doi:10.1016/j.jmb.2009.01.008
J. Mol. Biol. (2009) 386, 1240–1254
Available online at www.sciencedirect.com
Crystal Structures of Limulus SAP-Like Pentraxin
Reveal Two Molecular Aggregations
Annette K. Shrive 1 , Ian Burns 1 , Hui-Ting Chou 2 , Henning Stahlberg 2 ,
Peter B. Armstrong 2 and Trevor J. Greenhough 1 ⁎
1
Institute of Science and
Technology in Medicine,
School of Life Sciences,
Keele University,
Staffordshire ST5 5BG, UK
2
Molecular and Cellular
Biology, College of Biological
Sciences, University of
California at Davis, Briggs Hall,
1 Shields Avenue, Davis,
CA 95616, USA
Received 31 July 2008;
received in revised form
19 December 2008;
accepted 7 January 2009
Available online
18 January 2009
The serum-amyloid-P-component-like pentraxin from Limulus polyphemus,
a recently discovered pentraxin species and important effector protein of the
hemolymph immune system, displays two distinct doubly stacked cyclic
molecular aggregations, heptameric and octameric. The refined threedimensional structures determined by X-ray crystallography, both based on
the same cDNA sequence, show that each aggregate is constructed from a
similar dimer of protomers, which is repeated to make up the ring structure.
The native octameric form has been refined at a resolution of 3 Å, the native
heptameric form at 2.3 Å, and the phosphoethanolamine (PE)-bound
octameric form at 2.7 Å. The existence of the hitherto undescribed
heptameric form was confirmed by single-particle analysis using cryoelectron microscopy. In the native structures, the calcium-binding site is
similar to that in human pentraxins, with two calcium ions bound in each
subunit. Upon binding PE, however, each subunit binds a third calcium ion,
with all three calcium ions contributing to the binding and orientation of the
bound phosphate group within the ligand-binding pocket. While the
phosphate is well-defined in the electron density, the ethanolamine group is
poorly defined, suggesting structural and binding variabilities of this group.
Although sequence homology with human serum amyloid P component is
relatively low, structural homology is high, with very similar overall folds
and a common affinity for PE. This is due, in part, to a “topological”
equivalence of side-chain position. Identical side chains that are important
in both function and fold, from different regions of the sequence in human
and Limulus structures, occupy similar space within the overall subunit fold.
Sequence and structure alignment, based on the refined three-dimensional
structures presented here and the known horseshoe crab pentraxin
sequences, suggest that adaptation and refinement of C-reactive-proteinmediated immune responses in these ancient creatures lacking antibodybased immunity are based on adaptation by gene duplication.
© 2009 Elsevier Ltd. All rights reserved.
Edited by R. Huber
Keywords: Limulus; pentraxin; octamer; heptamer; X-ray crystal structures
Introduction
*Corresponding author. E-mail address:
[email protected].
Abbreviations used: PE, phosphoethanolamine; CRP,
C-reactive protein; SAP, serum amyloid P component; 3D,
three-dimensional; LPx, Limulus pentraxins; cryo-EM,
cryo-electron microscopy; FSC, Fourier shell correlation;
PC, phosphocholine; PEG, polyethylene glycol; MPD,
methyl-2,4-pentanediol.
The classical serum pentraxins C-reactive protein
(CRP) and serum amyloid P component (SAP)
are members of a conserved family of cyclic oligomeric Ca-binding proteins.1 In species as diverse
as humans and horseshoe crab, which has been
shown to have both CRP-like and SAP-like pentraxins, the serum pentraxins play an important role in
innate host defense.2–7 The combination of adaptive
immunoglobulin-based antibody immunity and single multifunctional CRP and SAP in humans contrasts sharply with invertebrates, where, in the
0022-2836/$ - see front matter © 2009 Elsevier Ltd. All rights reserved.
Multiple Aggregates for Limulus SAP-Like Pentraxin
absence of adaptive immunity, humoral components
such as the pentraxins provide an essential and effective strategy for recognizing and destroying diseasecausing pathogens.5,8,9 CRP has recently been shown
to be a predominant lipopolysaccharide-binding
acute-phase protein in the horseshoe crab in response
to Pseudomonas aeruginosa infection.10 Recently, a new
serum pentraxin named CrOctin, which is produced
in response to infection and targets bacterial phosphoethanolamine (PE), has been identified in the
horseshoe crab Carcinoscorpius rotundicauda.11 Phylogenetic analysis indicates that it is related to Limulus
SAP (the SAP-like pentraxin from Limulus hemolymph), but is a pentraxin distinct from SAP and
CRP.
Levels of circulating CRP are naturally high (∼ 1–
7 mg/ml) in the phylogenetically ancient horseshoe
crab Limulus polyphemus,12,13 where pentraxins are
the second most abundant proteins in hemolymph
after hemocyanin.7 SAP pentraxin levels corresponding to 8–19% of the total pentraxin in Limulus
hemolymph have been observed.14 Whereas numerous CRP proteins with differing ligand-binding
properties6,15–17 have been identified in horseshoe
crabs, to date, only one SAP-like protein has been
found (in L. polyphemus).18 No convincing evidence
has thus far been presented of multiple CRP or SAP
genes in mammals, although multiple sequences
have been suggested for the CRP found in the dogfish Mustelus canis.19,20 To our knowledge, multiple
genes or sequences for any SAP-like protein have so
far not been reported for any organism.
The three-dimensional (3D) polyalanine native
structure of a doubly stacked octameric molecular
form of Limulus SAP-like pentraxin, showing that
the structure of the pentraxin protomer is remarkably conserved through evolution, has previously
been reported.14,21,22 The sequence for Limulus SAP
has since been determined at the genetic level,18 thus
allowing full refinement and analysis of the 3D
structure. We report here the refined 3D structures
of the octameric Limulus SAP, with and without
bound ligand, along with the refined structure of a
novel previously unknown molecular form that
exhibits a doubly stacked heptameric aggregation.
We have also investigated the distribution of these
two molecular aggregations by cryo-negative staining electron microscopy.
Results
Gel electrophoresis
SDS-PAGE of the purified SAP protein reveals three
closely spaced but distinct bands that are thought to
be due to variable glycosylation.18 Analysis of the
purified SAP protein on native gels confirms the
presence of two major distinct molecular aggregates,
with bands at approximately 359 kDa and 395 kDa,
and with a less intense band corresponding to a highermolecular-mass aggregate at approximately 772 kDa
1241
(Fig. 1). Based on a monomeric molecular mass of
25 kDa (the protein molecular mass is 23.8 kDa,
excluding glycosylation18), the two major bands correspond to 14.4 and 15.8 subunits, respectively.
Overall structures and molecular aggregations
The previously reported 3D polyalanine structure
of a Limulus pentraxin, derived from crystals that
grew spontaneously from a pentraxin mix containing both CRPs and SAP, revealed a previously
undiscovered SAP-like pentraxin in the molecular
form of a doubly stacked octamer.14 New crystallization conditions have now been found using SAP
protein isolated and purified following the methods
described by Tharia et al.18 One of these conditions
produces diffraction-quality crystals not only of the
original octameric structure, as verified by structural
analysis of both native and PE-bound crystals, but
also of a new crystal form. Both crystal forms were
obtained from the same purified SAP sample, often
appearing in the same crystallization well. The
structure solution of this new crystal form reveals
a different molecular aggregation consisting of a
doubly stacked heptamer (Fig. 2). At no stage during
the refinement could significant differences between
the amino acids in the two molecular forms be
unambiguously identified from the electron density.
While this may be due to the resolution of the data,
only a single Limulus SAP cDNA sequence has so far
been found,18 and this was used throughout. The
native octameric form has been refined at a
resolution of 3 Å, the native heptameric form at
2.3 Å (see Fig. 3a), and the PE-bound octameric form
at 2.7 Å.
Fig. 1. Native PAGE (5–12.5%) of purified Limulus SAPlike pentraxin. The purified Limulus SAP was treated as
described under Materials and Methods, and the gel was
stained with Coomassie brilliant blue R-250. Lane 1, High
Molecular Weight Calibration Kit (porcine thyroglobulin
(Mr ∼ 669,000); equine ferritin (Mr ∼ 440,000); bovine
catalase (Mr ~232,000); bovine lactate dehydrogenase
(Mr ∼140,000); bovine serum albumin (Mr ∼66,000));
lanes 2 and 3: 5 μg and 2.5 μg of Limulus SAP pentraxin
purified from serum using 50 mM PE elution from a PEaffinity column subsequent to an initial 10 mM PC elution
to remove CRP proteins; lane 4, L-lactate dehydrogenase
from bovine heart (Mr ∼ 140,000); lane 5: catalase from
bovine liver (Mr ∼ 240,000); lane 6: ferritin from equine
spleen (Mr ∼ 440,000); lane 7: molecular mass standards as
in lane 1.
1242
Multiple Aggregates for Limulus SAP-Like Pentraxin
Fig. 2. The doubly stacked octameric and heptameric structures of Limulus SAP. Ribbon diagrams of Limulus SAP
viewed down and perpendicular to the molecular cyclic symmetry axis. (a) The doubly stacked octameric structure. (b)
The doubly stacked heptameric structure. This figure was generated using MOLSCRIPT.23 Calcium ions are shown in
green, and the hydrogen-bonded dimer is highlighted in orange.
The 3D X-ray structures of Limulus SAP pentraxin
reveal either a 16-mer (hexadecamer) or a 14-mer of
protomers arranged in doubly stacked cyclic rings
(Fig. 2). Within both molecular forms, pairs of protomers (one from each ring) form a common dimeric
unit, with the protomers linked together through an
extended β-sheet structure (Fig. 2). In the octameric
structure, the two protomers are crystallographically
related and thus identical, while in the heptameric
structure, the two are independent but essentially
identical. This dimer formation is a conserved feature
of both heptameric and octameric structures and
leads to the doubly stacked rings observed. The protomers themselves (Fig. 3b) exhibit the highly conserved pentraxin fold.14 In both molecular forms, the
calcium-binding site (and ligand-binding site) on
each protomer is located on the perimeter of the ring,
in contrast to the human pentraxins where these
sites are on one face of the pentameric ring, with the
effector face located on the opposite face.24 The electron density in all the structures indicates glycosylation at Asn81, but not at the other potential site
Asn118.18 Sugar density is poorly defined, and no
sugar has been built into the model.
Interactions at the subunit–subunit interface
within the cyclic rings are clustered towards the
central point of contact and are limited towards the
outer and inner surfaces of the cyclic molecule (Fig.
4a). The difference between the two cyclic arrangements appears to be a rotation of some 6° pivoted
about this central contact point such that in the
octameric arrangement, the distance between the
inner contact surfaces is increased, and the distance
between the outer contact surfaces is reduced
(Fig. 4b). In terms of specific interactions across
the interface, there are examples of hydrogen
bonds, hydrophobic interfaces, and possible salt
bridges (see Fig. 4). The average protomer–protomer contact interface area within the cyclic
arrangement is 598 Å2 in the heptamer and 568 Å2
in the octamer. The total contact interface area for
each protomer, excluding the dimer interface,
averages 910 Å2 in both the heptamer and the
octamer.25
The inner surface contacts are restricted to only
Gln177 NE2 (subunit 1)–His 213 O (subunit 2) and,
in the heptamers only, a relatively weak Trp169–
C-terminus (Leu217) interaction. The Leu217 side
chain appears to adopt a different rotamer in the two
molecular forms, but is inserted similarly in both
cases into a hydrophobic pocket in the neighboring
subunit. The Gln177-His213 and Trp169 NE1–
Multiple Aggregates for Limulus SAP-Like Pentraxin
1243
Fig. 3. (a) Stereo view of typical 2.3-Å resolution electron density for the heptamer. The electron density map (2Fo − Fc) is
calculated from the observed data Fo and the final model Fc. The map, calculated at 2.3 Å resolution, is
noncrystallographically averaged, solvent-flattened, and contoured at 1 rmsd. (b) The Limulus SAP protomer and the
pentraxin fold. Structure of the Limulus SAP protomer showing the labeling of the secondary structure elements. The figure
was generated from the native heptameric structure using MOLSCRIPT.23 Calcium ions are shown in green.
1244
Multiple Aggregates for Limulus SAP-Like Pentraxin
Fig. 4. (a) Subunit–subunit contacts within the cyclic aggregates.
The view is approximately parallel
with the cyclic rotational axis (perpendicular to the plane of the ring),
with the outer edge of the aggregate
at the top of the picture. The subunits shown are A and M in the
cyclic heptamer, but the interactions
shown are the sum of those that are
present throughout the three structures. This figure was generated
using MOLSCRIPT.23 (b) Overlay
of the octameric and heptameric
subunit–subunit interfaces. The
view is the same as that in (a). The
heptameric subunits A and M are in
cyan and orange, respectively, and
the octameric subunits (A and B,
respectively) are in gray following a
least squares fit of the main chains of
heptameric subunit A and octameric
subunit A. This figure was generated using MOLSCRIPT.23
C-terminus distances average 3.0 Å and 3.8 Å, respectively, in the octamer, and 2.8 Å and 3.0 Å, respectively, in the heptamer. The only example of a
possible specific contact towards the outer edge of
the cyclic rings is between Arg32 NH1 (subunit 1)
and Pro12 O (subunit 2), but this is highly variable
1245
Multiple Aggregates for Limulus SAP-Like Pentraxin
Fig. 5. PE in the ligand-binding site of the octameric
molecular form. Arrayed around the phosphate and the
three calcium ions (in green) are the calcium-coordinating
residues (see the text), while at the other end of the
binding pocket are the three key residues Tyr49, Asn85,
and Gln79. The four β-strands that form the shallow
binding pocket are labeled D, E, F, and G. This figure was
generated using MOLSCRIPT.23
due to somewhat poorly defined Arg32 side chains
(Fig. 4).
Around the central point of contact are clustered a
number of interactions involving Gln33, Arg112,
and Glu171 in subunit 1, and a variety of residues in
subunit 2 (see Fig. 4), although the distances suggest
that the interactions are relatively weak. Except for
Glu95 in subunit 2, which is often poorly defined
and whose conformation seems to vary, the residues
in this central contact region are all well-defined (for
the resolution of the data) in the electron density,
and there is no consistent pattern to the nonbonded
contact distance variations within or between the
structures. In this region, only Glu95 and Arg112
showed any significant difference in side-chain
conformation between the two molecular forms
(see Fig. 4b), with the side chain of His213 rotated
slightly between the two arrangements. The Arg112Glu95 interaction is poorly defined due to the illdefined nature of Glu95. The remaining contact
residues in subunit 1, Gln33 and Glu171, interact
with various residues in subunit 2. The strongest of
these interactions involve Glu171 OE1 and OE2,
with an H-bond averaging 2.7 Å and 2.9 Å in the
octamer and the heptamer, respectively, being
formed with His213 ND1, and with another averaging 2.8 Å and 2.7 Å in the octamer and the
heptamer, respectively, being formed with Trp99
NE1. The Gln33 NE2–Gly97 O H-bond is similar in
all structures, averaging 2.8 Å in both the octamer
and the heptamer.
Calcium- and ligand-binding site
The pentraxin calcium-binding site in both structures exhibits an overall topology similar to those in
the known calcium-bound pentraxin structures,
with (in the native structures) two calcium ions
present in the ligand-binding site in each protomer. In the native structures, calcium ion 1 is
6-coordinated (by Asp59 OD2, Asn60 OD1, Glu145
OE1, Glu145 OE2, Asp147 OD1, and the main-chain
Fig. 6. Stereo view of electron density in the calcium- and ligand-binding site of the PE-bound octameric structure. The
difference electron density map (Fo − Fc) is calculated from the observed data Fo and the final model Fc omitting the PE
ligand and the calcium ions. The map, calculated at 2.7 Å resolution, is noncrystallographically averaged, solventflattened, and contoured at 1 rmsd.
1246
Multiple Aggregates for Limulus SAP-Like Pentraxin
carbonyl of Gln146), and calcium 2 is 5-coordinated
(by Glu145 OE1, Asp147 OD1, Asp147 OD2, Glu154
OE2, and Glu157 OE1).
At the end of the shallow ligand-binding pocket
farthest from the calcium ions lie residues Gln79,
Asn85, and Tyr49 (Fig. 5). Asn85 makes hydrogenbond interactions with both Gln79 and Tyr49.
Farther round the binding pocket, Gln79 OE1
forms a hydrogen bond with ND2 of the calciumcoordinating Asn60. These residues appear to be
positioned in the pocket ready to form interactions
with ligands, including the amine group of a PEcontaining ligand, with the phosphate group interacting with the Ca ions in a similar manner to that
described for human pentraxin structures.21,26
Cocrystallization of the SAP octameric form with
PE ligand shows strong electron density for phosphate binding into the calcium-binding site, although the density for the remainder of the ligand is
poorly defined, suggesting alternative conformations (Fig. 6). In the PE-bound structure, there is
also additional density in the calcium-binding site,
which—based on the appearance and significance
of the electron density, the constituents of the crystallization buffers, and the location and nature of
proximal atoms in the structure—has been assigned
as a third calcium ion Ca3 (Figs. 5 and 6). The
average Ca3–O bond distance agrees with the characteristic value of 2.4 Å for Ca2 +-binding sites in
proteins.27 This novel third calcium, which is not
present in either of the native structures, is coordinated by Glu154, Glu157, and a phosphate oxygen
from the PE ligand. This phosphate oxygen also
coordinates to Ca2 (bringing Ca2 coordination up to
6), and a second phosphate oxygen coordinates to
Ca1 (bringing Ca1 coordination up to 7). The interactions between Ca3 and the two glutamates differ
between the two independent subunits A and B,
with Glu154x1 in subunit A, Glu154x2 in subunit B,
and Glu157x1 in both. This may be due to the crystal
contact in the region of the subunit A calciumbinding site. The temperature factors for the three
calcium ions are consistent across the two independent subunits, averaging 31 Å2 for both Ca1 and
Ca2, and 45 Å2 for Ca3. These values suggest that
Ca1 (7-coordinated) and Ca2 (6-coordinated) have
similar occupancies (similar also to main-chain protein atoms with an average B-factor of 30 Å2; see
Table 1), with Ca3 (3/4-coordinated) probably at a
similar level and with the higher B-factor being
consistent with reduced coordination.
Despite the somewhat amorphous appearance of
the density associated with the ethanolamine group,
the conformation that best fits the electron density
Table 1. Data collection and processing
Space group
Cell dimensions
Maximal resolution (Å)
Observations
Unique reflections
Completeness (%)
Rmergea
I/σ(I)
Highest resolution bin (Å)
Observations
Unique reflections
Completeness (%)
Rmergea
I/σ(I)
Protein atoms
(independent subunits)
Other atoms
Resolution range (Å)
Rconvb (%)
Rfreec (%)
rmsd for angles (°)
rmsd for bonds (Å)
Average B for main chains (Å2)
Average B for water
molecules (Å2)
Ramachandran plot (%)d
Favored
Outliers
a
PE-bound 16-mer
Native 14-mer
I422
a = 172.4 Å; b = 172.4 Å; c = 98.4 Å
2.7
78,485
20,530
99.6
0.084
5.0
2.85–2.70
10,348
2738
98.9
0.252
2.6
P21
a = 98.3 Å; b = 167.6 Å; c = 141.0 Å; β = 92.5°
2.3
374,693
180,371
89.7
0.080
5.4
2.42–2.30
20,703
9565
93.2
0.167
3.9
Native 16-mer
3336 (2)
Refinement
PE-bound 16-mer
3336 (2)
Native 14-mer
23,352 (14)
4 calcium ions
—
60.9–3.00
18.4
23.2
1.342
0.0071
30.28
13.43
6 calcium ions
16 ligand
60.9–2.70
21.0
24.3
1.356
0.0068
27.85
28 calcium ions
881 water
45.2–2.30
19.5
21.9
1.313
0.0060
13.57
94.9
0.0
97.7
0.0
98.2
0.0
Rmerge = ∑h∑j|Ih,j − Ih|/∑h∑j|Ih,j|, where Ih,j is the jth observation of reflection h.
Rconv = ∑h‖Foh| − |Fch‖/∑h|Foh|, where Foh and Fch are the observed and calculated structure factor amplitudes,, respectively, for
reflection h, except for those included in Rfree.
c
Rfree is equivalent to Rconv for an ∼ 5% subset of reflections not used in the refinement.
d
Defined according to MolProbity.
b
Multiple Aggregates for Limulus SAP-Like Pentraxin
1247
Fig. 7. Transmission electron microscopy imaging of SAP oligomers. (a) Cryo-EM image of a nonstained frozen
hydrated preparation. (b) Cryo-negative stain EM image. Stacks of ring-like particles were frequently observed under all
preparation conditions. The scale bars in (a) and (b) represent 100 nm. (c) The side view of the 3D reconstruction from cryonegative stain EM. (d) The top view of the model. The scale bars in (c) and (d) represent 11.8 nm. (e) The FSC plot of the final
reconstruction showing a resolution of 1.4 nm.
places the ethanolamine nitrogen adjacent to and
within hydrogen-bonding distance of the hydroxyl
group of tyrosine 49, Tyr49 OH–N, being 2.6 Å and
2.7 Å in subunits A and B, respectively. There are no
other specific interactions between the monoester
and the protein. Torsional rotation of the terminal PE
amine about the C–N bond brings the amine
nitrogen to within hydrogen-bonding distance of
both Gln79 and Tyr49.
Structural analysis by transmission electron
microscopy
To further investigate the two molecular aggregates, SAP particles were analyzed by cryo-electron
microscopy (cryo-EM) imaging of frozen hydrated
(Fig. 7a) and negatively stained frozen hydrated
preparations (Fig. 7b). Imaging of cryo-negatively
stained preparations showed doughnut-shaped
particles with an outer diameter of 10.4 ± 0.4 nm,
an inner hole of 4.2 ± 0.6 nm diameter, and thickness
of 5.8 ± 0.4 nm. These dimensions agree with those of
the heptameric X-ray structure (the outer diameter,
inner diameter, and thickness are 11.4 nm, 4.6 nm,
and 6.1 nm, respectively), but are smaller than the
octameric X-ray structure (dimensions of 13.1 nm,
6.2 nm, and 6.1 nm). These findings and dimensions
were confirmed by cryo-EM imaging of frozen
hydrated preparations (Fig. 7a). The ring-like SAP
oligomers were occasionally found to stack onto
long rods (Fig. 7b). This was observed most
frequently in the cryo-negatively stained preparations, but was also present in nonstained cryo-EM
images, indicating that this behavior was not staininduced. The average width of 22 selected rod-like
stacks was 10.3 ± 0.3 nm, consistent with the hepta-
meric molecular form rather than the octameric
molecular form.
From images of cryo-negatively stained preparations, 22,897 particles were semiautomatically
selected. After classification and averaging, class
average images showed the ring-like structure of
the SAP complexes under different orientations.
The heptameric and octameric X-ray crystallographic structures were used to generate 41 reference projections from each structure (Fig. 8a). The
two reference volumes from X-ray crystallography
had slightly different diameters (11.4 nm versus
13.1 nm). Multireference angular assignment of the
cryo-negatively stained single particle images to the
reference projections showed that 93% of the
particles had a significantly higher correlation
coefficient when aligned to the 7-fold symmetric
references (Fig. 8a–c). The 2409 particle images that
showed the particles in the ring-like “top-view”
orientation were selected after reference-free alignment and classified into eight classes that had 593,
523, 392, 312, 197, 187, 172, and 33 particles (Fig.
8d). No significant variation in ring diameters was
observed among these classes. The average image
and rotational power from the four most populated
classes are shown in Fig. 8d and e. This rotational
symmetry analysis indicates that the 7-fold symmetric form dominates the imaged particle population. Lower-symmetry contributions such as 2-, 3-,
and 4-fold could originate from residual tilt in the
imaged particles, while the intensity at 6- and 8fold could be due either to the residual intensity
from 7-fold symmetry particles or the existence of
6- and 8-fold symmetry particles. The resolution of
the final 7-fold symmetric 3D reconstruction (Fig.
7c and d), determined from cryo-negatively stained
1248
Multiple Aggregates for Limulus SAP-Like Pentraxin
Fig. 8. Analysis of the rotational symmetry of cryo-negative stain EM images. (a) Selected references derived from the
heptameric and octameric X-ray structures, projected in different directions and filtered to 20 Å resolution. (b)
Corresponding reference-based class averages from aligned cryo-negative stain EM images. The number of particle
images that were assigned to these references is indicated underneath the class averages. No particles were assigned to the
8-fold symmetric top-view class. (c) Selected individual particle images from those classes. (d) The averages of the particle
images of the four largest top-view classes obtained by reference-free alignment and correspondence analysis. (e) The
rotational power spectrum for each top-view class average.
particles, was 14 Å [0.5 Fourier shell correlation
(FSC); Fig. 7e].
Discussion
Overall structure
The Limulus SAP-like pentraxin structures reported here reveal a new pentraxin cyclic molecular
aggregation in addition to the cyclic octamer14 and
pentamer21,22 reported previously. At this resolution, they also reveal identical protomers aggregating to form two different molecular forms for the
same species. The calcium- and ligand-binding sites
of Limulus SAP are located on the perimeter of the
cyclic aggregate, in contrast to human CRP and SAP,
where they are located on one face of the pentamer.21,22 The presence of the ligand-binding sites
arrayed around the outside of these large molecules
provides the ideal geometric arrangement for efficient multiple recognition and binding, and presumably facilitates the action of Limulus SAP as an
agglutinin.17,28 This could not be achieved easily
with a protomer orientation similar to that in the
human pentraxins.
The pentraxin fold is highly conserved in the
pentraxins,14 but the dimer present in both Limulus
SAP structures reported here is thus far unique in
the family. Electron micrographs indicate that other
mammalian pentraxins are likely to be similar to
the human structures in that they form single ring
structures. Although these rings can aggregate and
form stacks,29,30 they do not produce a similar type
of hydrogen-bonded dimer structure within the
aggregation. Several animal lectins not only possess a monomeric fold that is related to the legume
lectin fold but also adopt similar quaternary
structures,31 often containing dimers of identical
1249
Multiple Aggregates for Limulus SAP-Like Pentraxin
subunits resulting in an extended β-sheet structure.
The similarities of the pentraxin protomer fold to
that exhibited by concanavalin A have been noted
previously,21 and although the Limulus SAP dimer,
formed across a crystallographic dyad in the
octameric form, is not identical with the canonical
legume lectin dimer, both exhibit an extended
β-sheet structure.
Variable aggregation
Several plausible explanations offer themselves
for the presence of two distinct molecular aggregates, but these must be viewed in the light of the
single Limulus SAP sequence reported,18 the resolution of the X-ray data, and the pooling of hemolymph from several animals prior to pentraxin
isolation and purification.
The native octameric form diffracts to a resolution of 3 Å, the native heptameric form to 2.3 Å,
and the PE-bound octameric form to 2.7 Å. At the
resolution of the octameric structures, and consistent with the single cDNA sequence reported
for Limulus SAP pentraxin,18 there is no definitive
evidence in the electron density for amino acid
differences between the two molecular species,
either in general or at the subunit–subunit
interfaces around the cyclic aggregates (Fig. 4).
Furthermore, the surface areas of these contact
interfaces are remarkably similar in both the
heptameric and the octameric structures, differing
by only 5% between the two aggregates. One
would be very hard-pressed indeed not to draw
conclusions from structural data that both aggregations are energetically and chemically favorable
and that there is a single translated sequence18
common to both forms. Indeed, only Lys41 (see
Fig. 4) of the interface residues was part of the
gene-specific primers on which the cDNA
sequence was based. Perhaps the most compelling
argument for a single gene is that the nature of
the subunit–subunit interface may be influenced
by differences in glycosylation.32 Two different
glycosylation states—perhaps of many14—in the
two molecular forms may affect the interface
sufficiently to drive aggregation in two different
but similar directions. Whatever the cause of the
variable aggregation, differences between the two
molecular forms in terms of stability and potential
to aggregate and precipitate would be expected,
offering an explanation for the dominance of the
heptameric form on cryo-EM images.
Despite the available evidence, the possibility that
two different genes code for Limulus SAP proteins
cannot be entirely discounted. These genes may
differ only slightly in sequence in a region(s)
associated with or affecting the subunit–subunit
interface, or may be differentially regulated through
differing 3′ untranslated regions.18 Further investigation of the origin of the two molecular forms of
Limulus SAP involving higher-resolution single
crystal studies and cryo-EM of pentraxins purified
from single animals is underway.
Calcium- and ligand-binding site
The calcium-binding configuration of the two
primary calcium ions Ca1 and Ca2 in the Limulus
SAP structures (Fig. 5) is similar to that in the two
human pentraxins.21,22 It is more similar, but not
identical, to that in human CRP22 rather than to that
in human SAP,21 in part due to the presence of
Glu154 (Glu147 in human CRP) rather than the
equivalent Asp in human SAP. Coordination of Ca1
is identical with that in human CRP, while Ca2
differs between the two in that Limulus SAP Asp147
coordinates to Ca2 via OD1 and OD2, whereas only
OD2 (human CRP Asp140) coordinates to Ca2 in
human CRP, and coordination is completed by
Gln150 OE1 in human CRP, equivalent to the Ca2–
Glu157 OE1 interaction in Limulus SAP.
Upon binding of PE, a third calcium ion (Ca3) is
bound in each protomer (Figs. 5 and 6). The phosphate is clearly seen in the electron density (Fig. 6)
interacting with all three calcium ions, with interactions with the standard pentraxin ion pair (Ca1 and
Ca2) being similar to those observed in other ligandbound pentraxin structures.21,26,30,33,34 Despite the
clarity of the phosphate, the monoester appears to
be weakly bound and ill-defined. While this may be
a result of insufficient PE ligand present in the
crystallization or cryobuffers, it may also reflect
alternative modes of binding arising from the nature
of the ligand-binding pocket in Limulus SAP,
particularly the presence of Gln79. The major PE
conformation inferred from the electron density
gives a hydrogen bond between the amine nitrogen
and Tyr49 OH, but a change in PE conformation
brings this N within hydrogen-bonding distance of
Gln79. With the side chains of Tyr49, Gln79, and
Asn85 arrayed around the extremes of the binding
pocket distant from the calcium sites (Fig. 5), it
seems likely that specificity is relatively broad and
that PE may not be the primary target of Limulus
SAP.
In contrast to the conserved calcium- (Ca1 and
Ca2) and phosphate-binding site, in terms of both
coordinating residues and overall structure, the
binding pocket extending from the phosphatebinding site in Limulus SAP shows little sequence
similarity to that in human SAP or, indeed, in any
other known pentraxin sequence. The pocket is
formed by the β-strands D, E, F, and G, with strand
D placed nearest to the calcium-binding site (Fig. 5).
It is of interest that the glycosylation site at Asn81
lies in the reverse turn connecting strands F and G,
which form the far wall of the binding pocket and
include Gln79 and Asn85, respectively. This may
affect the ligand binding specificity.
In human SAP, the lysine in strand G at position
hSAP79 (human SAP numbering) interacts with
hSAP Glu66 (strand E) and bound MOβDG and, in a
similar manner to the equivalent residue hCRP
Glu81 (human CRP numbering) in human CRP,
is likely to be an important determinant of
specificity.22,33 In Limulus SAP, there is a single
residue insertion in strand G, and the Limulus SAP
1250
Multiple Aggregates for Limulus SAP-Like Pentraxin
Fig. 9. Sequence and structure
alignment in the calcium- and
ligand-binding region. The arrow
indicates the insertion of Limulus
SAP (LSAP) Tyr49 side chain into
the ligand-binding pocket position
occupied by Tyr64 in human SAP
(hSAP) and by Phe66 in human
CRP (hCRP). The four β-strands
underlying the ligand-binding site
are double underlined and labeled
D, E, F, and G. The three known Limulus CRP sequences are shown as 1.1CRP, 3.3CRP, and 1.4CRP, with heterogeneity
shaded and with the two differences between 1.1 and 3.3 underlined.
side chain most structurally related to hSAP Lys79 is
LSAP Asn85 (LSAP numbering). LSAP Asn85 forms
hydrogen bonds with both LSAP Tyr49 (strand D)
and LSAP Gln79 (strand F) in Limulus SAP (Figs. 5
and 9). In human SAP, the topologically nearest
equivalent residue to Limulus SAP Gln79 is the key
SAP-conserved residue hSAP Tyr74, which makes a
major contribution to the formation of the SAPbinding pocket.21,30,33 The structural space occupied
by hSAP Tyr74 in human SAP is partially occupied
by both Gln79 and Asn85 in Limulus SAP.
The SAP tyrosine that lies at position hSAP64
(strand E) in human SAP is generally conserved as
either Y or F in mammalian SAP and CRP (Fig. 9),
respectively, and, in both pentraxins, it is central to
the formation of the ligand-binding pocket.21,22,26 In
the human SAP complex with dAMP,30 hSAP Tyr64
forms a hydrogen bond with the 2′-deoxyadenosine
ring oxygen atom. In human CRP, this F (hCRP
Phe66; human CRP numbering) forms the basis of
the hydrophobic pocket beneath the phosphocholine (PC)-binding site.22 In Limulus SAP, the equivalent residue in terms of the alignment of sequence
and fold is LSAP Ser65, while the equivalent residue
in terms of the positioning of the side chain is LSAP
Tyr49 (strand D), equivalent to a conserved serine in
mammalian CRP and SAP. Thus, despite the
absence in terms of sequence of this critical tyrosine,
the positioning of the side chain of LSAP Tyr49,
which relative to human SAP lies on the preceding
β-strand of the sheet underlying the ligand-binding
pocket, provides the required physical characteristics for the formation of the pocket and, where
appropriate, interaction with ligand (Fig. 9). Indeed,
the interaction of bound PE here with LSAP Tyr49
closely mirrors that between hSAP Tyr64 and dAMP
in human SAP. We have termed this feature
“topological homology,” with identical side chains
from different regions of two structures occupying
similar spaces in both. Another example of this is
provided by the remarkable spatial overlap between
the side chain of Limulus SAP Trp99 and that of the
pentraxin Trp at human SAP 203, which, in terms of
sequence and fold alignment, is equivalent to
Limulus SAP Met214. Thus, despite the modest
sequence homology with human SAP (22%), the
lack of sequence similarity in the binding pocket,
and the absence of some SAP and pentraxin motifs
associated with both structure and ligand binding
specificity (Fig. 9), topological homology contributes
towards the equivalence of both overall fold and
ligand binding to PE. In more general terms, topological equivalence of side chains will allow proteins
of the same family with low sequence homology to
adopt closely similar folds with common physical
characteristics and to recognize and bind the same
ligands.
Phylogenetic and immunological aspects
Phylogenetic analysis aligns the horseshoe crab
pentraxins with the long pentraxins rather than with
mammalian CRPs and SAPs,18,35 and shows that
gene duplication of CRP (or SAP), followed by
sequence divergence and evolution of CRP and/or
SAP function, occurred independently along the
chordate and arthropod evolutionary lines rather
than in a common ancestor.18 The diversification of
the horseshoe crab and mammalian CRPs and SAPs
may also be further reflected in the aggregation of
these pentraxins.
The horseshoe crab CRPs have been shown to have
varying ligand binding and cytolytic activities,6,17
although all appear to exhibit the CRP-characteristic
Ca-dependent binding to PC (although not necessarily PC agarose).17 The majority of the 10%
sequence difference between the Limulus CRP
sequences 1.4 and 1.1/3.315,16 lies in the region of
the ligand-binding pocket, particularly in strands D,
E, and F and the preceding 310 turn, with the two
amino acids differing between 1.1 and 3.3 also in
this region (Fig. 9). Significantly, the key ligandbinding pocket residue Tyr49 in Limulus SAP is
replaced by histidine in CRP1.1 and CRP3.3, and by
threonine in CRP1.4. This will directly affect ligandbinding properties. Alignment of all the known
Tachypleus tridentatus CRPs17 onto the Limulus SAP
and human CRP structures22 reveals considerable
sequence homology in the Ca-binding sites, but no
strong amino acid conservation in the region of the
PC-binding site, with significant variation between
the three different groups of T. tridentatus CRPs
again present in the ligand-binding pocket. Evidence from sequence and structure suggests that
adaptation of CRP-mediated immune responses in
these ancient creatures lacking antibody-based
1251
Multiple Aggregates for Limulus SAP-Like Pentraxin
immunity is based on adaptation through sequence
changes that follow gene duplication.
There is an emerging picture of the immunological
role of the pentraxins in Limulus. They can agglutinate microbes, which may serve to immobilize
invading microbes and prevent their systemic dissemination. 17,28 They can also form hydrophilic
channels across lipid bilayers, including the lipopolysaccharide-rich outer membrane of Gram-negative
bacteria,28 a function that may serve to mediate the
direct cytolytic destruction of invading microbes.
Additionally, the horseshoe crab pentraxins form
complexes with other proteins of the plasma that
have been shown to be bacteriolytic.8,9 The Limulus
SAP-like pentraxin is a significant constituent of
hemolymph and, as such, is clearly an important
part of host defense. It may, as in humans, play a
role in Ca-dependent DNA binding in serum36 or
may simply meet the need for agglutination and/or
cytolytic destruction of pathogens and/or parts of
dying cells, displaying surface PE or carbohydrate
moieties.
The discovery and characterization of two oligomeric states for Limulus SAP raise further questions.
It may be speculated that the different molecular
forms are functionally different (with variability of
expression or posttranslational modification linked
to a physiological response of the immune system to
infection or other external challenge), that they influence the stability or functioning of antimicrobial
molecular complexes,8,9 and that individual horseshoe crabs produce either the heptameric or the
octameric form and that this may be linked to
gender. An immune response linked to gender specificity has been reported for the serum pentraxin
CrOctin in the horseshoe crab C. rotundicauda.11 If
only one of the two forms of Limulus SAP is found in a
single animal and the source of the molecular difference is not gender-associated, it is tempting to further
speculate that the evolution of the SAP-like protein in
Limulus is at a crossroads, with the two forms
representing the past and the future, respectively.
Materials and Methods
Extraction and purification
Recently collected adult horseshoe crabs were obtained
from the Marine Resources Center of the Marine Biological
Laboratory (Woods Hole, MA). Blood was collected from
prechilled animals by cardiac puncture under sterile
lipopolysaccharide-free conditions, and blood cells were
removed immediately after bleeding.37 It is important to
avoid degranulation of the blood cells, since this releases
proteases38 and active-site protease inhibitors39–41 into the
serum. Animals were released into the ocean unharmed
after bleeding.
Crystallization and native gels
Isolation and purification of Limulus SAP for crystallization and native gels were carried out in accordance
with Tharia et al.18 Briefly, diluted pooled plasma was
applied to a PE-agarose column, and SAP was eluted
with 50 mM PE in calcium buffer after the removal of
CRP pentraxins by an initial isocratic elution with
10 mM PC chloride (Sigma) in calcium buffer (50 mM
Tris (pH 7.4), 150 mM NaCl, and 10 mM CaCl2). Both
elutions were performed at a flow rate of 0.5 ml/min
and monitored at 280 nm. The SAP was first dialyzed
versus approximately 100 vol of buffer [10 mM ethylenediaminetetraacetic acid, 50 mM Tris, and 150 mM
NaCl (pH 7.4)] with one change over 24 h, followed by
dialysis against the above calcium buffer with one
change over 24 h. The SAP was then concentrated
using a 10-kDa molecular mass cutoff centrifugal filter
device (Amicon Ultra–Millipore) at 2500g ready for use
in crystallization.
Cryo-electron microscopy
Limulus SAP for the cryo-EM study was prepared from
a stock of mixed Limulus pentraxins (LPx) isolated from
hemolymph as described previously42 and stored at 4 °C.
The first deposits in this pooled stock were made in 1985,
with use and replenishment over the past two decades.
The preparation of the LPx was carried out as follows:
Much of the hemocyanin was removed from plasma by
ultracentrifugation (141,000g for 16 h) or by incubation
with 3% polyethylene glycol (PEG) 8000 with centrifugation at 30,000g for 0.5 h. The supernatant was then made
10% in PEG 8000 and centrifuged as described above, and
the precipitate was redissolved in 0.1 M citrate (pH 6.7).
The remaining traces of the hemocyanin were precipitated
by the addition of zinc acetate to a final concentration of
0.1 M, followed by centrifugation at 30,000g for 0.5 h to
remove the hemocyanin precipitate. The hemocyanin-free
plasma was then precipitated with 10% PEG 8000 and
redissolved in 0.15 M NaCl, 50 mM Tris (pH 7.3), and
2 mM ethylenediaminetetraacetic acid. Calcium (CaCl2)
was added to a final concentration of 10 mM, and the
precipitate that formed was removed by centrifugation
(30,000g for 0.5 h). The supernatant was incubated with
0.2 vol of Sepharose 4B to remove the Sepharose-binding
proteins, followed by incubation with 0.1 vol of PE
agarose with gentle stirring. Incubation of resin with the
3–10% PEG 8000 cut of hemolymph in suspension avoided
the precipitation of the LPx that occurred when solutions
were applied to columns of PE agarose. After incubation,
the pentraxin-charged PE agarose was packed in a column
and washed with 1.0 M NaCl, 10 mM CaCl2, and 50 mM
Tris (pH 7.3), then the bound pentraxins were eluted with
0.1 M citrate (pH 7.0). The pooled fractions from PEagarose columns were stored in 0.1 M citrate at 4 °C.
Limulus SAP was purified from this sample using the
methods described by Tharia et al.18
Native gel
Limulus SAP was treated with a sample buffer
without SDS, reducing agent, or heating, and run at
loadings of 2.5 μg and 5 μg per lane on a discontinuous
native polyacrylamide gel (pH 6.8, 4%T stacking; pH 8.8,
5–12.5%T gradient resolving gel; 2.6%C) and electrophoresed at 200 V for 2 h using the Leammli buffer
system without SDS. A High Molecular Weight Calibration Kit (Amersham Biosciences) was used for molecular
weight markers. Additional separate lanes of ferritin
from equine spleen, Mr ∼ 440,000 (Sigma); catalase from
bovine liver, Mr ∼240,000 (Sigma/Fluka); and L-lactate
dehydrogenase from bovine heart, Mr ∼140,000 (Sigma)
1252
were included for confirmation of marker band molecular weight values. The gel was stained with Coomassie brilliant blue R-250. A Gel Doc-It Imaging System
(UVP) and the integral Labworks software were used for
estimation of the Limulus SAP molecular weight by
comparing the mobility of Limulus SAP bands with those
of calibration kit molecular weight markers on the gel.
Crystallization and data collection
The purification, crystallization, and data collection
details for the native octameric structure have been
described previously.14,43 Crystals of PE-bound octameric
SAP and the new native form were grown in sitting drops
using equal volumes of protein [7–12 mg/ml in 150 mM
NaCl, 10 mM CaCl2, and 50 mM Tris (pH 7.4)] and
precipitating solution [6% PEG 6000, 10 mM CaCl2, and
50 mM 4-morpholineethanesulfonic acid (pH 7)], with the
addition of 20 mM PE in the precipitating solution for the
ligand-bound crystals. Crystals were prepared for cryocooling by the successive addition of 2-μl aliquots of
increasing concentrations (5%, 10%, 15%, and 20%) of
methyl-2,4-pentanediol (MPD) in precipitant buffer, followed by addition of a further 2-μl aliquot of 20% MPD
cryobuffer, and an exchange of ∼ 10 μl of the resulting
buffer with 20% MPD cryobuffer. Data were collected
from a single crystal in each case with an ADSC Quantum
4R CCD detector on Daresbury SRS station 14.1. Integrated intensities were calculated with the program
MOSFLM.44 Data collection and processing statistics are
given in Table 1.
Structure solution and refinement
The known Limulus SAP sequence18 was built into the
previously determined native octameric polyalanine
structure.14 CCP4 programs44 were used throughout to
calculate maps, and manual model building was carried
out using the program O.45 The model was refined
initially using X-PLOR46 and completed using CNS.47
Noncrystallographic symmetry averaging was used
throughout, except for those residues involved in crystal
contacts. The refined native octameric structure was then
used as a starting model for both the PE-bound octameric
SAP and the new native form. Crystals of PE-bound SAP
were isomorphous to the native octameric structure, and
molecular replacement was not necessary. For the new P21
crystal form, molecular replacement was carried out using
AMORE.48 As the full 16-mer was too large for input into
AMORE, an octameric ring was used as initial search
model. The octameric search model did not produce any
obvious or successful solutions, and the Limulus dimer
from the refined octameric structure was then used as
search model. This produced seven (and only seven)
outstanding rotations, with three giving single translations, two giving two possible translations, and two giving
multiple translation possibilities. Moving progressively
through these solutions by refining the top solution, fixing
it, and searching for translations for the remaining
rotations gave a single and outstanding set of seven
rotations and translations. These subsequently packed
into a doubly stacked ring composed of seven dimers for a
starting model comprising 14 protomers. Model building
of the structures was carried out using maximum likelihood refinement with CNS47 and alternated with rounds
of manual model building with the program O.45 Mainchain noncrystallographic symmetry averaging was used
throughout. Topology and parameter files for ligand were
Multiple Aggregates for Limulus SAP-Like Pentraxin
obtained from the HIC-Up server.49 Refinement statistics
are given in Table 1, and the quality of the final structures
was verified by MolProbity.50 The structures have N 95%
residues in favored regions of the Ramachandran plot,
with no outliers. The electron density is poorly defined for
residues 184–187.
Transmission electron microscopy and image
processing
Cryo-EM grid preparations were made by applying
2.5 μl of purified SAP sample at 1 mg/ml concentration to
holey carbon film grids (Quantifoil Micro Tools, GmbH),
blotted, and vitrified by quick-freezing in liquid-nitrogencooled liquid ethane.51 For cryo-negative stain EM, grids
were prepared as described.52 Briefly, the sample was
diluted to 0.1 mg/ml, and 2.5 μl of sample was applied to
holey carbon film grids, which were then layered on top of
a 100-μl drop of saturated ammonium molybdate solution
(pH 7). After 60 s of staining, the grid was picked up,
blotted for 6 s, air-dried for 3 s, and vitrified by quickfreezing as described above. Cryo-EM grids were imaged
in a JEOL JEM-2100F using a Gatan-626 cryo-EM sample
holder. The microscope was operated at 200 keV, and
images were recorded at a nominal magnification of
50,000× under minimum dose conditions on a TVIPS F415
CCD camera.
For image processing, particles were selected automatically from eighty-one 4000 × 4000 CCD images in the
program Boxer53 and screened manually. To generate the
reference projection maps from the atomic X-ray diffraction models of the heptameric and octameric forms, the
atom coordinates from those models were used to generate
a volume that was projected into 41 different directions,
and the resulting images were Fermi-low-pass-filtered to
20 Å resolution. The cryo-negative stain EM particle
images were aligned against those references, classified,
and averaged in each class. Using the same data set,
particles that showed the “top view” were selected,
reference-free-aligned, and subjected to multivariate statistical analysis using correspondence analysis and hierarchical ascendant classification.54 These particles were
then grouped into eight classes and averaged. The class
averages were analyzed for their rotational symmetry by
transformation into polar coordinates, selection of ringcorresponding densities, and calculation of power
spectra.55 Both reference-based and reference-free analyses
were performed on the platform of SPIDER software.56
The cryo-negatively stained particle images were also
subjected to 3D reconstruction, using projection matching
and the EMAN and SPIDER software packages.53,56,57 The
initial model was made from the stack of the top-view
class average. After each iteration of the 3D reconstruction, D7 symmetry was applied to the reference model.
The resolution of the final reconstruction (Fig. 7c and d)
was determined by producing two reconstructions from
randomly assigned subgroups of particles and comparing
them by FSC, using the 0.5 FSC criterion.58 The handedness of the final model was chosen to follow the
orientation of the X-ray protomer structure.
Accession numbers
The coordinates and structure factors of both octameric and heptameric Limulus SAPs have been deposited
in the Protein Data Bank with accession numbers 3FLR
(octameric; native), 3FLT (octameric; PE-bound), and 3FLP
(heptameric). The single-particle D7 symmetric cryo-EM
Multiple Aggregates for Limulus SAP-Like Pentraxin
3D reconstruction was deposited in the Electron Microscopy Data Bank† under accession number 1598.
Acknowledgements
The authors wish to acknowledge support from
the Medical Research Council (A.K.S., T.J.G., and
I.B.), CLRC Daresbury Laboratory (T.J.G. and A.K.S.),
National Institutes of Health (grant U54 GM074929;
H.-T.C. and H.S.), and the National Science Foundation (grant 0344360; P.B.A.).
References
1. Gewurz, H., Zhang, X.-H. & Lint, T. F. (1995).
Structure and function of the pentraxins. Curr. Opin.
Immunol. 7, 54–64. (And references therein).
2. Volanakis, J. E. (1982). Complement activation by
C-reactive protein. Ann. N. Y. Acad. Sci. 389,
235–250.
3. Szalai, A. J., Agrawal, A., Greenhough, T. J. &
Volanakis, J. E. (1999). C-reactive protein: structural
biology and host-defence function. Clin. Chem. Lab.
Med. 37, 265–270.
4. Bickerstaff, M. C. M., Botto, M., Hutchinson, W. L.,
Herbert, J., Tennent, G. A., Bybee, A. et al. (1999).
Serum amyloid P component controls chromatin
degradation and prevents antinuclear autoimmunity.
Nat. Med. 5, 694–697.
5. Armstrong, P. B., Melchior, R. & Quigley, J. P. (1996).
Humoral immunity in long-lived arthropods. J. Insect
Physiol. 42, 53–64.
6. Armstrong, P. B., Swarnakar, S., Srimal, S., Misquith,
S., Hahn, E. A., Aimes, R. T. & Quigley, J. P. (1996).
A cytolytic function for a sialic acid-binding lectin
that is a member of the pentraxin family of proteins.
J. Biol. Chem. 271, 14717–14721.
7. Swarnakar, S., Asokan, R., Quigley, J. P. &
Armstrong, P. B. (2000). Binding of α2-macroglobulin and limulin: regulation of the plasma haemolytic system of the American horseshoe crab, Limulus
polyphemus. Biochem. J. 347, 679–685.
8. Ng, P. M., Le Saux, A., Lee, C. M., Tan, N. S., Lu, J.,
Thiel, S. et al. (2007). C-reactive protein collaborates
with plasma lectins to boost immune response against
bacteria. EMBO J. 26, 3431–3440.
9. Le Saux, A., Lon Ng, P. M., Yun Koh, J. J., Ping Low,
D. H., Leong, G. E. -L., Ho, B. & Ding, J. L. (2008). The
macromolecular assembly of pathogen recognition
receptors is impelled by serine proteases via their
complement control protein (CCP) modules. J. Mol.
Biol. 377, 902–913.
10. Ng, P. M., Jin, Z., Tan, S. S., Ho, B. & Ding, J. L.
(2004). C-reactive protein: a predominant LPSbinding acute phase protein responsive to Pseudomonas infection. J. Endotoxin Res. 10, 163–174.
11. Li, Y., Ng, M. L., Ho, B. & Ding, J. L. (2007). A femalespecific pentraxin, CrOctin, bridges pattern recognition receptors to bacterial phosphoethanolamine. Eur.
J. Immunol. 37, 3477–3488.
† http://www.ebi.ac.uk/msd
1253
12. Tennent, G. A., Butler, P. J. G., Hutton, T., Woolfitt,
A. R., Harvey, D. J., Rademacher, T. W. & Pepys, M. B.
(1993). Molecular characterization of Limulus polyphemus C-reactive protein: 1. Subunit composition.
Eur. J. Biochem. 214, 91–97.
13. Robey, F. A. & Liu, T. -Y. (1981). Limulin: a C-reactive
protein from Limulus polyphemus. J. Biol. Chem. 256,
969–975.
14. Shrive, A. K., Metcalfe, A. M., Cartwright, J. R. &
Greenhough, T. J. (1999). C-reactive protein and SAPlike pentraxin are both present in Limulus polyphemus
haemolymph: crystal structure of Limulus SAP. J. Mol.
Biol. 290, 997–1008.
15. Nguyen, N. Y., Suzuki, A., Boykins, R. A. & Liu, T.-Y.
(1986). The amino acid sequence of Limulus C-reactive
protein. J. Biol. Chem. 261, 10456–10465.
16. Nguyen, N. Y., Suzuki, A., Cheng, S.-M., Zon, G. &
Liu, T.-Y. (1986). Isolation and characterization of
Limulus C-reactive protein genes. J. Biol. Chem. 261,
10450–10455.
17. Iwaki, D., Osaki, T., Mizunoe, Y., Wai, S. N.,
Iwanaga, S. & Kawabata, S. (1999). Functional and
structural diversities of C-reactive proteins present in
horseshoe crab haemolymph plasma. Eur. J. Biochem.
264, 314–326.
18. Tharia, H. A., Shrive, A. K., Mills, J. D., Arme, C.,
Williams, G. T. & Greenhough, T. J. (2002).
Complete cDNA sequence of SAP-like pentraxin
from Limulus polyphemus: implications for pentraxin
evolution. J. Mol. Biol. 316, 583–597.
19. Robey, F. A., Tanaka, T. & Liu, T.-Y. (1983). Isolation and
characterization of two major serum proteins from the
dogfish, Mustelus canis, C-reactive protein and amyloid
P component. J. Biol. Chem. 258, 3889–3894.
20. Samudzi, C. T., Nguyen, N. Y. & Rubin, J. R. (1993).
Crystallization and preliminary X-ray diffraction
studies of dogfish C-reactive protein. Proteins Struct.
Funct. Genet. 15, 100–102.
21. Emsley, J., White, H. E., O'Hara, B. P., Oliva, G.,
Srinivasan, N., Tickle, I. J. et al. (1994). Structure of
pentameric human serum amyloid P component.
Nature, 367, 338–345.
22. Shrive, A. K., Cheetham, G. M. T., Holden, D., Myles,
D. A. A., Turnell, W. G., Volanakis, J. E. et al. (1996).
Three-dimensional structure of human C-reactive
protein. Nat. Struct. Biol. 3, 346–354.
23. Kraulis, P. J. (1991). MOLSCRIPT: a program package
to produce both detailed and schematic plots of
protein structures. J. Appl. Crystallogr. 24, 946.
24. Agrawal, A., Shrive, A. K., Greenhough, T. J. &
Volanakis, J. E. (2001). Topology and structure of the
C1q binding site of C-reactive protein. J. Immunol. 166,
3998–4004.
25. Krissinel, E. & Henrick, K. (2007). Inference of macromolecular assemblies from crystalline state. J. Mol.
Biol. 372, 774–797.
26. Thompson, D., Pepys, M. B. & Wood, S. P. (1999). The
physiological structure of human C-reactive protein
and its complex with phosphocholine. Struct. Fold.
Des. 7, 169–177.
27. Harding, M. M. (1999). The geometry of metal–ligand
interactions relevant to proteins. Acta Crystallogr. Sect.
D, 55, 1432–1443.
28. Harrington, J. M., Chou, H.-T., Gutsmann, T., Gelhau,
C., Stahlberg, H., Leippe, M. & Armstrong, P. B.
(2008). Membrane pore-formation by pentraxins.
Biochem. J. 413, 305–313.
29. Painter, R. H., de Escallon, I., Massey, A., Pinteric, L. &
Stern, S. B. (1982). The structure and binding cha-
1254
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
racteristics of serum amyloid protein (9.55 α1-glycoprotein). Ann. N. Y. Acad. Sci. 389, 199–213.
Hohenester, E., Hutchinson, W. L., Pepys, M. B. &
Wood, S. P. (1997). Crystal structure of a decameric
complex of human SAP with bound dAMP. J. Mol.
Biol. 269, 570–578.
Bouckaert, J., Hamelryck, T., Wyns, L. & Loris, M.
(1999). Novel structures of plant lectins and their
complexes with carbohydrates. Curr. Opin. Struct. Biol.
9, 572–577.
Amatayakul-Chantler, S., Dwek, R. A., Tennent, G. A.,
Pepys, M. B. & Rademacher, T. W. (1993). Molecular
characterization of Limulus polyphemus C-reactive
protein: II. Asparagine-linked oligosaccharides. Eur.
J. Biochem. 214, 99–110.
Thompson, D., Pepys, M. B., Tickle, I. & Wood, S. (2002).
The structures of crystalline complexes of human SAP
with its carbohydrate ligand, the cyclic pyruvate acetal
of galactose. J. Mol. Biol. 320, 1081–1086.
Pepys, M. B. , Herbert, J., Hutchinson, W. L., Tennent,
G. A., Lachmann, H. J, Gallimore, J. R. et al. (2002).
Targeted pharmacological depletion of serum amyloid P component for treatment of human amyloidosis. Nature, 417, 254–259.
Hughes, A. L. (1998). Protein phylogenies provide
evidence of a radical discontinuity between arthropod
and vertebrate immune systems. Immunogenetics, 47,
283–296.
Pepys, M. B. & Butler, P. J. G. (1987). Serum amyloid P
component is the major calcium-dependent specific
DNA binding protein of the serum. Biochem. Biophys.
Res. Commun. 148, 308–313.
Armstrong, P. B. (1985). Amebocytes of the American
“horseshoe crab” Limulus polyphemus. In Blood Cells of
Marine Invertebrates (Cohen, W. D., ed), pp. 253–258,
A. R. Liss, New York.
Iwanaga, S., Miyata, T., Tokunaga, F. & Muta, T.
(1992). Molecular mechanism of hemolymph clotting
system in Limulus. Thromb. Res. 68, 1–32.
Armstrong, P. B. & Quigley, J. P. (1985). Proteinase
inhibitory activity released from the horseshoe crab
blood cell during exocytosis. Biochim. Biophys. Acta,
827, 453–459.
Nakamura, T., Hirai, T., Tokunaga, F., Kawabata, S. &
Iwanaga, S. (1987). Purification and amino acid
sequence of Kunitz-type protease inhibitor found in
the hemocytes of horseshoe crab (Tachypleus tridentatus).
J. Biochem. 101, 1297–1306.
Donovan, M. A. & Laue, T. M. (1991). A novel trypsin
inhibitor from the hemolymph of the horseshoe crab
Limulus polyphemus. J. Biol. Chem. 266, 2121–2125.
Armstrong, P. B., Armstrong, M. T. & Quigley, J. P.
(1993). Involvement of α2-macroglobulin and Creactive protein in a complement-like hemolytic
system in the arthropod, Limulus polyphemus. Mol.
Immunol. 30, 929–934.
Multiple Aggregates for Limulus SAP-Like Pentraxin
43. Myles, D. A. A., Bailey, S., Rule, S. A., Jones, G. R. &
Greenhough, T. J. (1990). Preliminary crystallographic analysis of C-reactive protein from Limulus
polyphemus. J. Mol. Biol. 213, 223–225.
44. Collaborative Computational Project Number 4.
(1994). The CCP4 suite: programs for protein crystallography. Acta Crystallogr. Sect. D, 50, 760–763.
45. Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjeldgaard, M.
(1991). Improved methods for building protein models
in electron density maps and the location of errors in
these models. Acta Crystallogr. Sect. A, 47, 110–119.
46. Brunger, A. T. (1992). X-PLOR Manual, Version 3.1
Yale University, New Haven, CT.
47. Brunger, A. T., Adams, P. D., Clore, G. M., DeLano,
W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998).
Crystallography and NMR system: a new software
suite for macromolecular structure determination.
Acta Crystallogr. Sect. D, 54, 905–921.
48. Navaza, J. (1994). AMORE—an automated package
for molecular replacement. Acta Crystallogr. Sect. A, 50,
157–163.
49. Kleywegt, G. J., Henrick, K., Dodson, E. J. & van
Aalten, D. M. F. (2003). Pound-wise but pennyfoolish—how well do micromolecules fare in macromolecular refinement? Structure, 11, 1051–1059.
50. Lovell, S. C., Davis, I. W., Bryan Arendall, W., III, Paul,
I. W., de Bakker, J., Word, M. et al. (2003). Structure
validation by C-alpha geometry: phi, psi, and C-beta
deviation. Proteins Struct. Funct. Genet. 50, 437–450.
51. Dubochet, J., Adrian, M., Chang, J.-J., Homo, J.-C.,
Lepault, J., McDowall, A. W. & Schultz, P. (1988).
Cryo-electron microscopy of vitrified specimens. Q.
Rev. Biophys. 21, 129–228.
52. Adrian, M., Dubochet, J., Fuller, S. D. & Harris, J. R.
(1998). Cryo-negative staining. Micron, 29, 145–160.
53. Ludtke, S. J., Baldwin, P. R. & Chiu, W. (1999). EMAN:
semiautomated software for high-resolution singleparticle reconstructions. J. Struct. Biol. 128, 82–97.
54. Frank, J. (2006). Multivariate statistical analysis and
classification of images. In Three-imensional Electron
Microscopy of Macromolecular Assemblies (2nd edit),
pp. 145–192, Oxford University Press, New York.
55. Crowther, R. A. & Amos, L. A. (1971). Harmonic
analysis of electron microscope images with rotational
symmetry. J. Mol. Biol. 60, 123–130.
56. Frank, J., Radermacher, M., Penczek, P., Zhu, J., Li, Y.,
Ladjadj, M. & Leith, A. (1996). SPIDER and WEB: processing and visualization of images in 3D electron microscopy and related fields. J. Struct. Biol. 116, 190–199.
57. Penczek, P., Radermacher, M. & Frank, J. (1992). 3Dimensional reconstruction of single particles
embedded in ice. Ultramicroscopy, 40, 33–53.
58. Böttcher, B., Wynne, S. A. & Crowther, R. A. (1997).
Determination of the fold of the core protein of
hepatitis B virus by electron cryomicroscopy. Nature,
386, 88–91.