Endogenous lipid- and peptide-derived anti-inflammatory pathways generated with glucocorticoid and aspirin treatment activate the lipoxin A4 receptor

Abstract

Aspirin (ASA) and dexamethasone (DEX) are widely used anti-inflammatory agents yet their mechanism(s) for blocking polymorphonuclear neutrophil (PMN) accumulation at sites of inflammation remains unclear. Here, we report that inhibition of PMN infiltration by ASA and DEX is a property shared by aspirin-triggered lipoxins (ATL) and the glucocorticoid-induced annexin 1 (ANXA1)-derived peptides that are both generated in vivo and act at the lipoxin A(4) receptor (ALXR/FPRL1) to halt PMN diapedesis. These structurally diverse ligands specifically interact directly with recombinant human ALXR demonstrated by specific radioligand binding and function as well as immunoprecipitation of PMN receptors. In addition, the combination of both ATL and ANXA1-derived peptides limited PMN infiltration and reduced production of inflammatory mediators (that is, prostaglandins and chemokines) in vivo. Together, these results indicate functional redundancies in endogenous lipid and peptide anti-inflammatory circuits that are spatially and temporally separate, where both ATL and specific ANXA1-derived peptides act in concert at ALXR to downregulate PMN recruitment to inflammatory loci.

Full text

1296 NATURE MEDICINE • VOLUME 8 • NUMBER 11 • NOVEMBER 2002
ARTICLES
Inflammation plays a key role in many prevalent diseases such
as arthritis, and there is increased appreciation that atherosclero-
sis and Alzheimer disease also share uncontrolled inflammation
as part of their etiology
1
. The anti-inflammatory and analgesic
actions of ASA have been well known for more than 100 years
2–7
.
At low doses, ASA displays newly recognized beneficial actions
including prevention of cardiovascular diseases not shared by
other non-steroidal anti-inflammatory drugs
3
, that may also ex-
tend to cancer treatments
4,5
. Dating from the early work of
Hench et al.
8
, glucocorticoids are also widely used to treat in-
flammatory diseases. Although prolonged treatment with either
ASA or glucocorticoids is widely used, each is associated with an
unacceptably high level of adverse side effects
9,10
. Thus it is im-
portant to understand their action in order to have new ap-
proaches to controlling unwanted inflammation. Control of
PMN functions is critical in the amplification of inflammation.
ASA and glucocorticoids each independently share the ability to
inhibit a key first step in inflammation, namely leukocyte-en-
dothelial adhesion
11
. The molecular basis, however, underlining
their anti-PMN activity remains of interest because their elucida-
tion could shed light on novel pathways that might be useful in
sparing unwanted side effects of these agents.
Impact of combined ASA and DEX on neutrophil extravasation
To this end, we addressed whether ASA and glucocorticoids have
overlapping components in their therapeutic actions, namely
regulation of PMN trafficking (Fig. 1). Using a murine dorsal air
pouch inflammation that is widely studied and likened to fea-
tures of rheumatoid synovium
12
, we evaluated the combined im-
pact of ASA and DEX in regulating PMN recruitment to
inflammatory sites and exudate formation with the microbial
particle Zymosan A. Administration of both agents (10 µg DEX
and 1 mg ASA) into pouches gave statistically significant inhibi-
tion of Zymosan A-driven PMN infiltration (∼76%) (Fig. 1). For
direct comparison, when administrated alone, DEX (10 µg) gave
53% and ASA (1 mg) 36% inhibition (Fig. 1). This is consistent
with results of early studies with human rheumatoid arthritis pa-
tients where low-dose DEX and ASA in combination was found
to be a highly effective anti-inflammatory therapy
13
. Also, the re-
sults in Fig. 1 suggest that the inhibitory actions of DEX and ASA
in combination on PMN trafficking in vivo may involve path-
ways in addition to those already appreciated for these agents
when used independently
8–11
.
Of the various hypotheses proposed for glucocorticoid’s anti-
inflammatory action, attention initially focused on annexin 1
Endogenous lipid- and peptide-derived anti-inflammatory
pathways generated with glucocorticoid and aspirin
treatment activate the lipoxin A
4
receptor
MAURO PERRETTI
1
, NAN CHIANG
2
, MYLINH LA
1
, IOLANDA M. FIERRO
2
, STEFANO MARULLO
3
,
S
TEPHEN J GETTING
1
, EGLE SOLITO
4
& CHARLES N. SERHAN
2
1
Department of Biochemical Pharmacology, William Harvey Research Institute,
Bart’s and The London School of Medicine, Queen Mary University of London, Charterhouse Square, London, UK
2
Center for Experimental Therapeutics and Reperfusion Injury, Department of Anesthesiology,
Perioperative and Pain Medicine, Brigham and Women’s Hospital and
Harvard Medical School, Boston, Massachusetts, USA
3
Department of Cell Biology, ICGM, Pavilion Gustave Roussy, Paris, France
4
Department of Neuroendocrinology, Faculty of Medicine, Imperial Collegee, London, UK
M.P. and N.C. contributed equally to this paper.
Correspondence should be addressed to C.N.S.; email: cnserhan@zeus.bwh.harvard.edu
Published online 7 October 2002; doi:10.1038/nm786
Aspirin (ASA) and dexamethasone (DEX) are widely used anti-inflammatory agents yet their
mechanism(s) for blocking polymorphonuclear neutrophil (PMN) accumulation at sites of in-
flammation remains unclear. Here, we report that inhibition of PMN infiltration by ASA and DEX
is a property shared by aspirin-triggered lipoxins (ATL) and the glucocorticoid-induced annexin
1 (ANXA1)-derived peptides that are both generated in vivo and act at the lipoxin A
4
receptor
(ALXR/FPRL1) to halt PMN diapedesis. These structurally diverse ligands specifically interact di-
rectly with recombinant human ALXR demonstrated by specific radioligand binding and func-
tion as well as immunoprecipitation of PMN receptors. In addition, the combination of both ATL
and ANXA1-derived peptides limited PMN infiltration and reduced production of inflammatory
mediators (that is, prostaglandins and chemokines) in vivo. Together, these results indicate
functional redundancies in endogenous lipid and peptide anti-inflammatory circuits that are
spatially and temporally separate, where both ATL and specific ANXA1-derived peptides act in
concert at ALXR to downregulate PMN recruitment to inflammatory loci.
© 2002 Nature Publishing Group http://www.nature.com/naturemedicine

NATURE MEDICINE • VOLUME 8 • NUMBER 11 • NOVEMBER 2002 1297
ARTICLES
(denoted earlier as lipocortin 1)
14
. Annexin 1 (ANXA1) is a pro-
tein of 346 amino acids and a member of the annexin superfam-
ily that is a highly abundant PMN-derived protein induced by
glucocorticoids in vivo. Both ANXA1 as well as its enzymatically
generated ANXA1-derived peptide (for example. peptide Ac2-26,
or Ac-AMVSEFLKQAWFIENEEQEYVQTVK) are potent inhibitors
of PMN transmigration and phagocytosis in vitro and in vivo
15,16
.
Along these lines, in addition to inhibiting formation of both
prothrombotic and pro-inflammatory eicosanoids
6,7
, ASA treat-
ment was more recently shown to trigger formation of endoge-
nous anti-inflammatory lipid mediators
17
. These mediators are
generated, for example, when ASA acetylates cyclooxygenase-2
and initiates the generation of 15-epimeric lipoxin A
4
(LXA
4
),
termed aspirin-triggered LXA
4
(ATL). Like other autacoids, both
ATL and their lipoxygenase-derived carbon 15 epimers, the na-
tive lipoxins, are generated and act locally, followed
by rapid inactivation via metabolic conversion.
Metabolically stable analogs of LXA
4
and ATL were
designed to resist rapid enzymatic inactivation that
prolongs their actions. These compounds display po-
tent anti-PMN actions and act with high affinity (K
d
∼0.5 nM) at a G protein–coupled receptor, LXA
4
re-
ceptor (ALXR; also referred to as formyl peptide re-
ceptor-like 1 or FPRL1)
17
. Along with its endogenous
anti-inflammatory lipid ligands, ALXR/FPRL1 also
directly interacts with selective peptides that are re-
vealed during tissue damage in situ to evoke PMN
necrotaxis
18
. These independent observations raised
the possibility that DEX- and ASA-generated endoge-
nous mediators might share some common path-
ways governing leukocytes in inflammation. To test
this, we first questioned whether ANXA1-derived
peptides themselves might interact with human
PMN ALXR/FPRL1. The two pathways converge at a
shared receptor.
[
3
H]LXA
4
was prepared for direct ligand binding
with human embryonic kidney (HEK) 293 cells ex-
pressing human ALXR. Both ANXA1-derived peptide
Ac2-26 and a shorter peptide denoted Ac2-12 (Ac-
AMVSEFLKQAW) competed for specific [
3
H]LXA
4
binding with similar affinities (50% inhibitory con-
centration (IC
50
) ∼0.3 µM) (Fig. 2a). To determine
whether ANXA1-derived peptides directly bind re-
combinant ALXR/FPRL1, we prepared [
125
I-Tyr] Ac2-
26 peptide (corresponding to the major portion of
the N-terminus of ANXA1), which gave specific and
saturable binding with a K
d
value of ∼0.9 µM (Fig. 2b,
inset). The full-length ANXA1 protein itself also
competed for [
125
I-Tyr]Ac2-26 binding with similar affinity,
whereas the Ac2-12 peptide gave an IC
50
of ∼20 µM. In contrast, a
scrambled sequence peptide Ac2-6 did not compete for specific
[
125
I-Tyr]Ac2-26 binding (Fig. 2b). The structurally unrelated lipid
mediator LXA
4
also specifically competed for [
125
I-Tyr]Ac2-26
binding with HEK293 cells expressing ALXR/FPRL1 (Fig. 2b).
These ligands are selective for ALXR/FPRL1, as neither LXA
4
(ref.
18) nor Ac2-26 peptide competed for [
3
H]LTB
4
binding to human
LTB
4
receptor (BLT1), a related lipid mediator G protein–coupled
receptor (GPCR) (ref. 19), with BLT1-transfected HEK293 cells
(Supplementary Fig. A online).
For the purpose of direct comparison, isolated human PMNs
in suspension gave a K
d
of ∼1.3 µM for [
125
I-Tyr]Ac2-26, a value
comparable to that obtained with recombinant ALXR/FPRL1,
and LXA
4
competed for [
125
I-Tyr]Ac2-26 peptide binding to
human PMN (Fig. 2c). Serum amyloid protein A (SAA), a prote-
olytic product of the acute phase response
20
, was also tested be-
cause, at much higher concentrations than LXA
4
, it competes for
specific [
3
H]LXA
4
binding
18
, activates HEK293 cells expressing
ALXR/FPRL1 (ref. 21) and inhibits human PMN oxidative
burst
22
. SAA competed for specific [
125
I-Tyr]Ac2-26 binding with
an apparent IC
50
of ∼1 µM with human PMNs (Fig. 2c); this value
Fig. 1 Additive actions with DEX and ASA on PMN extravasation. DEX or
ASA were injected into murine air pouches 15 min before Zymosan A
(1 mg) injection. PMN accumulation was measured 4 h later by light mi-
croscopy. Results are the mean ± s.e.m of n = 8 per group. *, P < 0.05; **, P
< 0.01 versus Zymosan A alone. , Zymosan A alone; , +DEX; , +ASA;
, +DEX & ASA.
a
c
d
Fig. 2 Direct interaction of ANXA1 with ALXR: competitive [
125
I-Tyr]Ac2-26 and
[
3
H]LXA
4
binding. a and b, Human ALXR-transfected HEK293 cells (0.5 × 10
6
) were incu-
bated with [
3
H]LXA
4
(a) or [
125
I-Tyr]Ac2-26 (b) in the presence of an increasing concentra-
tion of unlabeled compounds: LXA
4
(), Ac2-26 (), Ac2-12 (), ANXA1 () or
scrambled Ac2-6 ( ). c and d, Suspension (c) or adherent human PMNs (d) were incu-
bated with [
125
I-Tyr]Ac2-26 in the presence of increasing concentration of unlabeled com-
pounds: LXA
4
(), Ac2-26 ( ) or SAA ( ). Insets in b and c, Scatchard plots of specific
[
125
I-Tyr]Ac2-26 binding. Results represent the mean ± s.e.m. from duplicates of n =3.
b
© 2002 Nature Publishing Group http://www.nature.com/naturemedicine

1298 NATURE MEDICINE • VOLUME 8 • NUMBER 11 • NOVEMBER 2002
ARTICLES
is comparable to that of ANXA1, suggesting that SAA and
ANXA1 share recognition sites on PMNs for their action. As
PMN adhesion is a key event that externalizes ANXA1, permit-
ting its inhibitory actions in PMN recruitment
23,24
, both lipid
(LXA
4
) and protein (SAA and ANXA1) ligand interactions with
endogenous ALXR/FPRL1 were evaluated with adherent PMNs
and compared with those of PMNs in suspension. Adherent
PMNs gave specific [
125
I-Tyr]Ac2-26 peptide binding with an ap-
parent K
d
of ∼0.9 µM (Fig. 2d), and both LXA
4
and SAA retained
their ability to compete for specific [
125
I-Tyr]Ac2-26 peptide
binding. Taken together, these results provide the first direct ev-
idence that ANXA1 as well as specific ANXA1-derived peptides
directly interact with human PMNs as well as recombinant
ALXR.
Next, we determined whether these ligands could evoke cellu-
lar responses via recombinant ALXR/FPRL1. We examined
chemotaxis with cells expressing ALXR/FPRL1 together with
Gqo chimera
18
. Although ATL analogs inhibit PMN
chemotaxis in vivo and in vitro, the presence of Gqo
chimera enables this ligand-receptor interaction to drive
chemotaxis as a reporter system in vitro. A stable analog
of both LXA
4
and ATL, denoted ATL analog 1 (ATLa1;
15(R/S)-methyl-LXA
4
) (Fig. 3a inset) gave clear chemo-
taxis responses (Fig. 3a), whereas the Ac2-26 peptide at
concentrations greater than 1,000 times that of ATLa1
did not. This peptide markedly inhibited ATLa1-evoked
reporter cell (Gqo chimera-ALXR) chemotaxis in a con-
centration-dependent manner (Fig. 3b). These findings
indicate that the ANXA1-derived peptide and ATLa1 can
act at the same recombinant ALXR/FPRL1, yet each lig-
and evoked different responses using this Gqo reporter
system. This was also demonstrable with human PMNs.
With the primary cells in vitro, the peptide Ac2-26 dis-
plays a chemotactic activity (Fig. 3c), and checkerboard
analysis indicated a chemokinetic activity as well (data
not shown). In contrast, ATLa1 alone at 10–100 nM did
not evoke chemotaxis, but clearly blocked Ac2-26 evoked
chemotaxis. Together with results with recombinant re-
ceptors (Fig. 3a and b), they indicated that Ac2-26 and
ATLa1 interact with the same PMN receptor (Fig. 3c).
Endogenous ANXA1/ALXR association in activated PMN
Recently, results from studies on the ANXA1-derived
peptides and inhibition of PMN chemotaxis in vitro im-
plicated that these peptides could interact with GPCRs
related at the nucleotide sequence level to ALXR/FPRL1—
namely one of the formyl peptide receptors (FPRs)
25
.
However, direct evidence for specific binding of the
ANXA1 peptides or full-length protein to either human
25
or mouse
26
FPRs was not demonstrated, and ANXA1 in-
teractions with circulating leukocytes or peritoneal
macrophages was only partially (∼40%) diminished in
FPR-deficient mice
26
. Thus, the involvement of FPRs in
the inhibitory actions of ANXA1 and its bioactive pep-
tides remained indirect because the conclusions drawn
relied heavily on the use of FPR receptor antagonist: the
Boc compounds that can also partially compete at
ALXR/FPRL1 (see Supplementary Fig. B online; panels a
and b showing that Boc1 compound blocked the effect of
SAA, a selective ALXR/FPRL1 ligand). Also, results from
FPR-deficient mice may not be conclusive because they
were not characterized for potential compensatory
changes in related murine receptors or direct ligand binding.
Moreover, these findings suggested that additional, possibly re-
lated, receptors were responsible for most of the inhibitory ac-
tions of ANXA1 in vivo and in human disease. Along these lines,
the ANXA1 peptide Ac2-26 as well as human recombinant
ANXA1 (hrANXA1) even at high concentrations (for example,
10 µM and 0.5 µM, respectively) did not affect the cellular re-
sponse of human PMNs produced by subsequent addition of
fMLP (85 ± 4% of the fMLP response; n = 4, P > 0.05) (Fig. 3d).
Thus, ANXA1 and fMLP each evoke responses via distinct and
separate receptors on human PMNs. Notably, the Ac2-26 peptide
response with PMNs (that is, calcium mobilization)
(Supplementary Fig. B online; panel c) was partially blocked by
genistein, a tyrosine kinase inhibitor that also modulates LXA
4
-
regulated PMN transmigration and adhesion
27
. The Ac2-26 re-
sponse further suggests that ANXA1 and LXA
4
can share some
intracellular signaling pathway(s) in PMNs (Figs. 2 and 3).
0.8
1.0
1.2
1.4
1.6
1.8
Chemotaxis index
#
HO OH
H
3
COH
COOCH
3
ATLa1
ATLa1
0.8
1.0
1.2
1.4
1.6
1.8
Chemotaxis index
*
*
*
*
0110
ANXA1-derived peptide
(Ac2-26,
µ M)
Inhibition of chemotaxis (%)
0
20
40
60
80
0 10 30 100 300
**
**
*
ANXA1-derived peptide
(Ac2-26, µM)
Fig. 3 ANXA1 peptides directly interact
with recombinant as well as endogenous
PMN ALXR. a, CHO-Gqo-ALXR cells were
added to the upper compartment of a mi-
crochamber (5 × 10
4
cells per well).
Chemotaxis was initiated by addition of Ac2-26 peptide (100 µM; ) or aspirin-trig-
gered lipoxin A
4
analog 1 (ATLa1; 100 nM; ) to the lower compartment. , vehi-
cle. b, CHO-Gqo-ALXR cells were pretreated with Ac2-26 peptide for 30 min at 37
°C and added to the upper compartment of a microchamber (5 × 10
4
cells per well).
Chemotaxis was initiated by addition of ATLa1 (100 nM) to the lower compartment.
c, Human PMNs were added to the upper compartment of a microchamber (5 × 10
4
cells per well). Chemotaxis was initiated by addition of Ac2-26 peptide (1–10 µM;
) to the lower compartment. In some cases, cells were treated with 10 () or 100
nM () ATLa1 for 30 min at 37 °C. #, P = 0.01, versus vehicle; *, P < 0.01, versus
Ac2-26 alone (a and c) or % inhibition of ATLa1- evoked chemotaxis (*, P = 0.02; **,
P < 0.01, versus ATLa1 alone in b). Data represent the mean ± s.e.m. from n =3 ex-
periments. d, Fura2-AM-loaded human PMNs (5 × 10
6
cells/incubation) were incu-
bated with the different stimuli, and rapid changes in intracellular [Ca
2+
] measured
by fluorimetry. Additions of agonists are denoted by arrows. hrANXA1: human re-
combinant annexin 1. Traces are representative of 3 independent experiments.
a b
c
d
© 2002 Nature Publishing Group http://www.nature.com/naturemedicine

NATURE MEDICINE • VOLUME 8 • NUMBER 11 • NOVEMBER 2002 1299
ARTICLES
Fig. 4 Both ATL and ANXA1 are generated in vivo and interact
with ALXR. a, Adherent human PMNs (5 × 10
6
) incubated (in 6-
well plates) in presence of 0.1 µM PMA or 0.1 µM fMLP for 30
min at 37 °C. b, Murine blood-borne PMNs (BL) or air pouch
(AP) cells (>95% PMNs) collected 4 h after injection of 1 mg
Zymosan A were immunoprecipitated (IP) with either the mono-
clonal antibody against ANXA1 (top) or polyclonal antibody
against ALXR peptide (bottom). Immunoprecipitated proteins
were probed with anti-ALXR (top) or anti-ANXA1 (bottom) by
western blotting (WB). Immunoblots represent 3 independent
experiments. DEX or ASA were injected into murine air pouches
with the indicated doses 15 min before Zymosan A (1 mg) injec-
tion. c–e, After 4 or 16 h, PMN accumulation (c), 15-epi-LXA
4
(d)
and ANXA1 generation (e) were determined. Insets, Representative western
blot exudates at 4 and 16 h. Results are the mean ± s.e.m of n = 8 per group.
The direct interaction between ANXA1 and ALXR/FPRL1 was
further substantiated via immunoprecipitation with human
PMNs. PMNs were activated with either PMA or fMLP to adhere
cells and bring ANXA1 on the cell surface
24
to detect ANXA1 in-
teraction with its putative receptor. With adherent PMNs, im-
munoreactive bands of ∼70 kD, corresponding to the mass of
human ALXR/FPRL1 (ref. 28), was co-immunoprecipitated by
specific antibodies against ANXA1, demonstrated by western-
blot analysis using specific antibody against ALXR (Fig. 4a,
upper panel). Also, immunoreactive bands of ∼36 kD, consistent
with molecular masses of ANXA1, were obtained by co-immuno-
precipitation with anti-ALXR, followed by western blot using
anti-ANXA1 (Fig. 4a, lower panel). In comparison, ANXA1 and
ALXR/FPRL1 were not co-immunoprecipitated with non-adher-
ent PMNs (control; see Fig. 4a). The need for PMN activation to
bring about ANXA1–ALXR/FPRL1 interaction was also seen with
murine cells, as a much stronger signal was detected in inflamed
air-pouch cells compared with blood-borne PMNs (Fig. 4b).
Thus, these results indicate direct protein–protein interaction
between endogenous ANXA1 and ALXR/FPRL1 when PMNs ad-
here in vitro and extravasate in vivo.
Generation of endogenous ATL and ANXA1
Next, we examined whether these anti-inflammatory mediators
(for example, ANXA1 and ATL) are generated in murine air-
pouch exudates when DEX and ASA are administered. In combi-
nation, low doses of DEX (3 µg) and ASA (300 µg) gave 42%
inhibition of Zymosan A-initiated PMN infiltration at both 4
and 16 hours (Fig. 4c). 15-epi-LXA
4
(ATL) was generated and pre-
sent in ASA-treated animals at 4 hours that diminished by 16
hours (Fig. 4d). Without administration of ASA, animals gave
very low 15-epi-LXA
4
(∼500 and ∼200 pg per pouch at 4 and 16
hours, respectively). In this regard, Zymosan A can stimulate
some 15-epi-LXA
4
generation via ASA-independent pathways, as
its precursor 15R-HETE is also produced by p450-dependent
mechanisms, and lipoxygenases can produce low levels of the R
epimers
29
. In comparison, ANXA1 generation was low at 4 hours
and much higher at 16 hours (Fig. 4e). Together, these results in-
dicate different time courses for the generation of lipid (that is,
ATL) versus protein (ANXA1) mediators during the host inflam-
matory response, and they suggest that ALXR/FPRL1 could regu-
late PMNs by interacting with each class of ligands within
specific phases of the inflammatory response. This spatial and
temporal difference between the two ligand-receptor pairs in vivo
likely explains the K
d
values for each observed in vitro (Figs. 2 and
4). As previously suggested
24,25
, a further consideration is needed
regarding the bioactive levels of ANXA1 and its peptides in vivo:
it is still speculative but highly likely that the actual amounts of
ANXA1 and its peptides within the microenvironment where
they are generated (that is, by the adherent PMNs and surround-
ing tissue matrix) may be much higher than that measured
within the solubilized materials quantified in biological fluids.
Synergism between ATLa and ANXA1 peptide
In view of these considerations, we tested in vivo whether ad-
ministration of exogenous ATL could inhibit Zymosan A-initi-
ated PMN infiltration and if ANXA1-derived peptides share this
inhibitory property. Intravenous delivery of ANXA1-derived
peptides (for example, Ac2-26, Ac2-12 and Ac2-6) inhibited PMN
infiltration in a dose-dependent fashion whereas the non-physi-
ologically scrambled Ac2-6 peptide sequence, which did not
compete at ALXR/FPRL1 (Fig. 2b), did not share this action (Fig.
5a). Intravenous delivery of the stable analog of ATL (5 µg per
mouse; ∼12 nmol), denoted ATL analog (ATLa; 15-epi-16-(para-
fluoro)-phenoxy-LXA
4
) gave ∼40% inhibition comparable with
that obtained with 100 µg (∼33 nmol) peptide Ac2-26 (see Fig.
5b) (compare inhibition of TNF-α-induced PMN infiltration
30
).
Functional synergism was observed with lower doses of ATLa (1
µg or 2.4 nmol) and peptide Ac2-26 (10 µg or 3.3 nmol) (Fig. 5b).
Results in Fig. 5b indicate that when administered, ATLa and
a b c
d e
*, P < 0.05 versus Zymosan A alone. For c–e: , Zymosan alone; , + 3 µg
DEX; , + 300 µg ASA; , + 3 µg DEX + 300 µg ASA.
© 2002 Nature Publishing Group http://www.nature.com/naturemedicine

1300 NATURE MEDICINE • VOLUME 8 • NUMBER 11 • NOVEMBER 2002
ARTICLES
Fig. 5 Synergism with ANXA1-de-
rived peptides and ATLa in vivo.
a, ANXA1-derived peptides were ad-
ministered at the reported doses via
the tail vein 15 min prior to injection
of 1 mg Zymosan A into 6-d-old air
pouches. PMN accumulation was
measured 4 h later. The number of mi-
grated PMNs in the control group (ve-
hicle-treated mice) was 7.5 ± 0.5 ×10
6
PMNs per mouse. Data are mean ±
s.e.m of 6–12 mice; *, P < 0.05 versus
control migration as calculated on
original values. , Ac2-26; , Ac2-12;
, Ac2-6; , Ac2-6 scramble.
b, Animals were treated i.v. with the
reported doses of ATLa (; 0.5–5 µg,
1.2–12 nmol; chemical structure
shown, R = para-fluoro-phenoxy),
Ac2–26 (; 5–100 µg, corresponding
to ∼1.5–33 nmol) or both ATLa and
Ac2-26 () 15 min before local injec-
tion of Zymosan A. PMN accumula-
tion was measured after 4 h. The
number of migrated PMNs in the con-
trol group (mice treated with
Zymosan A alone) was 6.5 ± 0.55 × 10
6
PMNs per mouse. Data are mean ± s.e.m of n = 6–8
mice per group; *, P < 0.05 versus control migration as calculated on original values. c, Air-
pouch biopsies. Sections from the top dose groups (that is, 5 µg of ATLa and 100 µg of Ac2-
26; see b) were prepared and were stained with H&E. Magnifications, ×10 for the top panel
and ×40 for the inset and lower panels.
Ac2-26 were each potent inhibitors of PMN infiltration, and syn-
ergism was achieved with a combination of sub-threshold doses
of both agents. Inhibition with ATL and ANXA1-derived pep-
tides was also demonstrated with other markers of inflammation
including macrophage inflammatory protein-1α (MIP-1α) and
prostaglandin E
2
(PGE
2
), which were both inhibited in these
mice (Supplementary Fig. C online).
Histological analysis showed that Zymosan A gave a markedly
increased number of PMNs in the skin-tissue linings surround-
ing the pouch cavity (Fig. 5c); this number was reduced by ATLa,
Ac2-26 and their combination. The degree of PMN tissue traf-
ficking was reduced to the level observed without Zymosan A
challenge. These findings, together with the radioligand binding
and immunoprecipitation results (Fig. 2 and 4), indicate that a
shared site of action on PMNs is responsible for the anti-PMN
properties of ATL and ANXA1 as well as their mimetics. Recently
it is reported that a phosphatidylserine-specific receptor was crit-
ical for the clearance of apoptotic cells and that this may be a
critical point for modulating resolution
31
. Along these lines,
lipoxin A
4
stimulates monocytes/macrophages in a nonphlogis-
tic fashion to enhance the uptake of apoptotic neutrophils
32
, an
activity also shared by glucocorticoids
33
.
Discussion
Production of ANXA1-derived peptides is activated in vivo dur-
ing the process of PMN extravasation
23
, and within bronchoalve-
olar lavage fluids from patients with cystic fibrosis, where
processing of ANXA1 at its N-terminal portion seems to be
cleaved by PMN-derived elastase
34
. Notably, several other natu-
rally occurring peptides also exhibit inhibitory actions in cell ad-
hesion and migration with other leukocyte subclasses.
Interleukin-2 (IL-2) peptides, for example, generated by PMN
elastase inhibit IL-2 induced T-cell adhesion and migration
35
.
Thus, endogenously generated peptides derived during inflam-
mation or from inflammation-induced proteins may serve as
feedback inhibitors to terminate and/or halt the amplification of
inflammation and could work together with lipid mediators
such as LXA
4
and aspirin-triggered lipid mediators to promote
resolution.
In summary, our results are the first to indicate that both
ANXA1 and ATL directly interact with human ALXR/FPRL1. By
convergence at the same anti-inflammatory receptor, these two
structurally distinct endogenous systems, namely lipid-derived
(for example, ATL) and protein-derived (for example, ANXA1)
mediators, limit PMNs in vivo. This overlap may have evolved to
ensure that inflammation loci remain local, walled off and self-
limited, as well as to protect from self-damage from within.
Because several related sequences are present in this cluster of
GPCRs (ref. 36) (for example, FPRL2 etc.), it is likely in view of
these results and those of others that additional endogenous lig-
ands may share these functions. Moreover, impingement of
these natural endogenous systems by popular therapies such as
ASA and DEX may, at least in part, underlie the proven efficacy
of the combination of DEX and ASA in controlling rheumatic
diseases
13
and the success of their individual use this past cen-
tury. These systems likely represent functional redundancies in
endogenous anti-inflammation circuits that unveil presently un-
appreciated mechanisms operative in governing PMN responses
in host defense, and they now open new avenues for approaches
such as combination therapies.
Methods
Inflammation in vivo. PMN migration and mediator formation were as-
sessed using 6-day-old murine air pouches in Swiss Albino mice (n = 8 per
group). After Harvard Medical Area IRB approval (Protocol #02570), and
approval by the Home Office UK Project License (PPL70/4804), compounds
a
b
c
© 2002 Nature Publishing Group http://www.nature.com/naturemedicine

NATURE MEDICINE • VOLUME 8 • NUMBER 11 • NOVEMBER 2002 1301
ARTICLES
(for example, ASA, DEX, ATLa, ANXA1 and ANXA1 mimetics) were deliv-
ered as bolus injections either into the tail vein in 100 µl PBS (ATLa, ANXA1
and its mimetics) or locally into air pouches in 100 µl PBS (ASA and DEX).
Zymosan A (1 mg/0.5 ml saline) was locally injected into the air pouches 15
min later to induce inflammation
37
. 4 or 16 h later, air pouches were
washed with 2 ml PBS containing 25 U/ml heparin, and PMNs were
counted by light microscopy following staining in Turk’s solution. Cell-free
lavage fluids were used for quantification of mouse MIP-1α and PGE
2
using
available ELISAs (R&D System, Oxford, UK and Amersham,
Buckinghamshire, UK). 15-epi-LXA
4
generation was also determined using
an ELISA specific for the 15R epimer as described
29
. ANXA1 generation was
determined by western-blot analysis using specific antibodies to ANXA1.
For microscopic analysis, tissues were obtained with a 6-mm tissue biopsy
punch and fixed in 10% buffered formaldehyde. Samples were then em-
bedded in paraffin, sliced and stained with H&E.
Preparation of labeled ligands. [11,12-
3
H]LXA
4
-methyl ester was pre-
pared with Schering AG (Berlin, Germany) using acetylenic-LXA
4
-methyl
ester as precursor (prepared with N.A. Petasis) essentially as described
18
,
and was a gift of H.D. Perez. [11,12-
3
H]LXA
4
-methyl ester was isolated and
purified using a Hewlett Packard (San Fernando, California) 1100 Series
Diode Array Detector (DAD) equipped with a binary pump and eluted on a
Beckman (Fullerton, California) Ultrasphere C18 column (250 × 4.5 mm,
5 µm)
18
. To minimize chemical degradation of [11,12-
3
H]LXA
4
-methyl ester,
it was used immediately after HPLC chromatography for binding experi-
ments. [
125
I-Tyr]Ac2-26 was prepared by custom synthesis with Phoenix
Pharmaceuticals (Belmont, California).
Radioligand binding. Human PMNs were obtained from donors according
to the Brigham and Women’s Hospital Human Research Committee
Protocol #88-02642 and the Ethical Committee (UK) ELCHAP/00/029.
Informed consent was obtained from all subjects. [
3
H]LXA
4
, [
3
H]LTB
4
and
[
125
I-Tyr]Ac2-26 binding was performed with freshly isolated human PMNs
or HEK293 cells (American Type Culture Collection, Rockville, Maryland)
transfected with human recombinant ALXR (GenBank Accession #U81501)
or BLT
18
. Cells were suspended in Dulbecco’s PBS with CaCl
2
and MgCl
2
(DPBS
++
). Aliquots (0.5 × 10
6
/ml HEK293 or 5 × 10
6
/ml PMNs) were incu-
bated with 1 nM of [11,12-
3
H]LXA
4
(∼60,000 c.p.m., specific activity ∼10
Ci/mmol), [5,6,8,9,11,12,14,15-3H]LTB
4
(∼20,000 c.p.m., specific activity
∼200 Ci/mmol, from NEN, Boston, Massachusetts) or 30 nM [
125
I-Tyr]Ac2-
26 (∼80,000 c.p.m., specific activity ∼1,171.5 Ci/mmol), in the absence or
presence of increasing concentrations of unlabeled compounds for 40 min
at 4 °C. The bound and unbound radioligands were separated by filtration
through Whatman GF/C glass microfiber filters (Fisher, Pittsburgh,
Pennsylvania)
18
. Non-specific binding was determined in the presence of
1 µM of unlabeled homoligands. To obtain adherent PMNs, freshly isolated
PMNs were plated in collagen-coated 6-well plates (5 × 10
6
/ml) in the pres-
ence of low-dose (1 nM) PMA for 30 min. Cells were washed with DPBS and
[
125
I-Tyr]Ac2-26 binding was carried out with adherent cells using the con-
ditions described above. The unbound radioligands were aspirated and
cells washed with DPBS. The adherent cells were then lysed with scintilla-
tion mixture and radioactivity determined.
Chemotaxis. Chemotaxis was carried out using a microchamber technique
(chemotaxis chamber from Neuro Probe, Cabin John, Maryland)
18
. Chinese
hamster ovary (CHO) cells expressing ALXR and Gqo chimera (from B.
Conklin) or freshly prepared human PMNs (ref. 24) were incubated with ve-
hicle or peptide Ac2-26 for 30 min at 37 °C and added to the upper wells of
chemotaxis chamber (5 × 10
4
cells per well). After incubation for 4 h at 37
°C, polycarbonate membrane was removed, stained with a Diff-Quick stain-
ing kit (Dade Behring, Newark, Delaware) and migrated cells enumerated.
Chemotaxis index was calculated as the mean number of cells that mi-
grated toward medium containing chemoattractant solution divided by the
mean number of cells migrated toward the medium containing vehicle
only.
Ca
2+
mobilization. Changes in the cytosolic-free calcium concentration
were monitored in human PMNs loaded with 1 µM fura 2/ace-
toxymethylester (Molecular Probes, Leiden, Holland) at 37 °C for 1 h. Fura
2 fluorescence was performed with aliquots of 5 × 10
6
PMNs in 2 ml HBSS,
using a fluorimeter (Jobin Yvon 3D, Lonjumeau, France), equipped with a
thermally controlled cuvette holder and a magnetic stirrer.
Immunoprecipitation and western blotting. PMNs were freshly prepared
from healthy volunteers by histopaque (Sigma-Aldrich Ltd., Poole, UK) gra-
dient separation
24
. Mouse blood-derived PMNs were obtained by negative
immunomagnetic separation
38
, whereas extravasated cells were collected
from the air pouch exudates (see below). Human PMNs were plated in 6-
well plates in the absence or presence of 0.1 µM PMA, or 0.1 µM fMLP to
produce ∼80% adhesion (15 min at 37 °C). In all cases, cell extracts pre-
pared from adherent PMNs were immunoprecipitated with either 10 µl rab-
bit serum against human ALXR peptide [ASWGGTPEERLK] at the second
extracellular loop
28
or 5 µl antibody against ANXA1 (ref. 24) in the presence
of phosphatase and proteases inhibitors. Immunoprecipitated proteins
were subject to electrophoresis on a 12% SDS–PAGE gel and transferred
onto ECL Hybond nitrocellulose membrane (Amersham, Buckinghamshire,
UK). Membranes were further hybridized for ANXA1 or ALXR detection
using the specific antibodies: anti-ANXA1 (1:10,000) or anti-ALXR rabbit
serum (1:13,000). Immunoreactive proteins were detected using an en-
hanced chemiluminescence ECL kit from Amersham (UK).
Statistical analysis. Results were expressed as the mean ± s.e.m. of distinct
experiments (performed in triplicate) or in animals. In vitro experiments
were analyzed by Student’s t-test with P < 0.05 taken as statistically signifi-
cant. Analysis of in vivo data was done with one-way ANOVA plus Dunnett’s
test.
Note: Supplementary information is available on the Nature Medicine website.
Acknowledgment
We thank B. Schmidt for microscopic analyses, N. Petasis for synthetic LXA
4
and ATL stable analogs (prepared for P01-DE13499), and M.H. Small for
expert assistance in manuscript preparation. This work was supported in part
by grants GM38765 and P01-DE13499 (to C.N.S.) and by grants P0567 and
P0583 of the Arthritis Research Campaign UK (to M.P.).
Competing interests statement
The authors declare that they have no competing financial interests.
RECEIVED 26 JULY 2002; ACCEPTED 18 SEPTEMBER 2002
1. Cotran, R.S. Inflammation: Historical perspectives. in Inflammation: Basic Principles
and Clinical Correlates 3rd edn. (eds. Gallin, J.I., Snyderman, R., Fearon, D.T.,
Haynes, B.F. & Nathan, C.) 5–10 (Lippincott Williams & Wilkins, Philadelphia,
1999).
2. Weissmann, G. Aspirin. Sci. Am. 264, 84–90 (1991).
3. Patrono, C. Aspirin: new cardiovascular uses for an old drug. Am. J. Med. 110,
62S–65S (2001).
4. Barnes, C.J., Hardman, W.E., Cameron, I.L. & Lee, M. Aspirin, but not sodium sali-
cylate, indomethacin, or nabumetone, reversibly suppresses 1,2-dimethylhy-
drazine-induced colonic aberrant crypt foci in rats. Dig. Dis. Sci. 42, 920–926
(1997).
5. Marcus, A.J. Aspirin as prophylaxis against colorectal cancer. N. Engl. J. Med. 333,
656–658 (1995).
6. Samuelsson, B. From studies of biochemical mechanisms to novel biological medi-
ators: Prostaglandin endoperoxides, thromboxanes and leukotrienes. in Les Prix
Nobel: Nobel Prizes, Presentations, Biographies and Lectures 153–174 (Almqvist &
Wiksell, Stockholm, 1982).
7. Vane, J.R. Inhibition of prostaglandin synthesis as a mechanism of action for aspirin-
like drugs. Nature (London) New Biol. 231, 232–235 (1971).
8. Hench, P.S., Kendall, E.C., Slocumb, C.H. & Polley, H.E. Proc. Staff Meet. Mayo Clin.
24, 181 (1949).
9. Boumpas, D.T. & Wilder, R.L. Corticosteroids. in Arthritis and Allied Conditions: A
Textbook of Rheumatology 14th edn. edn (ed. Koopman, W.J.) 827–847 (Lippincott
Williams & Wilkins, Philadelphia, 2001).
10. Derry, S. & Loke, Y.K. Risk of gastrointestinal haemorrhage with long term use of
aspirin: meta-analysis. Br. Med. J. 321, 1183–1187 (2000).
11. Cronstein, B.N. & Weissmann, G. Targets for antiinflammatory drugs. Annu. Rev.
Pharmacol. Toxicol. 35, 449–462 (1995).
12. Sin, Y.M., Sedgwick, A.D., Chea, E.P. & Willoughby, D.A. Mast cells in newly
formed lining tissue during acute inflammation: A six day air pouch model in the
mouse. Ann. Rheum. Dis. 45, 873–877 (1986).
13. Jick, H., Pinals, R.S., Ullian, R., Slone, D. & Muench, H. Dexamethasone and dex-
© 2002 Nature Publishing Group http://www.nature.com/naturemedicine

1302 NATURE MEDICINE • VOLUME 8 • NUMBER 11 • NOVEMBER 2002
ARTICLES
amethasone-aspirin in the treatment of chronic rheumatoid arthritis. Lancet 2,
1203–1205 (1965).
14. Flower, R.J. & Rothwell, N.J. Lipocortin-1: Cellular mechanisms and clinical rele-
vance. Trends Pharmacol. Sci. 15, 71–76 (1994).
15. Lim, L.H., Solito, E., Russo-Marie, F., Flower, R.J. & Perretti, M. Promoting detach-
ment of neutrophils adherent to murine postcapillary venules to control inflamma-
tion: effect of lipocortin 1. Proc. Natl. Acad. Sci USA 95, 14535–14539 (1998).
16. Maridonneau-Parini, I., Errasfa, M. & Russo-Marie, F. Inhibition of O
2
-generation by
dexamethasone is mimicked by lipocortin I in alveolar macrophages. J. Clin. Invest.
83, 1936–1940 (1989).
17. Serhan, C.N., Takano, T., Chiang, N., Gronert, K. & Clish, C.B. Formation of en-
dogenous “antiinflammatory” lipid mediators by transcellular biosynthesis:
Lipoxins and aspirin-triggered lipoxins inhibit neutrophil recruitment and vascular
permeability. Am. J. Respir. Crit. Care Med. 161, S95–S101 (2000).
18. Chiang, N., Fierro, I.M., Gronert, K. & Serhan, C.N. Activation of lipoxin A
4
recep-
tors by aspirin-triggered lipoxins and select peptides evokes ligand-specific re-
sponses in inflammation. J. Exp. Med. 191, 1197–1207 (2000).
19. Yokomizo, T., Izumi, T., Chang, K., Takuwa, T. & Shimizu, T. A G-protein-coupled
receptor for leukotriene B
4
that mediates chemotaxis. Nature 387, 620–624 (1997).
20. Gabay, C. & Kushner, I. Acute-phase proteins and other systemic responses to in-
flammation. N. Engl. J. Med. 340, 448–454 (1999).
21. Su, S.B. et al. A seven-transmembrane, G protein-coupled receptor, FPRL1, medi-
ates the chemotactic activity of serum amyloid A for human phagocytic cells. J. Exp.
Med. 189, 395–402 (1999).
22. Linke, R.P., Bock, V., Valet, G. & Rothe, G. Inhibition of the oxidative burst response
of N-formyl peptide-stimulated neutrophils by serum amyloid-A protein. Biochem.
Biophys. Res. Commun. 176, 1100–1105 (1991).
23. Oliani, S.M., Paul-Clark, M.J., Christian, H.C., Flower, R.J. & Perretti, M. Neutrophil
interaction with inflamed postcapillary venule endothelium alters annexin 1 ex-
pression. Am. J. Pathol. 158, 603–615 (2001).
24. Perretti, M. et al. Mobilizing lipocortin 1 in adherent human leukocytes downregu-
lates their transmigration. Nature Med. 2, 1259–1262 (1996).
25. Walther, A., Riehemann, K. & Gerke, V. A novel ligand of the formyl peptide recep-
tor: annexin I regulates neutrophil extravasation by interacting with the FPR. Mol.
Cell. 5, 831–840 (2000).
26. Perretti, M., Getting, S.J., Solito, E., Murphy, P.M. & Gao, J.L. Involvement of the
receptor for formylated peptides in the in vivo anti-migratory actions of annexin 1
and its mimetics. Am. J. Pathol. 158, 1969–1973 (2001).
27. Papayianni, A., Serhan, C.N. & Brady, H.R. Lipoxin A
4
and B
4
inhibit leukotriene-
stimulated interactions of human neutrophils and endothelial cells. J. Immunol.
156, 2264–2272 (1996).
28. Kang, Y., Taddeo, B., Varai, G., Varga, J. & Fiore, S. Mutations of serine 236–237
and tyrosine 302 residues in the human lipoxin A4 receptor intracellular domains
result in sustained signaling. Biochemistry 39, 13551–13557 (2000).
29. Chiang, N. et al. Aspirin-triggered 15-epi-lipoxin A
4
(ATL) generation by human
leukocytes and murine peritonitis exudates: Development of a specific 15-epi-LXA
4
ELISA. J. Pharmacol. Exp. Ther. 287, 779–790 (1998).
30. Clish, C.B. et al. Local and systemic delivery of a stable aspirin-triggered lipoxin pre-
vents neutrophil recruitment in vivo. Proc. Natl. Acad. Sci. USA 96, 8247–8252
(1999).
31. Fadok, V.A. et al. A receptor for phosphatidylserine-specific clearance of apoptotic
cells. Nature 405, 28–29 (2000).
32. Godson, C. et al. Cutting edge: Lipoxins rapidly stimulate nonphlogistic phagocy-
tosis of apoptotic neutrophils by monocyte-derived macrophages. J. Immunol. 164,
1663–1667 (2000).
33. Liu, Y. et al. Glucocorticoids promote nonphlogistic phagocytosis of apoptotic
leukocytes. J. Immunol. 162, 3639–3646 (1999).
34. Tsao, F.H.C., Meyer, K.C., Chen, X., Rosenthal, N.S. & Hu, J. Degradation of an-
nexin I in bronchoalveolar lavage fluid from patients with cystic fibrosis. Am. J.
Respir. Cell Mol. Biol. 18, 120–128 (1998).
35. Ariel, A. et al. IL-2 induces T cell adherence to extracellular matrix: Inhibition of ad-
herence and migration by IL-2 peptides generated by leukocyte elastase. J.
Immunol. 161, 2465–2472 (1998).
36. Yokomizo, T., Izumi, T. & Shimizu, T. Leukotriene B
4
: Metabolism and signal trans-
duction. Arch. Biochem. Biophys. 385, 231–241 (2001).
37. Perretti, M. et al. Acute inflammatory response in the mouse: Exacerbation by im-
munoneutralization of lipocortin 1. Br. J. Pharmacol. 117, 1145–1154 (1996).
38. Cotter, M.J., Norman, K.E., Hellewell, P.G. & Ridger, V.C. A novel method for isola-
tion of neutrophils from murine blood using negative immunomagnetic separa-
tion. Am. J. Pathol. 159, 473–481 (2001).
© 2002 Nature Publishing Group http://www.nature.com/naturemedicine