20, 2005
Published online: March 14, 2006
Abrahams, J.P., and Leslie, A.G.W. (1996). Methods used in the
structure determination of bovine mitochondrial F1ATPase. Acta
Crystallogr. D. Biol. Crystallogr. 52, 30–42.
Adams, L., Davey, S., and Scott, K. (2002). The DING protein: an au-
tocrine growth-stimulatory protein related to the human synovial
stimulatory protein. Biochim. Biophys. Acta 1586, 254–264.
Amann, K., Tornig, J., Kugel, B., Gross, M.L., Tyralla, K., El-Shak-
mak, A., Szabo, A., and Ritz, E. (2003). Hyperphosphatemia aggra-
vates cardiac fibrosis and microvascular disease in experimental
uremia. Kidney Int. 63, 1296–1301.
Belenky, M., Prasain, J., Kim, H., and Barnes, S. (2003). DING, a ge-
nistein target in human breast cancer: a protein without a gene.
J. Nutr. 133, 2497–2501.
Berna, A., Bernier, F., Scott, K., and Stuhlmuller, B. (2002). Ring up
the curtain on DING proteins. FEBS Lett. 524, 6–10.
Blass, S., Engel, J.M., and Burmester, G.R. (1999). The immunologic
homunculus in rheumatoid arthritis. Arthritis Rheum. 42, 2499–2506.
Brooks-Wilson, A., Marcil, M., Clee, S.M., Zhang, L.H., Roomp, K.,
van Dam, M., Yu, L., Brewer, C., Collins, J.A., Molhuizen, H.O.,
et al. (1999). Mutations in ABC1 in Tangier disease and familial
high-density lipoprotein deficiency. Nat. Genet. 22, 336–345.
Brunger, A.T., Adams, P.D., Clore, G.M., DeLano, W.L., Gros, P.,
Grosse-Kunstleve, R.W., Jiang, J.S., Kaszewski, J., Nilges, M.,
Pannu, N.S., et al. (1998). Crystallography & NMR system: a new
software suite for macromolecular structure determination. Acta
Crystallogr. D Biol. Crystallogr. 54, 905–921.
CCP4 (Collaborative Computational Project, Number 4) (1994). The
CCP4 suite: programs for protein crystallography. Acta Crystallogr.
D Biol. Crystallogr. 50, 760–763.
Table 1. Data Collection
Native U Peak U Remote
Unit cell (A˚)
Number of observed
Number of unique
(last bin) (%)
Rsym(last bin) (%)
I/s(I) (last bin)
Last resolution shell
a = 97.4, b = 87.9, c = 90.5
99.4 (91.8)98.8 (97.1) 95.1 (93.5)
Resolution range (A˚)
Number of reflections
used in refinement
Number of protein atoms
Number of water
Average B factor (A˚2)
Rwork= SkFoj 2 jFck/ = SjFoj, where Fodenotes the observed struc-
ture factor amplitude and Fcdenotes the structure factor amplitude
calculated from the model. Rfreeis as for Rwork, but it is calculated
with 5% of the randomly chosen reflections omitted from the refine-
A Human Phosphate Binding Apolipoprotein
Chang, G., and Roth, C.B. (2001). Structure of MsbA from E. coli:
a homolog of the multidrug resistance ATP binding cassette (ABC)
transporters. Science 293, 1793–1800.
Chesne, M.L. (2002). Utilisation de la diffusion anomale de l’uranium
pour de ´terminer la structure de macromolecules biologiques. The-
sis, University Joseph Fourier, Grenoble I, France.
Connelly, P.W., Draganov, D., and Maguire, G.F. (2005). Paraoxo-
oxynitrite. Free Radic. Biol. Med. 38, 164–174.
Contreras-Martel, C., Carpentier, P., Morales, R., Renault, F.,
Chesne-Seck, M.L., Rochu, D., Masson, P., Fontecilla-Camps,
J.C., and Chabriere, E. (2006). Crystallization and preliminary X-ray
diffraction analysis of the human phosphate binding protein. Acta
Crystallogr. F 62, 67–69.
Cowtan, K. (1994). Joint CCP4 and ESF-EACBM. Newslett. Protein
Crystallogr. 31, 34–38.
Cserzo, M., Wallin, E., Simon, I., von Heijne, G., and Elofsson, A.
(1997). Prediction of transmembrane a-helices in prokaryotic mem-
brane proteins: the dense alignment surface method. Protein Eng.
De la Fortelle, E., and Bricogne, G. (1997). Maximum-likelihood
heavy-atom paramater refinement for multiple isomorphous re-
placement and multiwavelength anomalous diffraction methods.
Methods Enzymol. 276, 472–494.
Deakin, S., Moren, X., and James, R.W. (2005). Very low density lipo-
proteins provide a vector for secretion of paraoxonase-1 from cells.
Atherosclerosis 179, 17–25.
Dorozhkin, S.V., and Epple, M. (2002). Biological and medical signif-
icance of calcium phosphates. Angew. Chem. Int. Ed. Engl. 41,
Draganov, D.I., Teiber, J.F., Speelman, A., Osawa, Y., Sunahara, R.,
and La Du, B.N. (2005). Human paraoxonases (PON1, PON2, and
PON3) are lactonases with overlapping and distinct substrate spec-
ificities. J. Lipid Res. 46, 1239–1247.
Felder, C.B., Graul, R.C., Lee, A.Y., Merkle, H.P., and Sadee, W.
(1999). The Venus flytrap of periplasmic binding proteins: an ancient
protein module present in multiple drug receptors. AAPS PharmSci
Fokine, A., Morales, R., Contreras-Martel, C., Carpentier, P., Re-
nault, F., Rochu, D., and Chabriere, E. (2003). Direct phasing at low
resolution of a protein co-purified with human paraoxonase
(PON1). Acta Crystallogr. D Biol. Crystallogr. 59, 2083–2087.
Fuhrman, B., Volkova, N., and Aviram, M. (2005). Paraoxonase 1
(PON1) is present in postprandial chylomicrons. Atherosclerosis
Gan, K.N., Smolen, A., Eckerson, H.W., and La Du, B.N. (1991). Puri-
fication of human serum paraoxonase/arylesterase. Evidence for
one esterase catalyzing both activities. Drug Metab. Dispos. 19,
Gerdes, R.G., and Rosenberg, H. (1974). The relationship between
the phosphate-binding protein and a regulator gene product from
Escherichia coli. Biochim. Biophys. Acta 351, 77–86.
Gordon, D.J., and Rifkind, B.M. (1989). High-density lipoprotein: the
clinical implications of recent studies. N. Engl. J. Med. 321, 1311–
Hain, N.A., Stuhlmuller, B., Hahn, G.R., Kalden, J.R., Deutzmann, R.,
and Burmester, G.R. (1996). Biochemical characterization and mi-
crosequencing of a 205-kDa synovial protein stimulatory for T cells
and reactive with rheumatoid factor containing sera. J. Immunol.
Higgins, C.F. (1992). ABC transporters: from microorganisms to
man. Annu. Rev. Cell Biol. 8, 67–113.
Holm, L., and Sander, C. (1994). Searching protein structure data-
bases has come of age. Proteins 19, 165–173.
Jones, P.M., and George, A.M. (2002). Mechanism of ABC trans-
porters: a molecular dynamics simulation of a well characterized nu-
cleotide-binding subunit. Proc. Natl. Acad. Sci. USA 99, 12639–
Baud, D., Renault, F., Le Maire, M., Chabrieres, E., et al. (2002). Olig-
omeric states of the detergent-solubilized human serum paraoxo-
nase (PON1). J. Biol. Chem. 277, 33386–33397.
Kabsch, W. (1993). Automatic processing of rotation diffraction data
from crystals of initially unknown symmetry and cell constants.
J. Appl. Crystallogr. 26, 795–800.
Khersonsky, O., and Tawfik, D.S. (2005). Structure-reactivity studies
of serum paraoxonase PON1 suggest that its native activity is lacto-
nase. Biochemistry 44, 6371–6382.
Kumar, V., Yu, S., Toback, F.G., and Lieske, J.C. (2004). Renal epi-
thelial cells constitutively produce a protein that blocks adhesion
of crystals to their surface. Am. J. Physiol. 287, 373–383.
Ledvina, P.S., Yao, N., Choudhary, A., and Quiocho, F.A. (1996).
Negative electrostatic surface potential of protein sites specific for
anionic ligands. Proc. Natl. Acad. Sci. USA 13, 6786–6791.
Luecke, H.,and Quiocho, F.A.(1990).High specificity ofaphosphate
transport protein determined by hydrogen bonds. Nature 347, 402–
Mao, B., Pear, M.R., McCammon, J.A., and Quiocho, F.A. (1982).
Hinge-bending in L-arabinose-binding protein. The ‘‘Venus’s-fly-
trap’’ model. J. Biol. Chem. 257, 1131–1133.
Medveczky, N., and Rosenberg, H. (1971). Phosphate transport in
Escherichia coli. Biochim. Biophys. Acta 241, 494–506.
Myers, E.W., and Miller, W. (1988). Optimal alignments in linear
space. Comput. Appl. Biosci. 4, 11–17.
Perrakis, A., Harkiolaki, M., Wilson, K.S., and Lamzin, V.S. (2001).
tallogr. 57, 1445–1450.
Riah, O., Dousset, J.C., Bofill-Cardona, E., and Courriere, P. (2000).
Neurobiol. 20, 653–664.
Roussel, A., Fontecilla-Camps, J.C., and Cambillau, C. (1990).
TURBO-FRODO: a new program for protein crystallography and
modeling. XVthIUCr Congress, Bordeaux, France.
Scott, K., and Wu, L. (2005). Functional properties of a recombinant
bacterial DING protein: comparison with a homologous human pro-
tein. Biochim. Biophys. Acta 1744, 234–244.
L.W., Furlong, C.E., Costa, L.G., Fogelman, A.M., et al. (1998). Mice
lacking serum paraoxonase are susceptible to organophosphate
toxicity and atherosclerosis. Nature 394, 284–287.
Smolen, A., Eckerson, H.W., Gan, K.N., Hailat, N., and La Du, B.N.
(1991). Caracteristics of the genetically determined allozymic forms
Suhre, K., and Sanejouand, Y.H. (2004). ElNemo: a normal mode
web-server for protein movement analysis and the generation of
templates for molecular replacement. Nucleic Acids Res. 32, 610–
Surin, B.P., Jans, D.A., Fimmel, A.L., Shaw, D.C., Cox, G.B., and
Rosenberg, H. (1984). Structural gene for the phosphate-repressible
phosphate-binding protein of Escherichia coli has its own promoter:
complete nucleotide sequence of the phoS gene. J. Bacteriol. 157,
ary relationships among extracellular solute-binding receptors of
bacteria. Microbiol. Rev. 57, 320–346.
Teiber, J.F., Draganov, D.I., and La Du, B.N. (2004). Purified human
serum PON1 does not protect LDL against oxidation in the in vitro
assays initiated with copper or AAPH. J Lipid Res 45, 2260–2268.
Vyas, N.K., Vyas, M.N., and Quiocho, F.A. (2003). Crystal structure of
M. tuberculosis ABC phosphate transport receptor: specificity and
charge compensation dominated by ion-dipole interactions. Struc-
ture 11, 765–774.
Watson, A.D., Berliner, J.A., Hama, S.Y., and La Du, B.N. (1995). Pro-
tective effect of high density lipoprotein associated paraoxonase.
Inhibition of the biological activity of minimally oxidized low density
lipoprotein. J. Clin. Invest. 96, 2882–2891.
Weebadda, W.K., Hoover, G.J., Hunter, D.B., and Hayes, M.A. Download full-text
(2001). Avian air sac and plasma proteins that bind surface polysac-
charides of Escherichia coli O2. Comp. Biochem. Physiol. 130, 299–
Weeks, C.M., and Miller, R. (1999). The design and implication of
SnB v2.0. J. Appl. Crystallogr. 32, 120–124.
Yao, N., Ledvina, P.S., Choudhary, A., and Quiocho, F.A. (1996).
Modulation of a salt link does not affect binding of phosphate to
its specific active transport receptor. Biochemistry 35, 2079–2085.
Young, D.B., and Garbe, T.R. (1991). Lipoprotein antigens of Myco-
bacterium tuberculosis. Res. Microbiol. 142, 55–65.
short hydrogen bond in very high resolution structures of protein
receptor–phosphate complexes. Nat. Struct. Biol. 7, 519–522.
The atomic coordinates and structure factors of the HPBP structure
have been deposited with the Protein Data Bank with the accession
A Human Phosphate Binding Apolipoprotein