Abbreviations CHO, Chinese hamster ovary; FITC, fluorescein isothiocyanate; IL, interleukin; Iris, Ixodes ricinus immunosuppressor; LPS, lipopolysaccharide; MCP-1, monocyte chemoattracta
Trang 1multifunctional serpin from the saliva of the tick
Ixodes ricinus
Pierre-Paul Prevot1, Alain Beschin2,3, Laurence Lins4, Je´roˆme Beaufays1, Ame´lie Grosjean5,6,
Le´a Bruys2,3, Benoıˆt Adam4, Michel Brossard5,6, Robert Brasseur4, Karim Zouaoui Boudjeltia5,6, Luc Vanhamme1,7,* and Edmond Godfroid1,*
1 Laboratoire de Biologie Mole´culaire des Ectoparasites, Universite´ Libre de Bruxelles, Gosselies, Belgium
2 Department of Molecular and Cellular Interactions, VIB, Brussels, Belgium
3 Laboratory of Cellular and Molecular Immunology, Vrije Universiteit Brussel, Belgium
4 Centre de Biophysique Mole´culaire et Nume´rique, Faculte´ Universitaire des Sciences Agronomiques de Gembloux, Belgium
5 Laboratoire de Me´decine Expe´rimentale, Universite´ Libre de Bruxelles, Montigny-le-Tilleul, Belgium
6 Institut de Zoologie, Universite´ de Neuchaˆtel, Switzerland
7 Parasitologie Mole´culaire, Institut de Biologie et de Me´decine Mole´culaires (IBMM), Universite´ Libre de Bruxelles, Gosselies, Belgium
Ticks are hematophagous arachnid ectoparasites,
sec-ond only to mosquitoes as pathogen vectors worldwide
[1] Ixodes ricinus is widely found in Europe and is
able to take its blood meal from a variety of
verte-brates, ranging from amphibians to mammals, and
including domestic animals and humans [2] It is the
one of the vectors of Borrelia burgdorferi, the agent of Lyme disease I ricinus is characterized by a long-last-ing blood meal, leavlong-last-ing time for its host to activate defensive reactions such as pain (stimulating scratch-ing) and hemostasis (repairing the wound and involv-ing coagulation), as well as innate, adaptive immune
Keywords
inflammatory; receptor binding domain;
sepsis; serpin; tick
Correspondence
E Godfroid, Laboratoire de Biologie
Mole´culaire des Ectoparasites, Institut de
Biologie et de Me´decine Mole´culaires
(IBMM), Universite´ Libre de Bruxelles, rue
des professeurs Jeener et Brachet 12,
B-6041 Gosselies, Belgium
Fax: +32 2 650 9900
Tel: +32 2 650 9934
E-mail: edmond.godfroid@ulb.ac.be
*These authors contributed equally to this
work
(Received 4 February 2009, revised 2 April
2009, accepted 3 April 2009)
doi:10.1111/j.1742-4658.2009.07038.x
Serine protease inhibitors (serpins) are a structurally related but function-ally diverse family of ubiquitous proteins We previously described Ixodes ricinus immunosuppressor (Iris) as a serpin from the saliva of the tick I ricinus displaying high affinity for human leukocyte elastase Iris also displays pleotropic effects because it interferes with both the immune response and hemostasis of the host It thus inhibits lymphocyte prolifera-tion and the secreprolifera-tion of interferon-c or tumor necrosis factor-a by periph-eral blood mononuclear cells, and also platelet adhesion, coagulation and fibrinolysis Its ability to interfere with coagulation and fibrinolysis, but not platelet adhesion, depends on the integrity of its antiproteolytic reactive center loop domain Here, we dissect the mechanisms underlying the inter-action of recombinant Iris with peripheral blood mononuclear cells We show that Iris binds to monocytes⁄ macrophages and inhibits their ability
to secrete tumor necrosis factor-a Recombinant Iris also has a protective role in endotoxemic shock The anti-inflammatory ability of Iris does not depend on its antiprotease activity Moreover, we pinpoint the exosites involved in this activity
Abbreviations
CHO, Chinese hamster ovary; FITC, fluorescein isothiocyanate; IL, interleukin; Iris, Ixodes ricinus immunosuppressor; LPS,
lipopolysaccharide; MCP-1, monocyte chemoattractant protein-1; PBMC, peripheral blood mononuclear cell; PDB, Protein Data Bank; RBD, receptor binding domain; RCL, reactive center loop; rIris, recombinant Iris; serpins, serine protease inhibitor; TNF-, tumor necrosis factor.
Trang 2reactions In response, ticks have evolved a battery of
countermeasures [3–7], mainly involving salivary
pro-teins, several of which are induced during the feeding
process and injected at the site of the wound Some of
these proteins have been identified in several tick
spe-cies [8–17] They comprise a variety of enzymes
inter-fering with different physiological processes We have
previously reported the construction and analysis of a
cDNA subtractive library which led to the
identifica-tion of I ricinus immunosuppressor (Iris), a protein
expressed in the salivary glands and secreted in the
saliva during the blood meal [18–20] Structural
analy-sis and site-directed mutageneanaly-sis confirmed that Iris is
a member of the serine protease inhibitor (serpin)
family The protein structure of serpins is characterized
by three b sheets (A, B and C) and eight or nine
a helices [21] A typical feature of serpins is the
reac-tive center loop (RCL), a protein motif of 20 amino
acids, located near the C-terminus of the protein This
motif contains a scissile bond between the so-called
residues P1 and P1¢, which is cleaved by the target
pro-tease This cleavage triggers structural rearrangement
of both the protease and the inhibitor in a suicide
mechanism that irreversibly complexes and inactivates
both interacting partners All these features have been
uncovered in Iris [21] However, because serpins are
involved in a wide variety of physiological processes,
prediction of the function of Iris in tick saliva based
solely on its belonging to this family was precluded
We previously expressed Iris as a recombinant
pro-tein and showed that it inhibits serine proteases such
as human leukocyte elastase, tissue plasminogen
acti-vator, thrombin and factor Xa Affinity rate constants
and inhibition values indicated that Iris preferentially
targets the human leukocyte elastase [21] In addition,
in agreement with its enzymatic activity, Iris was
shown to interfere with coagulation and fibrinolysis
[21] These effects were dependent on the protease
inhibitory function of the RCL domain of Iris, because
the Leu339Ala (P2) mutant devoid of antiprotease
activity did not influence coagulation or fibrinolysis
However, Iris also increases the platelet adhesion time
[21]; an activity not affected by the mutation This
sug-gests that domains other than the RCL are involved in
this distinct protein activity To add to the
multifunc-tionality of this protein, Iris has also been found to
modulate cytokine production by human peripheral
blood mononuclear cells (PBMC) [18] Leboulle et al
demonstrated that soluble protein extracts of Chinese
hamster ovary (CHO)-KI cells expressing recombinant
Iris (rIris) inhibited the production of tumor necrosis
factor (TNF-)a, interleukin (IL)-6, IL-8 and
inter-feron-c triggered by lipopolysaccharide (LPS)
Here, we further address the multifunctional charac-ter of Iris More precisely, we dissect its anti-inflamma-tory activity The results indicate that Iris inhibits TNF-a production by direct interaction with the monocyte⁄ macrophage populations of PBMC More-over, its ability to interfere with TNF-a production is independent of its antiprotease activity Finally, we also provide evidence that Iris, as an inhibitor of TNF-a production, may be used as a therapeutic tool against endotoxemic shock
Results
Iris inhibits TNF-a production by PBMC
We have previously reported that the tick salivary ser-pin Iris could interfere with the immune response Indeed, total soluble protein extracts from CHO-KI cells expressing rIris–6His inhibited the production of TNF-a by PBMC which had been activated by the Toll-like receptor (TLR)-4 trigger, LPS [18] In order
to dissect the mode of action of Iris, we purified large amounts of the protein expressed in a baculovirus sys-tem We first investigated the ability of Iris to suppress production of TNF-a by PBMC stimulated with LPS Because levels of TNF-a production by PBMC vary between healthy donors, prior experiments on multiple donors were performed to determine the appropriate rIris concentration range for these studies This range was found to be 10–400 nm (results not shown) In all experiments, dexamethasone was used as a control to inhibit LPS-induced TNF-a production [22] The results shown in Fig 1 (from three independent
*** ***
***
***
***
Fig 1 Effect of rIris on TNF-a production by LPS-stimulated human PBMC PBMC were activated by LPS (10 lgÆmL)1) for 4 h
in the presence or absence of the indicated rIris concentrations TNF-a concentrations were then measured in the harvested super-natants DEX, dexamethasone (10 l M) was used as positive control for inhibition of TNF-a production; NS, nonstimulated cells; LPS, LPS-stimulated cells Values are the mean of three experiments (± SD) *P < 0.05, **P < 0.01, ***P < 0.001 compared with controls (one-way ANOVA).
Trang 3experiments performed on one representative in five
donors using the appropriate concentration range)
indicate that the pure recombinant protein inhibited
TNF-a production by PBMC which had been
stimu-lated by LPS This effect was dose dependent, reaching
a maximum at 200–400 nm Iris Iris also impaired the
spontaneous release of TNF-a by PBMC (Fig 1)
Moreover, the ability of Iris to impair TNF-a
pro-duction could be inhibited by polyclonal anti-Iris
serum [20] Indeed, as shown in Fig 2, increasing
concentrations of anti-Iris serum progressively restored
TNF-a levels in the PBMC supernatant to the values
induced by LPS stimulation in the absence of Iris This
effect was presumably related to the specific
neutraliza-tion of Iris because the preimmune serum remained
without effect
Furthermore, Iris inhibited not only the production
of TNF-a by PBMC activated via the TLR4⁄ LPS pathway, but also, to varying extents, by peptidogly-can (PGN), poly(I : C) and ODN 2006, which are recognized by TLR2, TLR3 and TLR9, respectively (Table 1) Dexamethasone (10 lm) was used as a positive control and, as expected, inhibited TNF-a production by 95%
Together, these data show that Iris inhibited TNF-a production by PBMC and suggested that this activity was independent of the TLR stimulus
Iris binds to monocytes/macrophages
To investigate which PBMC subset population was the target of rIris, flow cytometry experiments were per-formed Fluorescent-labeled Iris did not bind on CD3) (T lymphocytes), or CD19) (B lymphocytes) positive cells (results not shown) By contrast, Iris binding was detected on both the CD11b+ CD14+ CD16+ and CD11b+ CD14+ CD16) monocyte subsets (Fig 3) Moreover, the binding of fluorescent-labeled Iris could
be competed out by co- or preincubation with anti-Iris serum but not with preimmune sera Furthermore, bind-ing of fluorescent-labeled Iris to monocytes could be abrogated by co- or preincubation with nonlabeled Iris, but not by co- or preincubation with LPS (not shown)
Inhibition of LPS-induced TNF-a release is independent of the antiprotease activity of Iris
We next asked whether the inhibiting effect of Iris on TNF-a released by LPS-activated PBMC was depen-dent on its serpin activity For that purpose, rIris, its cleaved elastase-complexed form or the mutant L339A (the latter two lacking antiprotease activity) were
***
***
**
Fig 2 Polyconal anti-rIris serum neutralize the capacity of Iris to
inhibit the release of TNF-a by LPS-activated PBMC PBMC were
incubated for 4 h with or without LPS (10 lgÆmL)1) in the presence
(+rIris) or absence ( )rIris) of 100 n M rIris that had been
preincuba-ted for 5 min at 37 C with the indicated dilutions of anti-Iris serum.
TNF-a concentrations were measured in the harvested
superna-tants ⁄ , no antibody added; PI, preimmune serum Values are the
mean of three experiments (± SD).
Table 1 Effect of recombinant Ixodes ricinus immunosuppressor (rIris) on production of tumor necrosis factor-alpha (TNF-a) by peripheral blood mononuclear cells (PBMC) stimulated via distinct Toll-like receptors (TLRs) PBMC (2 · 10 5 cellsÆwell)1) were incubated with the indicated chemicals [lipopolysaccharide (LPS), 10 lgÆmL)1; peptidoglycan (PGN), 10 lgÆmL)1; poly(I : C) 10 lgÆmL)1or ODN 2006, 2 lgÆmL)1]
in complete RPMI-1640 medium supplemented or lacking rIris (400 n M ) Cells were left at 37 C, 5% CO 2 for 4 or 24 h, as indicated Culture supernatants were harvested and TNF-a dosed by ELISA Numbers are the mean of three experiments (± SD) of three individuals per experiment.
TNF-a production
(pgÆmL)1)
PBMC stimulation LPS (100 ngÆmL)1) PGN (2 lgÆmL)1) Poly(I : C) (10 lgÆmL)1) ODN 2006 (10 lgÆmL)1)
*P < 0.05, **P < 0.01, ***P < 0.001 compared with controls (one-way ANOVA).
Trang 4added to LPS-stimulated PBMC before evaluating
TNF-a production
As illustrated in Fig 4, all forms of Iris tested,
whether retaining the serpin activity (native rIris) or not
(L339A mutant, cleaved elastase-complexed Iris), had a
similar effect: their addition to LPS-stimulated PBMC
resulted in a progressive (dose-dependent) reduction in
TNF-a release compared with the cells stimulated with
LPS alone The effect was maximal at 400 nm, reducing
TNF-a release to a level lower than that produced by
PBMC cultured without LPS stimulation (not shown)
These results showed that inhibition of TNF-a release
by Iris does not require its serine protease inhibitor
activity or integrity of the RCL domain
Exosites mediate the inhibitory effect of Iris on
TNF-a production
The similar effects of wild-type native, cleaved
prote-ase-complexed and RCL mutant forms of Iris on
TNF-a production suggested that Iris activity was
independent of serpin activity⁄ domain
Therefore, we sought to predict the domains of Iris
distinct from the RCL domain that may potentially
CD16 10
4
10 5
Comp-FITC-A
10 3
0
CD14 CD14 –
CD14 +
80
100
B C
A
60
100
80
60
20
40
20
Comp-FITC-A
Comp-FITC-A
0
0 10 3 10 4 10 5
0 10 3 10 4 10 5
0 10 3 10 4 10 5 0
Fig 3 Iris interacts with the mono-cyte ⁄ macrophage PBMC population PBMC were incubated with fluorescent-labeled rIris, CD14 and CD16 antibodies (A) Expres-sion of CD16 and CD14 on PBMC (B) CD14 ) populations gated in (A) were analyzed for Iris binding (dotted line, CD14 ) )CD16+)cells; tinted line, CD14 ) CD16)cells) (C) CD14+ populations gated in (A) were analyzed for Iris binding (dotted line, CD14 ) CD16+ )cells; bold line, CD14+ CD16+ cells, tinted line,
CD14+ CD16 ) )cells) FACS profiles are representative of one of five individuals tested in two independent experiments Numbers in the FACS profiles indicate the percentage of cells within the indicated gates.
0
200
400
600
800
1000
Concentration (n M )
L339A Iris + Elastase Elastase
Fig 4 Dose-dependent effects of wild-type and mutated rIris on TNF-a release by LPS-stimulated PBMC PBMC were activated for
4 h by LPS (10 lgÆmL)1) in the presence or absence of wild-type rIris (Iris), mutant rIris (L339A), rIris–elastase complex (Iris + elas-tase) or elastase alone (elaselas-tase) at the indicated concentrations TNF-a concentrations were measured in the harvested cell superna-tants In order to prepare the Iris–elastase complex, the two pro-teins were incubated in equimolar quantities for 1 h at 37 C Only samples with a recorded elastase activity < 25% compared with the control were used Values are the mean of three experiments (± SD).
Trang 5interact with other proteins by molecular modeling.
Two different types of sequence-based methods were
used On the one hand, potential immunogenic
domains were predicted, based on sequence analysis of
Iris, using a combination of DeLisi & Berzofsky’s [23],
Eisenberg et al.’s [24] and HCA [25] methods This
approach, predicting accessible, charged amphipathic
fragments [26,27], identified seven putative
epi-topes⁄ immunogenic fragments located between posi-tions 7–21 (ep1), 66–79 (ep2), 85–98 (ep3), 105–120 (ep4), 127–143 (ep5), 290–306 (ep6) and 312–325 (ep7)
of Iris (Fig 5) On the other hand, the RDB method identified eight regions within Iris as putative pro-tein⁄ protein interaction sites (Fig 5) located between positions 18–27 [receptor binding domain (RBD)1], 61–71 (RBD2), 91–98 (RBD3), 111–116 (RBD4), 125–
133 (RBD5) 139–150 (RBD6) 190–198 (RBD7) and 223–227 (RBD8), respectively We relied on a 3D model of Iris that we had established previously [21] (Fig 5) to address the location and accessibility of these antigenic and RBD domains
Five of the putative interacting domains identified
by either method were overlapping (RBD1–RBD5) One (RBD3) was not considered able to form a good interacting domain because of a bad Berzofsky score The properties of the four remaining selected peptides are summarized in Table 2 The analysis indicated that: (a) domains 62–67 (overlapping antigenic peptide ep2), 128–131 and 142–147 (both overlapping peptide ep5) defined particularly good interacting domains; and (b) antibodies against peptides 2 and 3 should interfere with a common interaction site (Fig 5) The four peptides described in Table 2 (ep1, ep2, ep4, ep5) were synthesized and used to immunize rab-bits The resulting antisera were evaluated for their neutralizing effect on the ability of Iris to block LPS-induced TNF-a production in PBMC The antibody titers of the different sera were similar (not shown) Table 3 shows that antibodies targeting peptides ep2 and ep4 reduced the inhibition of TNF-a production
in a dose-dependent manner, the anti-ep4 serum being more potent than the ep2 serum Conversely, anti-ep1 serum had no effect on the inhibitory action of Iris Finally, the action of anti-ep5 serum could not be analyzed because it inhibited the production of TNF-a
by itself, i.e in the absence of Iris (data not shown) From these results, we conclude that the interaction site responsible for the anti-inflammatory effects of Iris
is a conformational region covering domains RBD2 and RBD4, notably composed of helices D (67–79)
Fig 5 Potential antigenic epitopes and RBD prediction in Iris The
3D structure of Iris is represented as a blue ribbon and the P1
resi-due important for the antiprotease activity is represented in orange
(true volume) The axis of helices D and E and sheet1A is
high-lighted by a green, red or pink arrow, respectively (A) The
predicted epitopes are represented in yellow and numbered as
fol-lows: (1) 7–21, (2) 66–79, (3) 85–98, (4) 105–120, (5) 127–143, (6)
290–306, (7) 312–325 (B) The predicted RBD are represented in
yellow and numbered as follows: (1) 21–25, (2) 62–67, (3) 92–99,
(4) 111–115, (5) 128–131, (6) 142–147, (7) 192–196, (8) 223–228.
Table 2 Potential amphipathic domains within Ixodes ricinus immunosuppressor (Iris) predicted both as immunogenic epitopes and recep-tor-binding domain The first column indicates the amino acid positions in the protein sequence Angle, the calculated angle between the helix axis and the plane of a model membrane ASA, accessible surface area +, the peptide has an adequate mean surface accessibility
‡ 30%.
Peptide Sequence Total number aa Number positive charge Number negative charge Number polar aa Angle () ASA
Trang 6and E (104–114) and sheet 1A (117–121) (Fig 5).
These domains are distinct from the RCL (amino acids
324–340) involved in the antihemostatic action of Iris
[21] and are not affected by the structural
rearrange-ment during protease inhibition
Iris delivery inhibits LPS-induced septic shock
Because of its ability to interfere with TNF-a release,
Iris seemed a good candidate to counteract
endo-toxemia The in vivo half-life of Iris was determined to
verify whether it was appropriate for use in an animal
model of endotoxemia Figure 6 shows that the 125
I-labeled Iris concentration in plasma remained stayed
similar for at least 20 h after i.p injection, decreasing
to 20% of the maximum observed value 44 h after administration In order to address the ability of Iris
to counteract endotoxemia, we used a model of murine sepsis following LPS injection Mice were separated into two groups and treated with Iris (30 mgÆkg)1, i.p.)
or NaCl⁄ Pirespectively Two hours later, endotoxemia was induced by LPS injection (40 mgÆkg)1, i.p.) As shown in Fig 7, the mortality rate in the Iris-treated group ( 50%) was significantly lower than in the NaCl⁄ Pi-treated group ( 80%) (P < 0.001) Further-more, mean survival time in the Iris-treated group was
48 h, compared with 24 h in the control group, indi-cating that Iris increased both survival rate and sur-vival time In order to discount the possibility of a BSA-like effect for rIris, we injected mice with the same amount of ovalbumin, another serpin As expected, there was a significant difference in mortality rate between rIris- and ovalbumin-treated groups (P = 0.0019), whereas no difference was found between the NaCl⁄ Pi- and ovalbumin-treated groups However, when administrated after the induction of endotoxemia, Iris remained without effect (results not shown) In addition, Iris had no beneficial effect on caecum ligature puncture-induced sepsis (results not shown) Because Iris was able to reduce TNF-a release
in a LPS-activated PBMC culture, we asked whether the protective effect of Iris was related to interference with the cytokine storm usually associated with septic death Figure 8 shows that the administration of Iris significantly inhibited TNF-a release (P < 0.001), and
to a lesser extent the release of monocyte chemo-attractant protein-1 (MCP-1) (P < 0.01) and IL-6 (P < 0.05) in the blood following LPS injection
Table 3 Effect of neutralization of Ixodes ricinus
immunosuppres-sor (Iris) activity on tumor necrosis factor-alpha (TNF-a) production
induced by lipopolysaccharide (LPS) in peripheral blood
mononu-clear cells by polyclonal anti-rIris serum Results are expressed as
percentage of cytokine concentration recorded in the presence of
LPS and absence of rIris (+LPS; )rIris) ⁄ , no antibody added;
)LPS, no LPS added TNF-a expression values are expressed in
percentage relative to control ( )rIris; +LPS)
Antibodies
dilution
TNF-a production
Fig 6 Half-life of 125 I-labled Iris in the blood 125 I-labled Iris (10 lg;
107c.p.m.) was administrated i.p Blood samples were collected at
the indicated times by cardiac puncture, and platelet-poor plasma
was prepared by centrifugation Aliquots were counted in a gamma
counter Counts per minute per 500-lL aliquots are plotted against
time Values are the mean of three experiments (± SD).
0 20 40 60 80
PBS Ovalbumin
(h)
Fig 7 Iris treatment protects against LPS-induced toxic shock Mice were injected i.p with rIris (30 mgÆkg)1) or NaCl ⁄ P i Two hours later, septic shock was induced by i.p administration of
E coli serotype O111:B4 LPS (40 mgÆkg)1) Survival was recorded and plotted against time as percentage of injected animals surviving (n = 40).
Trang 7IL-10 production was not affected This is in
agreement with our previous observation that in vitro
Iris inhibits the LPS-induced increase in TNF-a,
inter-feron-c, IL-6 and IL-8, although IL-10 levels are not
affected [18] The IL-1b concentration remained too
low to detect any statistically significant difference and
was apparently unaffected during the time course of
the experiment (data not shown) Again, this
con-firmed our previous in vitro measurements [18] In
summary, Iris is able to increase⁄ prolong both survival
rate and survival time in mice undergoing LPS-induced
endotoxemic shock This correlated with a reduction
of endotoxemic cytokine production
Discussion
Iris as a multifunctional tick saliva protein
The currently documented functions of serpins and
tick saliva suggest a role for tick serpin(s) in the
modu-lation of immune response, coagumodu-lation, fibrinolysis,
complement regulation and inflammation or
angiogen-esis [28,29] In particular, the I ricinus
immunosup-pressor protein Iris, which is induced in tick saliva
during the blood meal [18–20], was suggested to
dis-turb the TH1⁄ TH2 balance by inhibiting interferon-c
production It was also shown to preferentially target
the human leukocyte elastase or pork pancreatic
elas-tase [21] and, to a lesser extent, tissue plasminogen
activator, coagulation factor X and thrombin As such,
Iris may act in physiological processes relevant for the
tick blood meal, disturbing its serpin activity
hemosta-sis through interference with fibrinolyhemosta-sis, contact
phase-activated pathway of coagulation and, to a
lesser extent, platelet aggregation Iris may also exert anti-inflammatory activity because soluble protein extracts of CHO-KI cells expressing rIRIS inhibited production of inflammatory cytokines like TNF-a trig-gered by LPS [18]
In this study, we documented that affinity-purified rIris was able to block TNF-a produced by PBMC activated by various TLR agonists, namely LPS (TLR4), poly(I : C) (TLR3), ODN 2006 (TLR9) and PGN (TLR2) Moreover, we indicated how Iris may exert its blocking effect: Iris was found to interact physically with the two major monocyte fractions
of PBMC, namely CD11b+ CD14+ CD16) and CD11b+ CD14+ CD16+ cells [30] Appropriate con-trols discarded the possibility of an action mediated by interference with LPS, reagents used in the dosage, TNF-a itself or TNF-a half-life (results not shown) Finally, we also showed that delivery of Iris in vivo significantly lowered the mortality rate and increased the survival time of mice undergoing LPS-induced sep-tic shock This effect was not observed upon injection
of the same amounts of ovalbumin, another serpin, used as a control, arguing against a BSA-like effect (Fig 7) The observed protective effect of Iris was correlated with the inhibition of TNF-a, MCP-1 and IL-6 production, all of which participate in the cyto-kine storm associated with LPS endotoxinemia [31,32]
Of note, Iris had no activity in caecum ligature punc-ture-induced sepsis (not shown) This is reminiscent of the lack of action of specific anti-TNF-a IgG in the latter pathology and in agreement with the lack of involvement of TNF-a in this model [33] However, Iris had a significant beneficial effect only when admin-istrated before LPS-induced endotoxemic shock (not
Fig 8 Effect of rIris on LPS-induced
cyto-kine release in vivo Mice were injected
i.p with rIris (30 mgÆkg)1) or NaCl ⁄ P i Two
hours later, septic shock was induced by
i.p administration of E coli serotype
O111:B4 LPS (40 mgÆkg)1) TNF-a, IL-6,
MCP-1 and IL-10 levels were measured in
the platelet-poor plasma collected by cardiac
puncture at the indicated times post
injec-tion of LPS Results are expressed as
means ± SEM of six mice per group for
each time point *P < 0.05, **P < 0.01,
***P < 0.001 compared with controls
(one-way ANOVA).
Trang 8shown) This may be expected, because maximal
TNF-a levels TNF-are recorded very shortly (90 min) TNF-after
LPS-induced endotoxemic shock, suggesting the need for
immediate action in order to interfere
Exosites mediate the anti-inflammatory
action of Iris
There are straightforward connections between the
enzymatic function of Iris – a specific inhibitor of
leu-kocyte elastase – and some of its putative physiological
activities In this regard, the proinflammatory effects
of fragments generated from extracellular matrix
deg-radation by elastase are well documented For
exam-ple, degradation products of elastin or heparan sulfate
proteoglycan can act as chemoattractants towards
inflammatory cells or activate TLR4, respectively
[34,35] TLR4 activation by LPS, responsible for fever,
shock and death in sepsis, is thought to be prevented
in vivo by the extracellular matrix By cleaving matrix
proteins, elastase liberates TLR4 from this
extracellu-lar matrix constraint, favoring its interaction with
potential ligands and activation of the inflammatory
immune reaction [36] Leukocyte elastase has also been
found to modulate chemokine and cytokine activity,
activate cell-surface receptors and cleave the
antiadhe-sive coat of neutrophils [34,37] Therefore, by
inhibi-ting leukocyte elastase, Iris could clearly interfere with
the inflammatory response Through its inhibition of
tissue plasminogen activator, factor X and thrombin,
Iris may also interfere directly with coagulation
The reported data nevertheless suggested that some
functions of Iris were independent of its enzymatic
inhibitory activity Indeed, both wild-type Iris and the
mutant devoid of serpin activity (L339A; P2) were
found to similarly increase platelet adhesion time [21]
This was reminiscent of the biological activity of native
alpha-1-antitrypsin, which was shown to be
indepen-dent of its inhibitory activity on serine proteases [38]
Moraga et al [39] demonstrated that cleaved (devoid
of activity) alpha-1-antitrypsin still has an effect on
IL-6 and TNF-a production by monocytes⁄
macro-phages This action was inferred to the presence of
exosites within alpha-1-antitrypsin Similar
assump-tions could be made regarding Iris
To test this hypothesis, we first evaluated whether
the ability of Iris to inhibit TNF-a released by PBMC
activated by LPS was dependent on the activity⁄
integ-rity of the RCL (anti-proteasic domain) Wild-type Iris
and its inactive mutant L339A inhibited release of the
inflammatory cytokine to the same extent This is in
sharp contrast to the effect of rIris on fibrinolysis, an
activity lost in the RCL mutant [21] This
indepen-dence of the protease inhibitory activity of the anti-inflammatory activity of Iris was further sup-ported using the cleaved⁄ inactivated Iris protein obtained by incubation with its target serine protease
We observed that an elastase⁄ Iris complex retains its inhibitory activity on the production of TNF-a This further confirms that Iris devoid of its inhibitory activ-ity still inhibits TNF-a production It further indicates that Iris preserves its anti-inflammatory activity even after a conformation change
Because an intact RCL domain seems dispensable for the inhibitory action of Iris on TNF-a production, the contribution of exosites was evaluated These can
be predicted using either the RBD method [26] or by epitope predicting (according to the Berzofsky, Eisen-berg and HCA methods) According to the most strin-gent criteria of the two methods, four antigenic peptides were selected within Iris and synthesized to generate specific antibodies Antibodies targeting two
of these peptides were shown to impair the ability of Iris to inhibit TNF-a by LPS-activated PBMC Anti-ep2 serum had an activity twice as low as that of anti-ep4 serum Because the different sera had the same antibody titer and were used at the same dilu-tion, this might translate into a real biological differ-ence and be related to the 3D location of these epitopes The antigenic fragments are close in the 3D structure and correspond to helices D and E-s1A (posi-tions 2, helix D, and 4, helix E-s1A, in Fig 5A) Therefore, an interacting site, involved in the anti-inflammatory function of Iris, corresponds to a region involving helices hD and hE However, it cannot be ruled out that only the helix E-s1A region is implicated
in inhibition of TNF-a release, because antibodies tar-geting helix E seem more potent inhibitors (although displaying the same affinity); the effects observed with antibodies targeting helix D may be related to an arti-fact caused by steric hindrance, antibodies masking helix E when binding to helix D (Fig 8) Further mutational studies could be performed to verify whether the helix D region is truly implicated in the immunomodulatory function of Iris However, these different mutations (with one or more mutations on a large peptide of 16 amino acids) might weaken the protein structure and distort the functional analysis
To summarize, we showed that Iris exerts an anti-inflammatory action, blocking the release of TNF-a by various TLR agonist-activated monocytes, independent
of the RCL and antiprotease activity This action is mediated by (an) exosite(s) (RBD2 and RBD4), nota-bly composed of helices D (67–79) and E (104–114) and sheet 1A (117–121), likely to mediate direct inter-action with monocytes Thus, Iris might be beneficial
Trang 9for the parasite, by interfering with hemostasis [21],
and for the host by blocking excessive serine protease
activity during acute inflammation and regulating the
expression of pro- and anti-inflammatory mediators
This may prove particularly useful when removed from
the parasite–host interaction context because it may
form the basis of a drug acting in pathological
situa-tions involving the overexpression of TNF-a
Materials and methods
Preparation of rIris, mutant L339A and the
cleaved forms of rIris
Purified recombinant wild-type Iris (rIris) (Fig S1) and
mutant L339A were produced in a baculovirus expression
system, according to Prevot et al [21] The cleaved form of
rIris was obtained by incubation for 1 h at 37C in the
presence of equimolar amounts of pancreatic elastase
Puri-fication buffers were prepared with Limulus amoebocyte
lysate reagent water (Lonza, Valais, Switzerland) Purified
proteins were diluted in NaCl⁄ Pi (pH 7.4) and tested for
endotoxin contamination using the QCL-1000 kit (Lonza)
Endotoxin levels were < 0.4 enzyme unitsÆmg)1 of protein
in all preparations used Samples containing endotoxin
amounts superior to that threshold were loaded on
Detoxi-Removal endotoxin gel columns according to the
manufac-turer’s instructions (Pierce, Rockford, IL, USA) to ensure
the removal of endotoxins This was followed by dialysis
against buffers appropriate for the following experiments
Protein concentrations in the endotoxin-purified batches
were determined using a microBCA kit (Pierce) according
to the manufacturer’s instructions
In order to control the protein activity of the different
forms of rIris, elastase inhibition was assessed, as described
by Prevot et al [20] Briefly, rIris, L339A mutant and the
cleaved form of rIris were incubated with pancreatic
elas-tase at an equimolar ratio for 10 min at room temperature,
in 0.1 m Tris buffer, pH 7.5 After addition of a
chromo-genic elastase substrate [succinyl-(Ala)3-p-nitroanilide;
Sigma, St Louis, MO, USA] to a final concentration of
0.5 mm, absorbance was measured at 405 nm for 280 s
Absorbance values used to calculate elastase inhibition were
corrected with controls containing buffer and substrate
only Inhibition values were 70% and 0% for rIris and
L339A mutant, respectively
Cell isolation and culture
PBMC were isolated from buffy coats using
Ficoll–Leuco-sep tubes (Greiner Bio One, Stuttgart, Germany) according
to the manufacturer’s instructions Briefly, heparinated
blood samples from three healthy human donors were
cen-trifuged at 400 g for 35 min, at 18C in Leucosep tubes
Cells at the interface were then collected and washed three times in NaCl⁄ Pi Cell numbers were determined using a Burker counting chamber
PBMC were seeded in 96-well culture plates (2· 105cellsÆ well)1; Falcon, Becton Dickinson, Plymouth, UK) and acti-vated by the indicated stimulus (PGN, 10 lgÆmL)1; ODN 2006, 2 lgÆmL)1; poly(I : C), 10 lgÆmL)1; LPS,
100 ngÆmL)1) in a total volume of 200 lL complete
RPMI-1640 medium in the presence or absence of the different form
of rIris (0–400 nm) Cells were incubated at 37C, 5%
CO2for various times (4–24 h) depending on the cytokines
to be assayed Culture supernatants were conserved at –
80C before analysis for their contents in cytokine Dexa-methasone (10 lm; Sigma) was used as a positive control for inhibition of cytokines production
Flow cytometry rIris was labeled with fluorescein isothiocyanate (FITC–Iris)
or allophycocyanin (APC–Iris) using the Alexa Fluor488 and 647 Protein Labeling kits (Molecular Probes, Carlsbad,
CA, USA), respectively Labeled rIris (100 nm) was incu-bated with 5· 105cells, LPS stimulated (100 ngÆmL)1; when indicated), for 30 min at 4C in the dark Cells were then washed in NaCl⁄ Pibefore incubation with the appro-priate antibodies (CD3, CD19, CD11b, CD14, CD16) labeled with R-Phycoerythrin or FITC (Biocytex, Marseille, France) for another 30 min at 4C in the dark After wash-ing twice with NaCl⁄ Pi, cells were subjected to FACS anal-ysis using a FACSCalibur flow cytometer (Becton Dickinson, San Jose, CA, USA)
To ascertain the specificity of the labeled rIris fluorescent signal, we performed a competition test between labeled and unlabeled rIris (both 100 nm) Briefly, the labeled rIris signal on PBMC was measured in presence of equimolar concentration of unlabeled rIris
Cytokine measurements Cytokine production was assessed using ELISA kits (human TNF-a or mouse MCP-1, TNF-a, IL-6, IL-10, and IL-1b) (eBioscience, San Diego, CA, USA) according to the manufacturer’s instructions
Polyclonal anti-rIris serum production Anti-rIris serum was produced as described by Prevot
et al.[20] Briefly, female New Zealand white rabbits (Har-lan, the Netherlands) were injected subcutaneously with
50 lg of purified rIris in NaCl⁄ Pi (pH 7.5), emulsified in Freund’s adjuvant The first injection was performed in complete Freund’s adjuvant and two boosters in incom-plete Freund’s adjuvant Injections were performed at 3-week intervals Animals were bled on the day of injection
Trang 10and 2 weeks after the last booster Two rabbits were
mock-immunized following the same protocol with
NaCl⁄ Piin Freund’s adjuvant as negative controls Animal
care and experimental procedures were carried out in
accordance with the Helsinki Declaration (Publication 85–
23, revised 1985), local institutional guidelines (laboratory
license number LA 1500474) and the Belgian law of
August 14th, 1986 as well as the royal decree of November
14th, 1993 on the protection of laboratory animals
Iris neutralization assays
rIris was preincubated with dilutions of immune and
con-trol (mock immunized or preimmune sera) sera for 10 min
at 37C in NaCl ⁄ Pi The effect of these samples on
cyto-kine production by PBMCs activated by LPS was then
assessed as described above
Prediction of interaction sites
The prediction of binding sites from the protein sequence
was made using the RBD method [26] This method is
derived from the Eisenberg’s method [24] based on the
calcu-lation of the mean hydrophobicity <H> and the mean
hydrophobic moment <l> for each amino acid by moving
a five-residue window along the sequence The method
pre-dicts accessible and charged domains potentially involved in
an interaction and is described in detail in Gallet et al [26]
Prediction of epitopes
The method involves the search for hydrophilic amphipathic
helices based on the primary sequence of the protein [27]
Amphipathic domains were predicted using a combination of
various available methods, such as DeLisi & Berzofsky’s [23],
Eisenberg et al.’s [24] and HCA [25] methods The first
method relies on known antigenic sites, the second allows the
detection of residues located at the protein surface, whereas
in the latter, amphipathic domains are visualized on a
bidimensional representation of the protein sequence
Furthermore, the secondary structure of the protein was also
predicted using different algorithms, such as NPSA (http://
npsa-pbil.ibcp.fr), PROF [40] and Psipred [41]
Fragments corresponding to appropriate criteria of
anti-genicity, amphipathicity and helicity are then reconstructed
in 3D and minimized using hyperchem 6.0 (Hypercube
Inc, Gainesville, FL, USA) Their interaction with a model
membrane is simulated by the IMPALA method [42] This
step allows an evaluation of the hydrophobic⁄ hydrophilic
segregation (i.e the amphipathicity) of the helices The
mean surface accessibility of the predicted epitopes on the
Iris 3D structures was calculated by averaging the accessible
surface area of the peptide residues using the Shrake and
Rupley method, as described previously [43]
Anti-peptide serum production Rabbits were immunized by an injection of 0.1 mg of each synthetic antigenic peptide coupled to KLH Two subsequent boosters were given at a 2-week interval Sera were collected
1 week after the last booster The antibody titer was mea-sured by ELISA as described previously [20] Briefly, 250 ng
of rIris in NaCl⁄ Piwas initially coated onto 96-well plates (Nunc, Rochester, NY, USA) overnight at 4C Wells were then saturated for 1 h in NaCl⁄ Pi⁄ 0.1% Tween 20 ⁄ 1% BSA
at room temperature The coated plates were incubated with various dilutions of immune or preimmune sera for 2 h at room temperature A secondary biotinylated anti-IgG (dilu-tion 1 : 10 000) was added for 1 h, followed by peroxidase-coupled streptavidin (1 : 10 000) for 30 min at room temper-ature Finally, the TMB chromogen (Sigma) was added for
10 min Absorbance was then read at 450 and 630 nm with a Model 680 microplate reader (Bio-Rad, Hercules, CA, USA) Values were expressed as antibody titers as defined by the serum dilution at the inflection point of the curve
Iodination of Iris
125
I-labeled Iris was prepared by iodination with [125I] sodium iodide (Perkin Elmer, Walthman, MA, USA) at
1 mCiÆmg)1of protein, using Iodo-gen (Pierce; 100 lgÆmg)1
of protein) following the manufacturer’s instructions Briefly, 2 Iodo-Gen beads were washed in 1 mL NaCl⁄ Pi, incubated for 5 min with 20 lL sodium iodide (10 mCiÆmL)1), and then with 200 lg Iris (1 mgÆmL)1) for
15 min at room temperature Free iodide was removed using Zeba Desalt Spin Columns (Pierce) following the manufacturer’s instructions
Determination of125I-labeled Iris half-life in rat blood
The in vivo blood persistence of125I-labeled Iris was evalu-ated after i.p administration in female Whistar Hanover rats (200 g) of 107c.p.m (corresponding to 10 lg Iris, resuspended in 200 lL NaCl⁄ Pi) Blood was collected 3, 20,
40, 60 and 120 h later by cardiac puncture and citrated (13 mm, final concentration) Platelet-poor plasma was then obtained by centrifugation and 500-lL aliquots were placed
in glass test tubes Radioactivity was measured using a gamma counter (LKB, Wallac, Finland) All animals were maintained and handled according to local and national ethical guidelines
Animal model of septic shock Two groups, each containing 40 female NMRI mice (30–
35 g), were injected i.p with: (a) 500 lL Iris (30 mgÆkg)1) dialyzed against NaCl⁄ Pi, or (b) NaCl⁄ Pi alone for the