| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
* Department of Biological Sciences,
Department of Microbiology, National University of Singapore, Singapore 117543
1Correspondence: Department of Biological Sciences, National University of Singapore, 10, Kent Ridge Crescent, Singapore 117543. E-mail: dbsdjl{at}nus.edu.sg
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
secretion with different potency. LAL assay revealed that at least two
molecules of S1 bind cooperatively to one LPS molecule, with Hill’s
coefficient of 2.42. The LPS binding by S3 is independent and
noncooperative. The modified S
1 and S3 peptides exhibited
increased LPS neutralization potential although its LPS binding
affinities indicated only a 10-fold improvement. Hence, the structural
difference of the four sushi peptides conferred different efficiencies
in LPS neutralization without altering their binding affinity for LPS.
Circular dichroism spectrometry revealed that the four peptides
underwent conformational change in the presence of lipid A,
transitioning from a random coil to either an -helical or ß-sheet
structure. Two factors are critical for the sensitivity of Factor C to
LPS: 1) the presence of multiple binding sites for LPS
on a single Factor C molecule; and 2) high positive
cooperativity in LPS binding. The results showed that in the design of
an improved LPS binding and neutralizing peptide, charge balance of the
peptide is a critical parameter in addition to its structure.—Tan,
N. S., Ng, M. L. P., Yau, Y. H., Chong, P. K. W., Ho, B., Ding, J. L. Definition of endotoxin binding
sites in horseshoe crab Factor C recombinant sushi proteins and
neutralization of endotoxin by sushi peptides.
Key Words: Limulus amebocyte lysate • lipid A • Drosophila S2 cells • LPS
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. The neutralization of LPS-mediated toxic injury has
been considered for a long time as a possible therapeutic target in
patients. In LPS molecules, three genetically, biochemically, and
antigenically distinct regions are present: O-specific polysaccharide
chain, the core oligosaccharide, and the lipid A. The O-specific chains
are composed of highly variable repeating oligosaccharide units, are
immunogenic, and give rise to high number of serotypes. When the
O-antigen is lost, the core glycolipid is exposed. Although
structurally less varied than the O-antigen, it is still responsible
for the various chemotypes found in many Enterobacteriaceae, for
example, Salmonella spp (2)
. The
proinflammatory bioactivities of LPS reside in the most structurally
conserved glucosamine-based phospholipid known as lipid A
(3)
. Consequently, neutralization of endotoxin via lipid A
represents an important aspect of a logical, multifaceted approach to
treating this complex clinical syndrome.
Trace amounts of LPS in solution can activate a coagulation
cascade found in Limulus amebocyte lysate (LAL). Three
serine protease zymogens—Factor C, Factor B, proclotting enzymes, and
one clottable protein, coagulogen—have been purified and
characterized. In the presence of LPS, the LPS-sensitive Factor C,
serine protease zymogen, is autocatalytically activated. The active
Factor C then activates zymogen Factor B to active Factor B, which
subsequently activates proclotting enzyme to clotting enzyme. The
resulting clotting enzyme converts soluble coagulogen, an invertebrate
fibrinogen-like substance, to an insoluble coagulin gel
(4)
. Being the initial activator of the clotting cascade,
Factor C functions as a biosensor that responds to LPS or lipid A. It
is conceivable that Factor C has an LPS binding region that exhibits
exceptionally high affinity for lipid A. Consequently, the LPS binding
domain derived from Factor C would bind and neutralize lipid A
biotoxicity and be cross-reactive to other bacterial LPS, hence its
application in immunotherapy for patients with gram-negative bacterial
septicemia.
Our laboratory has cloned the homologous Factor C cDNAs from
Carcinoscorpius rotundicauda (CrFC) (5)
. Factor
C is a novel mosaic protein with five sushi domains, an EGF-like, a
C-type lectin-like, and a serine protease domain. In addition to these
domains, a Cys-rich region and a Pro-rich region have also been found
in the NH2-terminal and COOH-terminal portions of
the H chains, respectively (Fig. 1
) (6)
. Recently, we have established that the
amino-terminal fragment of CrFC harbor multiple lipid A binding sites.
We have expressed and characterized a secreted amino-terminal region of
CrFC, termed SSCrFCES (7)
. This 38 kDa protein, which
represents the high-affinity LPS binding region of Factor C, exhibits
high positive cooperativity of binding to multiple lipid A molecules,
with a Hill’s coefficient of 2.2. SSCrFCES can inhibit
endotoxin-induced LAL coagulation reaction and suppress LPS-induced
cytokine [tumor necrosis factor (TNF-) and interleukin 8]
production by THP-1 and normal human peripheral blood mononuclear
cells. This region of Factor C, which consists of the cysteine-rich,
EGF-like, and three sushi domains of Factor C, protects
galactosamine-sensitized mice from LPS-induced lethality
(7)
. The sushi domain, also called the
ß2-glycoprotein I-like domain, has two
disulfide bonds. In this study we sought to further localize and assess
the multiple endotoxin binding sites via expression of smaller
functional sushi domains of SSCrFCES and synthetic peptides derived
from SSCrFCES. We measured the peptide-mediated inhibition of
LPS-induced LAL and suppression of LPS-induced TNF- secretion by
human THP-1 cells. The structure-activity relationship of the peptides
was studied using circular dichroism (CD) analysis. Finally, we
evaluated the protection provided by these peptides on
galactosamine-sensitized mice against a lethal endotoxin challenge.
|
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
was purchased from PharMingen (San Diego, Calif.). Cell Titer
96 AQueous for cytotoxic assay was from Promega (Madison, Wis.). Pure
GFP protein was purchased from Clontech (Palo Alto, Calif.).
Oligonucleotides were synthesized by Genosys Biotechnologies,
Inc. (Woodlands, Tex.). Enzymes for DNA manipulation and polymerase
reactions were purchased from NEB (Beverly, Mass.) and Boehringer
Mannheim (Mannheim, Germany). DNA purification kits used were from
Qiagen (Chatsworth, Calif.). Pyrogen-free water for making buffers was
from Baxter (Morton Grove, Ill.).
Peptides
Factor C-derived peptides were synthesized and purified by
Genemed Synthesis, Inc. (San Francisco, Calif.). The first
peptide, 'N'-GFKLKGMARISCLPNGQWSNFPPKCIRECAMVSS-'C',
corresponding to residue 171–204 at the NH2 terminus of
CrFC, is designated S1 (171–204). S1 has a molecular weight [MW] of
3758. The second peptide
('N'-HAEHKVKIGVEQKYGQFPQGTEVTYTCSGNYFLM-'C') corresponding to
residue 268–301 is designated S3 (268–301), with [MW] 3892. Two
lysine mutations were introduced to S1 and S3 resulting in to
S1(171–204177,179) [MW, 3727], and S3(268–301276,278)
[MW, 3962], respectively. These peptides were purified to >
95% purity.
Construction of secreted sushi::EGFP fusion protein
The 3 recombinant expression vectors are named
pAc5.1S123EGFP, pAc5.1S1EGFP, and pAc5.1S3EGFP containing sushi 123,
sushi 1, and sushi 3 domain of CrFC, respectively (Fig. 1A
).
The construction of clones in this study was facilitated by the advent
of a modified pEGFP-N1, involving the insertion of a secretory signal
(SS) upstream and in-frame with the enhanced green fluorescent protein
(EGFP) (8)
, termed pSSEGFP (7)
. The first
cloning step involved the polymerase chain reaction (PCR) amplification
of a 616 bp dsDNA fragment containing sushi 123 domains of CrFC. The
template for PCR was a recombinant pAc5.1/V5-HisA plasmid carrying a
full-length CrFC, pAc5.1/CrFC. The blunt-end PCR product was introduced
into pBluescript II SK (+) for propagation, after which the sushi
fragment was cleaved and inserted whole (i.e., Sushi 123) or in part
(i.e., Sushi 1 and Sushi 3) into pSSEGFP. The sushi domain(s) were
inserted in-frame between SS and EGFP. The tripartite construct was
then transferred into pAc5.1/V5-HisA via the EcoRV and
NotI sites. The start and stop codons are located in the SS
and EGFP, respectively.
Stable expression of sushi::EGFP fusion protein in
Drosophila S2 cells
Expression constructs and the selection vector pCoHygro were
purified via EndoFree Plasmid Kits. The vectors were introduced into
Drosophila S2 cells (9)
by calcium phosphate
coprecipitation method (10)
. Drosophila S2
cells were routinely maintained in DES Expression medium supplemented
with 10% FBS. The cells were incubated at 25°C in a humidified
incubator. For transfection, 3–6 x 106
cells were introduced with 20 µg of plasmid DNAs at a ratio of 19
pAc5.1Sushi-EGFP:1 pCoHygro. After transfection, cells were maintained
in calcium phosphate solution for 24 h. Subsequently, cells were
washed twice with complete medium to remove the calcium phosphate
solution and allowed to recover for 2 days. Selection for stable cell
lines was performed by the addition of 350 µg/ml of insect-cell
tested hygromycin (Gibco, BRL) over a period of 3 wk. Stable cell lines
expressing fusion protein were adapted to serum-free medium by weaning
over 3–4 passages. To test for recombinant protein expression, 30 µl
of medium were electrophoresed on a 12% sodium dodecyl
sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and proteins were
electrophoretically transferred to PVDF membrane (Millipore, Bedford,
Mass.). The recombinant sushi-EGFP fusion proteins were detected using
anti-GFP antibody (Clontech) as described by the manufacturer.
Purification of sushi::EGFP fusion proteins from culture
medium
Stable recombinant Drosophila cells secreting high
levels of sushi123, sushi1, and sushi3 were grown in serum-free medium
to generate conditioned medium containing the sushi::EGFP
recombinant proteins. The sushi::EGFP fusion proteins were
purified via anion exchange chromatography on the ÄKTA explorer
(Pharmacia, Uppsala, Sweden). Typically, 2 l of conditioned medium
were initially subjected to successive ultrafiltration using a 100 kDa
and 10 kDa cutoff with the Pellicon system (Millipore). The 10-fold
concentrated medium was mixed with an equal volume of 0.2 M HEPES, pH
6.7. Fifty milliliters of the buffered medium was injected into a 60 ml
Q-Sepharose Fast Flow column equilibrated with 0.1 M HEPES, pH 6.7 via
a 50-ml Superloop (ÄKTA, Pharmacia) at a flow rate of 2 ml/min.
The bound proteins were subsequently washed with 5 column volumes of
the same buffer. Bound proteins were eluted with a 600 ml linear
gradient of NaCl (0–1.0 M) in 0.1 M HEPES, pH 6.7. Fractions
containing the fusion protein was identified by SDS-PAGE. These
fractions were pooled, desalted, and concentrated using Centriprep 10
(Millipore) at 1500 g for 30 min at 15°C. The
concentration of the recombinant proteins were quantitated by
densitometric scan of the Western blot against 100 ng of pure GFP
protein (Clontech).
Surface plasmon resonance (SPR) analysis of protein/peptide
interaction with lipid A
Recognition of lipid A by secreted sushi::EGFP fusion
proteins were performed with BIAcore 2000 biosensor instrument using
HPA sensor chip. Briefly, 100 µl of lipid A at 0.1 mg/ml in
phosphate-buffered saline (PBS) was sonicated at 37°C for 2 min prior
to being immobilized to an HPA sensor chip (Pharmacia) according to the
manufacturer’s specification. In all experiments, pyrogen-free water
was used as the running buffer at a flow rate of 20, 50, or 100
µl/min. The binding response was measured as a function of time.
After injection of various concentrations of sushi::EGFP, a
solution of anti-GFP antibody, diluted in PBS to 400 µg/ml, was also
injected to cause a shift in SPR in order to further confirm that
sushi-EGFP protein binds to lipid A. For regeneration, 100 mM of NaOH
solution was injected for 5 min.
To investigate the role of disulfide bonds in LPS binding affinity of
sushi::EGFP proteins, the proteins were treated with 1 mM DTT
for 20 min at 50°C to reduce the disulfide bonds. The DTT was
subsequently removed via ultrafiltration through Microcon 10
(Millipore) at 10,000 g for 30 min at 15°C. The reduced
linearized polypeptides were used for SPR studies as described above.
As a positive control, the binding constant of polymixin B was
determined using the same experimental conditions. The affinity
constant was calculated using BIAevaluation version 3.0.2 and
reconfirmed using CLAMP (11)
. The mean values were
obtained from three independent experiments.
Circular dichroism analysis
CD spectra were recorded in water, 50% trifluoroethanol and in
small unilamellar vesicles (SUV). SUV composed of phosphatidylcholine
and lipid A in molar ratio of 3:1 was prepared by sonication
(12)
. CD experiments were carried out using a Jasco-J-715
CD spectropolarimeter. Spectra were recorded in quartz cell cuvettes of
0.1 cm path length. The parameters used were band width = 2 nm;
step resolution = 0.5 nm; response = 1 s; scan
speed = 10 nm/min; scan width = 320–180 nm. The temperature
within the sample chamber was maintained at 25°C with a continuous
nitrogen flow rate of 5 l/min. Minor contributions of circular
dichroism and scattering from small unilamellar vesicles were
eliminated by subtracting lipid spectra of the corresponding
peptide-free suspension. Calibration was carried out with D-camphor
sulfonic acid.
LAL inhibition assay for determining the ENC50
This assay uses the initial part of the LAL endotoxin reaction
to activate an enzyme, which in turn releases p-nitroaniline from a
synthetic substrate, producing a yellow color. Throughout the assay,
the absorbance at 405 nm of each well of the microplate was monitored.
The absorbance reading from LAL reagent alone was used as blank. The
50% endotoxin-neutralizing concentration (ENC50)
reflects the potency of peptides; a low ENC50
indicates high potency.
Briefly, 25 µl of endotoxin solution at 10 EU/ml was mixed with an equal volume of peptides at various concentrations in LAL reagent water in disposable endotoxin-free glass dilution tubes (BioWhittaker) and incubated at 37°C for 30 min. The reaction mixtures were each carefully dispensed into the appropriate wells of an endotoxin-free microtiter plate (Costar, Cambridge, Mass.). Fifty microliters of freshly reconstituted LAL reagent was dispensed into the wells using an 8-channel multipipettor. The absorbance at 405 nm of each well of the microtiter plate was monitored after 45 min. The concentration of peptides corresponding to 50% inhibition was designated ENC50. The mean values were obtained from three independent experiments.
Hill’s plots were performed by graphing log10
peptide concentration against log10
[(Y)/(1.0–Y)], where Y was the fractional inhibition of procoagulant
activity observed in the chromogenic assay (13)
. Y equaled
the percent inhibition divided by 100. Thus, a Y of 0.75 corresponded
to 75% inhibition of procoagulant activity.
Suppression of LPS-induced TNF- secretion in human THP-1 cells
All cell lines were grown at 37°C in a humidified environment
in the presence of 5% CO2. THP-1 cells were
grown in RPMI 1640 medium supplemented with 10% FBS, penicillin (100
U/ml), and streptomycin (100 µg/ml). The cells were maintained at a
density of 2.5 x 105–6 cells/ml. THP-1
cells were prepared for experiment by addition of a concentrated
stock solution of PMA (0.3 mg/ml in dimethyl sulfoxide) to cell
suspension to give a final concentration of 30 ng/ml PMA and 0.01%
dimethyl sulfoxide (14)
. PMA-treated cell suspensions were
immediately plated into 96-well microtiter plate at a density of 4 x 105 cells/ml and allowed to differentiate for
48 h at 37°C. Immediately before stimulation by 10 ng/ml LPS or
LPS preincubated with various concentrations of peptides, the culture
medium was removed; the cells were washed twice with serum-free RPMI
1640 and incubated at 37°C. At indicated times, the culture medium
was collected. Human TNF- concentrations in the supernatants were
assayed using ELISA as suggested by the manufacturer. The difference
between the test and control groups was subjected to Student’s
t test. The values were obtained from at least three
independent experiments.
Cytotoxicity of peptides in eukaryotic cells
2 x 104 THP-1 monocytes in 50 µl
of RPMI 1640 were mixed in a microtiter plate with 50 µl of twofold
serial dilutions of peptides ranging in concentrations from 1.25 to 320
µM in PBS and incubated for 60 min at 37°C. To determine
cytotoxicity induced by the peptides, 20 µl of CellTiter96
AQueous One Solution Reagent (Promega) was added
into each well for 90 min at 37°C.
[3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium
(MTS) is bioreduced by metabolically active cells into a colored
formazan product that is soluble in tissue culture medium (15
, 16)
. For detection, the absorbance was measured at 490 nm. To
determine the ratio of cell lysis induced by peptides, two controls
were used. Complete lysis (100%) was achieved by incubating cells in
PBS containing 0.2% Tween-20 instead of medium only. This absorbance
value corresponded to the background, as those cells could not
metabolize MTS. The second control representing 0% lysis was
determined by incubating cells in medium only. The
LD50 was calculated as the concentration of
peptides necessary to lyse 50% of the cells. The experiment was done
in triplicate.
LPS-induced lethality in galactosamine-sensitized mice.
Mice are typically resistant to endotoxin. However, the
sensitivity of mice to endotoxin can be enhanced > 1000-fold by
coinjection with a liver-specific inhibitor, galactosamine
(17)
. In our study, intraperitoneal (i.p.) injection of 2
ng of E. coli 055:B5 LPS together with 15 mg of
galactosamine hydrochloride (Sigma) in 0.2 ml of saline induced 100%
lethality in C57BL/6J mice (18–25 g) within 7 h. Seventy-five
micrograms of peptides were preincubated with 2 ng of LPS at 37°C for
30 min prior to i.p. injection together with 15 mg
galactosamine-sensitized mice. Lethality was observed at time intervals
over 72 h after injection. Statistics were performed using the
Kaplan-Meier test (18)
.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
75% transformants. Stable
recombinant Drosophila cell lines were obtained and the
distribution of the protein was identified using Chemiluminescent
Western blot. The result revealed the presence of secreted recombinant
proteins in the culture medium. The sizes of sushi123-EGFP,
sushi1-EGFP, sushi3-EGFP, and control EGFP corresponded to the expected
calculated sizes of 49, 35, 35, and 27 kDa, respectively (Fig. 1B
).
Purification of sushi::EGFP fusion proteins using anion
exchange chromatography
Production of recombinant proteins in serum-free medium
drastically reduces the amount of contaminating proteins during
purification. Initial batchwise ion exchange purification of the three
proteins revealed that at pH 6.7, all three sushi::EGFP
fusion proteins bind to the anion exchanger, along with several high
molecular weight contaminants of similar size. Consequently, the 100
kDa and 10 kDa ultrafiltration steps were used to reduce the amount of
such contaminants. The partially enriched medium was injected into a
Q-Sepharose anion exchange column and the bound proteins were
subsequently eluted over a salt gradient extended over 10 column
volumes. This resulted in a 60–80% purification of the protein of
interest. A Coomassie-stained SDS-PAGE of partially purified sushi123,
sushi1, and sushi 3 is shown in Fig. 1C
.
LPS binding regions of Factor C reside in its sushi domains
To rapidly determine the Kd values of several
proteins for both 4'-monophosphoryl- and diphosphoryl lipid A from
different species, five different concentrations of proteins in water
were injected across lipid A-coated HPA chip. To avoid mass transport
problems, a low-capacity coated surface was used in addition to high
flow rates of 50 and 100 ml/ml. Injection of the sonicated lipid A over
the HPA sensor chip resulted in a low-capacity coated surface as
indicated by an increase of 322 resonance unit (RU). Figure 2A
shows a binding profile of sushi123-EGFP to E.
coli 1,4'-diphosphoryl lipid A. Unique biphasic association and
dissociation profiles were observed, suggesting the presence of
multiple lipid A binding sites. Figure 2B
shows a typical
binding profile of sushi1-EGFP to E. coli 1,4'-diphosphoryl
lipid A. The Kd values for binding of sushi-EGFP
proteins to various lipid A are illustrated in Table 1A
. At least one LPS binding site resides in each of sushi 1
and 3 domains. It is interesting to observe that sushi1 associates with
lipid A 10-fold slower
(kass=3.7x104
M-1s-1) than sushi3
(kass=1.5x105
M-1s-1). However, sushi1
remains associated with lipid A by 100-fold
(kdiss=5.3x10-6
s-1) stronger than sushi3
(kdiss=5.9x10-4
s-1). Comparatively, the initial interaction of
sushi123 with lipid A exhibited a kinetic profile similar to that of
sushi3. The cooperativity effect prolongs the time that lipid A
interacts with the second LPS binding site in sushi123
(kdiss2=6.7x10-7
s-1). The binding constant of reduced linearized
sushi1 and sushi3 proteins displayed a 10,000-fold decrease in
Kd value (Table 1)
. The binding constant of control
peptide polymixin B for lipid A was determined to be 7.1 x
10-7 M, which is consistent with previously
published values (19)
.
|
|
Interaction of S1, S1, S3, and S3 peptides with E.
coli lipid A
Two potential LPS binding regions, S1 and S3, residing in sushi 1
and sushi 3 domains, respectively, were synthesized as 34-mer peptides.
The LPS binding potentials retained by such synthetic peptides were
again verified using SPR analysis (Table 1B)
. It is clear that in these linear peptides, the absence of structural
architecture conferred by disulfide bonds leads to an apparent decrease
in affinity for LPS by of 10,000-fold in S1 and 100-fold for S3 (Table 1B)
. The mutated peptides S1 at best only resulted in a 10-fold
increase in its affinity of lipid A, whereas no improvement could be
observed for S3. Nonetheless, we have localized two LPS binding
motifs to 34 residue regions in Factor C. These four peptides were used
for further studies.
|
Conformational studies
The CD studies revealed that S1 and S3 adopt a random coil
conformation in water. The mutated peptides S1 and S3 also
conformed largely to random coil structures, with only 4.7% -helix
and 28.7% ß-sheets unless structure-promoting exogenous elements
were also present. These include a simple solvent like 50%
trifluoroethanol or complex membrane where anionicity simulates that of
bacterial membrane, such as phosphatidylcholine:lipid A liposomes.
-Helical structure was induced in S1 (43%) and S3 (46.5%) upon
interaction with 0.75 nM lipid A even though the lipid A concentration
used was clearly below the critical micelle concentration (Table 2
). The introduction of two lysine residues to S1 caused a decrease in
-helical structure from 43% to 16%, with a concomitant increase in
ß-sheets (10.8%) resulting in a 10-fold increase in lipid A binding.
Similar mutations to S3 caused a complete structural changeover from a
highly ß-sheeted structure (90.6%) to a helical conformation
(46.5%), without any significant difference in affinity constant.
|
Inhibition of endotoxin-induced LAL reaction
We used a sensitive and precise Limulus chromogenic
assay to examine the ability of the four peptides to bind 10 EU of
endotoxin per ml (LPS, E. coli 055:B5). The 50%
endotoxin-neutralizing concentration (ENC50) is
taken as the concentration of peptides that reduces the absorbance by
50%. A low ENC50 indicates high potency of
endotoxin neutralization. The ENC50 values of the
four peptides were determined to be S1 (2.25 µM), S3(1 µM),
S1(0.875 µM), and S3(0.625 µM). Figure 3A
shows that the LAL inhibition curves for the four peptides
are different. Whereas the S1 peptide binding isotherm is sigmoidal,
the S1, S3, and S3 peptides are not. Because sigmoidal curves
suggest cooperativity, we also graphed that data as a Hill’s plot
(Fig. 3B
). By Hill’s plot analysis, the S1 peptide had a
linear slope of 2.42 (r=0.98), suggesting positive
cooperativity between at least two peptide molecules and a single LPS
molecule. In contrast, the LPS binding event in S1, S3, and S3
occurred independently rather than cooperatively.
|
Factor C-derived peptides inhibit LPS-stimulated hTNF- release
from THP-1 cells
Results from our in vitro binding studies suggested
that the four Factor C-based peptides would be potent inhibitors of the
LPS activation of monocytes. To test this prediction, we measured the
ability of S1, S1, S3, and S3 to inhibit hTNF- production by
THP-1 cells incubated with 10 ng/ml of LPS in a serum-free system
containing various concentrations of peptides. As shown in Fig. 4
, both modified S1 and S3 are more potent inhibitors than their
corresponding parental peptides, S1 and S3. S1 and S3 require
53.3 and 45.8 µM, respectively, to achieve >50% inhibition of
LPS-induced production of TNF-. Twice as much of S3 (94 µM) as
S3 (45.8 µM) is required to achieve 50% inhibition. S1 could not
achieve 50% inhibition with the range of concentrations tested.
|
S1, S1, S3, and S3 showed minimal cytotoxicity to eukaryotic
cells
The toxicity of S1, S1, S3, and S3 to mammalian cells was
tested by incubation with human monocytes and analyzed using MTS, a
compound that could only be metabolized by healthy cells. S1, S1,
and S3 had minimal effect on cell permeabilization (data not shown). At
the highest concentration of 320 µM, only 5–10% cell lysis was
observed. Comparatively, the most potent LAL assay inhibitor, S3,
causes a 25–30% cell lysis.
Effects of peptides on endotoxin-induced lethality in
galactosamine-sensitized mice
Injection (i.p.) of 2 ng of E. coli 055:B5 LPS per
mouse induced 100% lethality in the galactosamine-sensitized mice
within 7 h. Preincubation of 75 µg each of S1, S1, or S3 with
LPS for 30 min prior to i.p. injection resulted in a 20–55%
protection against LPS-induced lethality (Fig. 5
). Kaplan-Meier plot clearly shows a prolonged survival of the mice
receiving LPS together with S1, S1, or S3 compared to LPS alone. No
significant differences were observed among them. However, 75 µg of
S3 was sufficient to confer 100% protection to mice. S3 is
significantly more effective in protection as compared to S1, S1,
and S3 as analyzed using Kaplan-Meier analysis and log rank pairwise
test (P<0.05).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Although much work has been reported on the
endotoxin-activated coagulation cascade (4
, 21
, 22)
, no
effort was made to understand or explain the sensitivity of Factor C to
trace amounts of LPS. Such studies have been plagued by two major
difficulties: 1) the capacity to produce biologically active
recombinant Factor C; 2) the uncoupling of LPS binding from
subsequent autocatalytic activation. In an earlier study we were able
to express a 38 kDa amino-terminal fragment of Factor C that represents
the LPS binding domain, termed SSCrFCES. A novel SS was used to direct
the secretion of SSCrFCES into the culture supernatant of
Drosophila cells, and hence it was readily purified to
homogeneity as a stable monomeric protein. SSCrFCES exhibits high
positive cooperativity of binding to two or three lipid A molecules,
with a Hill’s coefficient of 2.2 (7)
.
To delineate the LPS binding region, smaller fragments of SSCrFCES were
subcloned and expressed as EGFP fusion proteins (Fig. 1A
).
Expression technology similar to that used to produce functional
SSCrFCES was used in this study to express sushi::EGFP fusion
proteins. The recombinant sushi::EGFP fusion proteins were
produced as fully functional secreted products. Purification of
sushi::EGFP proteins from serum-free medium via anion
exchange chromatography resulted in 60–80% purity.
Several precautions were taken to avoid mass transport limitations and
related inhomogeneities within the sensor chip during SPR studies.
First, injection of sonicated lipid A solution over the HPA sensor chip
resulted in an overall increase in 322 RU. The sonication step was
designed to achieve a more homogeneous sample. The use of low-capacity
coated surface decreases the demand for proteins, thus minimizing
concentration gradients in the flow cell. Second, high flow rates of 50
and 100 µl/ml were used to increase the transport rate of protein to
the surface and within the lipid A-coated layer. SPR studies indicate
that sushi 1 and 3 domains each harbor an LPS binding site. It is
evident that sushi 1 (Kd=1.4x10-10 M)
has a 10-fold higher affinity for lipid A than sushi 3
(Kd=3.9x10-9 M). Positive
cooperativity in sushi 123 was reflected in a 1000-fold increase in
affinity for lipid A
(Kd2=2.7x10-12 M). Sushi 123
showed comparable affinity constant to SSCrFCES. A similar phenomenon
was observed in all the diphosphoryl-lipid A we tested (Table 1A)
.
Comparatively, sushi 1 and 3 domains have lipid A binding affinities
that are matched by only three other reported LPS binding proteins,
namely, bactericidal/permeability increasing protein (BPI)
(23)
, Limulus anti-LPS factor
(24)
, and LPS binding protein (25)
(Table 3
). However, sushi 123, which contains at least two LPS binding sites
that show positive cooperativity, exhibits affinities that are
unparalleled by any known LPS binding proteins to date. Its
Kd2 value is only 10-fold weaker than
streptavidin–biotin interaction. Among the three proteins that showed
comparable affinities to sushi 1 and 3, only BPI has two LPS binding
sites; nevertheless, it has not been reported to exhibit positive
cooperativity. Contrary to the 1 Factor C:1 lipid A stoichiometric
ratio reported by Nakamura et al. (26)
, we have proved
that Factor C protein harbors at least two LPS binding sites. It is
also interesting to note that binding to 4'-monophosphoryl-lipid A by
the three sushi recombinant proteins decreased by 10- to 100-fold. This
was translated into 2.8-fold higher concentration (nM) required for
50% activation of Factor C as compared to its diphosphoryl-lipid A
counterpart (26)
.
|
Closer examination of the amino acid sequences within sushi 1 and 3
domains revealed two possible LPS binding regions. Consequently, two
34-mer peptides were synthesized: S1 (171–204) and S3 (268–301).
Again, SPR was used to rapidly verify their LPS binding potential.
Linear S1 and S3 peptides exhibited a 10,000- and 100-fold decrease,
respectively, in their affinity for LPS as compared to their
corresponding recombinant proteins. In accordance with other LPS
binding protein-derived peptides, a significant decrease in affinity
for LPS was also reported (Table 3)
. In this study, we show that S1 and
S3 peptides are able to bind LPS with varying LPS-neutralizing
potencies. Both S1 and S3 inhibit LPS-induced LAL activity. Chromogenic
Limulus assay revealed that S1 binds to LPS; this binding
showed positive cooperativity between at least two molecules of S1 and
one molecule of LPS (Hill’s coefficient=2.42). This apparent
cooperativity even with linear peptides is surprising, but offers
another strata of sensitivity of Factor C to LPS unmatched by any known
protein. A Hill’s plot of S3 has a coefficient of 0.99, indicating
simple and noncooperative binding to LPS. Suppression of LPS-induced
hTNF- secretion from THP-1 by S1 and S3 revealed surprising results.
Although SPR studies show that S3 has a 10-fold stronger affinity for
LPS than S1, the latter can only weakly inhibit LPS-induced hTNF-
production; 50% inhibition of TNF- production could not be reached
with the indicated range of concentrations used. At 100 µM of S1, a
mere suppression of 20% was observed. The main reason is the
‘sequestering’ of a single LPS molecule by at least two S1. S3 can
inhibit 50% hTNF- production at 94 µM. This represents a quantum
of 9500-fold excess of S3 compared to SSCrFCES (7)
.
In an attempt to increase LPS binding affinities of S1 and S3 peptides,
two amino acids in each peptide were replaced with lysine residues and
termed S1 and S3, respectively. These mutations were based on
computational analysis that an increase in lysine residues at specific
sites may improve LPS binding (27)
. SPR studies revealed
that these mutations did not improve S3 affinity, but decreased the
Kd values of S1 by only 10-fold. However, LAL
inhibition assay clearly showed that 2.6- and 1.6-fold less of S1
and S3, respectively, was required to achieve
ENC50. The inhibition binding profiles are
noticeably different from their parental peptides. Graphing S1 data
as a Hill’s plot showed that the increase in LPS binding resulted in a
lower Hill’s coefficient of 1.08, suggesting that methionine and
arginine residues are involved in determining cooperativity efficiency.
It is logical to conclude that the 10-fold increase in LPS binding
affinity of S1 resulted in a lower ENC50. In
addition, it is conceivable that the stoichiometric ratio of 1:1 rather
than 2:1 peptide to lipid A molecule would lead to a more effective
inhibition. Further studies on S1 and S1 using nuclear magnetic
resonance would probably reveal how these two amino acids interact with
LPS molecule. S3 showed a more typical hyperbolic binding curve, and
the Hill’s coefficient of 0.91 suggest that its binding to LPS was
simple and noncooperative. Since no differences in LPS binding affinity
(Kd for S3=5.1x10-7 M;
Kd for S3=6.6x10-7 M)
were observed for S3 and S3, the differences in potency to inhibit
LPS-induced LAL activity are likely to be attributed to two factors:
1) the different structural conformations of S3 and S3;
2) the presence of two overlapping pseudo-receptor sites for
LPS in S3 (KVKIKVK). As expected, both S1 and S3
have increased efficacy in suppressing LPS-induced hTNF- production.
The peptides S1 and S3 also conform to a largely random coil
structure, with only 4.7% -helix and 28.7% ß-sheet,
respectively. In the presence of phosphatidyl choline:lipid A
liposomes, a complex membrane whose anionicity simulates that of
bacterial membrane, mutation of S1 to S1 exhibited a decrease in
-helical structure from 43% to 16%, with a concomitant increase in
ß-sheets (10.8%) resulting in a 10-fold increase in lipid A binding.
Similar mutations in S3 resulted in a complete changeover from a highly
ß-sheeted structure (90.6%) to an -helical conformation (46.5%),
without any significant difference in affinity constant. However, the
apparent structural differences in the four sushi peptides conferred
different efficiencies in neutralizing LPS biotoxicity, since mutation
in the S3 peptide, which resulted in a complete structural changeover,
showed a marked improvement in suppressing LPS-induced TNF-
production. However, the structure of the sushi peptides does not
improve binding to LPS, as reflected by the similar affinity constant.
It would appear that Factor C protein possesses LPS binding domains
with both -helical and ß-sheets, where positive cooperativity
occurs between these two regions. This is the first report of two
peptides with comparable Kd values that adopt a
completely different conformation in association with LPS, which
translated into different potencies in neutralizing LPS biotoxicity.
We also showed a protective effect of S1, S1, S3, and S3 against
LPS-induced septic shock in mice. Mice receiving LPS preincubated with
these peptides, especially S3, clearly demonstrated attenuation of
lethality. This indicates a beneficial intervention on parameters that
determine long-term survival and may provide a window of time for other
therapeutic support. The possibility of a peptide being able to protect
against the severe clinical symptoms of LPS-induced septic shock is a
promising development. The next step would be the development of
peptoids (peptide mimics) that are resistant to degradation by
proteases in vivo, thus creating an anti-endotoxin peptoid
with prolonged half-life.
The display of properties such as LPS binding, neutralization, and
suppression of LPS-induced hTNF- production clearly indicate that
the linear 34-mer S1 and S3 peptides mimic, albeit with less
efficiency, the properties exhibited by the larger recombinant fragment
of Factor C, SSCrFCES. Consequently, it is likely that mutations
performed on these peptides would be reflected in a similarly mutated
SSCrFCES. Sushi proteins devoid of disulfide bonds were still capable
of binding to lipid A albeit with a 10,000-fold reduction in affinity.
This indicates that a rigidly held disulfide-bonded structure present
in sushi domains is important but may not be absolutely essential for
LPS binding activity. The results also suggest that the maintenance of
peptide negative-positive charge balance is a critical parameter, in
addition to its structure, in the design of an improved LPS binding and
neutralizing peptide. The modified S1 and S3 peptides also
exhibit increased LPS neutralization potential, although their LPS
binding affinities derived from SPR studies at best indicated only a
10-fold improvement. Although it is tempting to attribute this
improvement in LPS-neutralizing potency of S1 and S3 to the
replacement of specific amino acids by lysine residues, as suggested by
Hong et al. (27)
, one cannot ignore the possibility that
mutations to other residue of the peptide might elicit the same effect
observed here. Thus, the anti-endotoxin property of a peptide is also
affected by the peptide:lipid A stoichiometric ratio and does not
necessarily correlate with increased affinity for LPS. The
characterization of the minimal endotoxin binding motif of Factor C has
provided a basis for designing small molecules that could have
prophylactic and/or therapeutic properties in humans for the management
of septic shock.
The results from this study show that the sensitivity of Factor C for
LPS stems from two ratiocinations: 1) the presence of
multiple binding sites for LPS on a single Factor C molecule and
2) high positive cooperativity in LPS binding. Although any
individual LPS binding sushi domain of Factor C has affinity constants
comparable to other reported LPS binding proteins, this attribute is
not sufficient to account for the extreme sensitivity of Factor C to
LPS. SPR studies indicate that sushi 1 and 3 domains each harbor unique
LPS binding sites. It is evident that sushi 1
(Kd=1.4x10-10 M) has a
10-fold higher affinity for lipid A than sushi 3
(Kd=3.9x10-9 M). An
advantage of SPR is that the kinetic rate of association and
dissociation constants can be calculated. LPS tends to bind to sushi 3
(kass=1.5x105
M-1s-1) 10-fold faster
than to sushi 1
(kass=3.7x104
M-1s-1). However, the
sushi3:LPS complex
(kdiss=5.9x10-4
s-1) dissociates 100-fold faster than the
sushi1:LPS complex
(kdiss=5.3x10-6
s-1). Sushi 123 showed an affinity constant
comparable to SSCrFCES (7)
. Positive cooperativity
among the multiple LPS binding sites in sushi 123 was reflected in
a 1000-fold increase in affinity for lipid A
(Kd2=2.7x10-12 M). This
Kd2 value is only 10-fold weaker than the
streptavidin–biotin interaction. The kinetic rate constant calculated
for sushi123 revealed some mechanistic details of LPS binding to Factor
C. It is clear that the association and dissociation of sushi123
(kass1=1.6x105
M-1s-1;
kdiss1=2.7x10-4
s-1) are very similar to those of sushi3 alone.
Presumably, the effect of LPS binding to sushi3 region of sushi123
prolongs the dissociation time of another LPS molecule from sushi1 by
1000-fold (kass2=2.5x105
M-1s-1;
kdiss2=6.7x10-7
s-1), with a concomitant 10-fold increase in
association rate. With these linear peptides, a 100- to 10000-fold
decrease in affinity was observed compared to their parental
recombinant proteins. Both S1 and S3 can inhibit LAL reaction and
LPS-induced hTNF- secretion with a different potency. The two
synthetic peptides mimic the LPS binding characteristics displayed in
SSCrFCES (7)
. Surprisingly, LAL assay revealed that at
least two S1 bind cooperatively to one LPS molecule, with a Hill’s
coefficient of 2.42. The LPS binding by S3 is independent and
noncooperative. This provides a new level of understanding of
LPS-induced Factor C activation.
In view of our current accumulated understanding of LPS binding and
activation of Factor C, we propose a novel two-prong
amplification-activation pathway in which endotoxin interacts with
Factor C. The schematic representation is as illustrated in Fig. 6
. Our model explains the sensitivity of Factor C to femtogram level of
endotoxin. We postulate that sushi 1 and 3 domains of Factor C play
different roles in its activation. Sushi 1 domain functions as an ‘LPS
epitope-presenting’ site, whereas the ‘LPS-capturing’ role of sushi
3 aids in increasing the affinity of sushi 1 by 1000-fold.
Consequently, a simple event involving the occupancy of a single LPS
molecule binding to Factor C protein would result in a novel activation
of the two-prong amplification pathways of Factor C. First, the binding
of LPS to either one of the LPS binding sites in sushi 1 or 3 domain
would activate Factor C. This is supported by the Kd
values calculated from separate SPR studies of sushi 1 and 3 domains,
which showed only a 10-fold difference in their affinity for LPS.
Furthermore, when both domains exist together, no difference in
association rate was observed. The singly bound LPS molecules in sushi
3 domain would result in a cooperative binding of a second LPS molecule
to the sushi 1, hence creating a new ‘nucleus’ for further activation
of Factor C. This is supported by our first observation that SSCrFCES
exhibits high positive cooperativity of binding to multiple lipid A
molecules, with a Hill’s coefficient of 2.2 (7)
. SPR
studies also support this fact with a remarkable display of a 1000-fold
increase in affinity. Second, our observation shows that sushi 1
‘nucleus’, which is actually a tightly bound LPS molecule playing an
LPS eptitope-presenting function, can exhibit cooperative binding to
similar LPS binding sites of other Factor C proteins. This is reflected
in the positive cooperativity of multiple S1 (Hill’s coefficient of
2.42) to one LPS molecule. Such cooperativity leads to rapid
amplification of activation of multiple Factor C proteins from just two
molecules of LPS, thus explaining Factor C’s ability to detect trace
levels of LPS in solution.
|
|
ACKNOWLEDGMENTS |
|---|
Received for publication September 28, 1999.
Revision received January 26, 2000.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. Gonzalez, Y. Gueguen, D. Destoumieux-Garzon, B. Romestand, J. Fievet, M. Pugniere, F. Roquet, J.-M. Escoubas, F. Vandenbulcke, O. Levy, et al. Evidence of a bactericidal permeability increasing protein in an invertebrate, the Crassostrea gigas Cg-BPI PNAS, November 6, 2007; 104(45): 17759 - 17764. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Li, B. Ho, and J. L. Ding Recombinant Factor C competes against LBP to bind lipopolysaccharide and neutralizes the endotoxicity Innate Immunity, June 1, 2007; 13(3): 150 - 157. [Abstract] [PDF]
|
|
|
|
 |
|
 T. Koshiba, T. Hashii, and S.-i. Kawabata A Structural Perspective on the Interaction between Lipopolysaccharide and Factor C, a Receptor Involved in Recognition of Gram-negative Bacteria J. Biol. Chem., February 9, 2007; 282(6): 3962 - 3967. [Abstract] [Full Text] [PDF]
|
|
). The formation of an ‘LPS-eptiope presenting’ nucleus allows
multiple LPS binding sites to interact cooperatively with this second
LPS molecule (
- 
), resulting in an ‘amplified’
multi-activation of numerous Factor C molecules. Incidentally, if the
initial LPS capture event happens to sushi 1 domain, multi-activation
of Factor C proteins would also occur. Hence, the sensitivity of Factor
C to femtograms of LPS is the result of multiple LPS binding sites and
high positive cooperativity. Activation of Factor C is
temperature-dependent as shown by Nakamura et al. (35)