Primary afferent second messenger cascades interact with specific integrin subunits in producing inflammatory hyperalgesia.

ger
ng
, Je
ing,
f Neu
University of California at San Francisco, Campus Box 0440, Room C-555 521, San Francisco, CA 94143-0440, USA
of sensory neurons. Integrins mediate bidirectional trans- 1992; Tomaselli et al., 1993), a2 (Andrew et al., 1992;
Pain 115 (2005)E-mail address: levine@itsa.ucsf.edu (J.D. Levine).Received 29 November 2004; received in revised form 11 February 2005; accepted 22 February 2005
Abstract
We recently reported that hyperalgesia induced by the inflammatory mediator prostaglandin E2 (PGE2) requires intact a1, a3 and b1
integrin subunit function, whereas epinephrine-induced hyperalgesia depends on a5 and b1. PGE2-induced hyperalgesia is mediated by
protein kinase A (PKA), while epinephrine-induced hyperalgesia is mediated by a combination of PKA, protein kinase C3 (PKC3) and
mitogen-activated protein kinase/extracellular signal-regulated kinase (MAPK/ERK). We hypothesized that inflammatory mediator-induced
hyperalgesia involves specific interactions between different subsets of integrin subunits and particular second messenger species. In the
present study, function-blocking anti-integrin antibodies and antisense oligodeoxynucleotides were used to elucidate these interactions in rat.
Hyperalgesia produced by an activator of adenylate cyclase (forskolin) depended on a1, a3 and b1 integrins. However, hyperalgesia induced
by activation of the cascade at a point farther downstream (by cAMP analog or PKA catalytic subunit) was independent of any integrins
tested. In contrast, hyperalgesia induced by a specific PKC3 agonist depended only on a5 and b1 integrins. Hyperalgesia induced by agonism
of MAPK/ERK depended on all four integrin subunits tested (a1, a3, a5 and b1). Finally, disruption of lipid rafts antagonized hyperalgesia
induced by PGE2 and by forskolin, but not that induced by epinephrine. Furthermore, a1 integrin, but not a5, was present in detergent-
resistant membrane fractions (which retain lipid raft components). These observations suggest that integrins play a critical role in
inflammatory pain by interacting with components of second messenger cascades that mediate inflammatory hyperalgesia, and that such
interaction with the PGE2-activated pathway may be organized by lipid rafts.
q 2005 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.
Keywords: Inflammation; Cytoskeleton; Extracellular Matrix; Caveolae; Cholesterol; Sphingomyelin
1. Introduction
Inflammatory mediators, acting through G-protein-
coupled receptors, sensitize the peripheral terminals of
primary afferent nociceptors to produce a decrease in pain
threshold (Reichling and Levine, 1999). Inflammation also
produces changes in the extracellular matrix (Bradley et al.,
2000; Wallquist et al., 2002) which surrounds the terminals
and intracellular second messenger pathways (Hynes,
2002), and we recently reported that integrin function in
nociceptive sensory neurons is essential in the development
and maintenance of inflammatory pain (Dina et al., 2004).
Integrin receptors are heterodimers composed of an a
and b subunit (Humphries, 2000). At least seven different a
integrin subunits are expressed in mammalian sensory
neurons or sensory nerve fibers, including a1 (Andrew et al.,Primary afferent second messen
integrin subunits in produci
Olayinka A. Dina, Tim Hucho
David B. Reichl
Departments of Medicine and Oral and Maxillofacial Surgery, Division ocascades interact with specific
inflammatory hyperalgesia
nny Yeh, Misbah Malik-Hall,
Jon D. Levine*
roscience and Biomedical Sciences Program, NIH Pain Center (UCSF),
191–203
www.elsevier.com/locate/pain1993), a6 (Andrew et al., 1992; Tomaselli et al., 1993), and
a7 (Werner et al., 2000). In a recent study, we found that
hyperalgesia produced by PGE2 requires intact function of
0304-3959/$20.00 q 2005 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.pain.2005.02.028
* Corresponding author. Tel.: 1 415 476 5108; fax: C1 415 476 6305.membrane signaling between the extracellular matrix Khalsa et al., 2000; Tomaselli et al., 1993), a3 (Andrew
et al., 1992; Tomaselli et al., 1993), a5 (Tomaselli et al.,

et al., 1999b). Hyperalgesia induced by epinephrine is
second messenger cascades such interactions with integrins
saline. Stock solutions of forskolin, 8-Br-cAMP, cPKA and
integrin subunit (clone Ralph 3.2) was obtained from Santa Cruz
ain 112. Methods
2.1. Animals
Male Sprague–Dawley rats (200–300 g; Charles River, Hollister,
CA) were used in these experiments. The rats were housed in the
Animal Care Facility of the University of California, San Francisco,
under a 12-h light/dark cycle with food and water available ad
libitum. Care and use of rats conformed to National Institutes of
Health guidelines and were approved by the University of California,
San Francisco Committee on Animal Research. All efforts were made
to minimize number of animals used and their suffering.
2.2. Mechanical nociceptive threshold testing
The nociceptive flexion reflex was quantified using the
Randall–Selitto paw pressure test (Randall and Selitto, 1957) in
which a linearly increasing mechanical force (Analgesymeter,
Stoelting, Chicago, IL) is applied to the dorsum of the rat’s hind
paw (Taiwo et al., 1989). The nociceptive threshold was defined as
the force in grams at which the rat withdrew its paw, and baseline
paw-withdrawal threshold was defined as the mean of three
readings before test agents were injected. Each paw was treated as
an independent measure and each experiment was performed on a
separate group of rats. Except where stated otherwise, rats were
treated with only one agonist or antagonist injected intradermally
in the dorsum of the hind paw. Nociceptive threshold was
measured 30 min after the administration of hyperalgesic agents.occur.mediated by PKC3, extracellular signal-regulated kinases
1 and 2 (ERK1/2), and PKA (Aley et al., 2001a; Cesare
et al., 1999; Dina et al., 2001; Khasar et al., 1999a). These
observations suggest that integrins do not broadly control
nociception, but rather that specific integrin subunits
interact with particular second messenger species. The
present study employed specific second messenger agonists
and integrin antagonists as well as membrane fractionation
methods to elucidate at what stage in the hyperalgesica1, a3 and b1 integrin subunits, while hyperalgesia induced
by epinephrine requires a5 and b1 (Dina et al., 2004). The
nature of interactions between these integrin subunits and
second messenger pathways that underlie the hyperalgesia-
inducing effects of these agents are not known.
Integrins are capable of interacting with multiple second
messenger signaling systems that can mediate hyperalgesia
and sensitization of nociceptors, including PKA (Dormond
et al., 2002; Meyer et al., 2000; Rangarajan et al., 2003;
Suzuki et al., 2002), PKC3 (Berrier et al., 2000; Ivaska et al.,
2003) and MAPK (Howe et al., 2002; Short et al., 2000;
Stupack and Cheresh, 2002). Hyperalgesia induced by
prostaglandin E2 (PGE2) is mediated by cAMP-activated
PKA (Aley and Levine, 1999; Dina et al., 2001; Khasar
O.A. Dina et al. / P192All behavioral testing was performed between 10 and 4 pm.Biotechnology, Santa Cruz, CA. Monoclonal antibodies were
dissolved in distilled water to 1 mg/10 ml (Dina et al., 2004).
2.5. Antisense oligodeoxynucleotides
Antisense and mismatch oligodeoxynucleotides (ODN) target-
ing a1, a3, a5 or b1 integrin subunits were purchased from
Invitrogen, San Francisco, CA. The antisense ODN sequence,
corresponding GenBank accession number, and ODN position
within the cDNA sequence, corresponding mismatch sequence and
the number of mismatched bases (mismatched bases denoted by
bold face) for each integrin subunit are given as follows: rat a1
(23-mer, 5 0-CAA CAT TGA AGG AGA CGC AGA AG-3 0,
X52140 and 491–513, 5 0-CAT GAT TCT AGC TGA CCC TGA
AG-3 0, eight bases mismatched); mouse a3 (23-mer, 5 0-GTG GAG
GAT TTA GAA CGG CAA GC-3 0, AB080229 and 438–460,
5 0-GTC GAG CAT TTT GAT CGG CAT GGK3 0, six bases
mismatched); human a5 (21-mer, 5 0-CTG AGA ATC CGA AGA
AGG AGC-3 0, NM 002205 and 192–213, 5 0-CTG TCA ATG CGA
TGA ACG TGG-3 0, seven bases mismatched) and rat b1 (24-mer,
5 0-CAA ATT CAT CTT TTC GCA GCG TCC-3 0, U12309 and
30–53, 5 0-CAA TTT GTT GAT TTG CCA GCG TCC-3 0, seven
bases mismatched). A search of the National Center for
Biotechnology Information database identified no other sequences
in Rattus norvegicus homologous to those used in this experiment.
Prior to use, ODN was lyophilized, reconstituted in nuclease-freepseudo3RACK were stored at K20 A˚C while stock solutions of
recombinant active MEK1C were stored at K80 A˚C.
2.4. Anti-integrin monoclonal antibodies
Monoclonal antibodies against the a1 (clone Ha31/8), a5
(HMa5-1), and b1 (Ha2/5) integrin subunits were obtained from
PharMingen, San Diego, CA. Monoclonal antibody against the a32.3. Drugs
The adenylate cyclase activator forskolin, and 8-bromo-
adenosine 3,5 0-cyclic monophosphate (8-Br-cAMP) were obtained
from Sigma. The catalytic subunit of protein kinase A (cPKA), a
38-kDa serine/threonine protein kinase (Slice and Taylor, 1989),
was obtained from New England BioLabs, Beverly, MA. Pseudo-
receptor octapeptide, NH2-HDAPIGYD-COOH, for activated
protein kinase C epsilon (pseudo3RACK) (Dorn et al., 1999) was
synthesized by SynPep, Dublin, CA. Dominant active rabbit MAP
kinase kinase (MEK1C) (Saxena et al., 1999) was obtained from
Upstate Biotechnology, Lake Placid, NY. To facilitate the entry of
the membrane impermeant peptides into the peripheral terminals of
neurons in the skin, these agents are preceded by 2.5 ml distilled
water in the same syringe to cause a transient, hypo-osmotic
permeabilization (Borle and Snowdowne, 1982; Burch and
Axelrod, 1987).
Stock solutions of the hyperalgesic agents were: 1 mg/ml
forskolin in dimethylsulfoxide/saline; and 1 mg/ml 8-Br-cAMP,
100 U/ml cPKA, 1 mg/ml pseudo3RACK and 1 U/ml MEK1C in
physiological saline. Final dilution of forskolin and 8-Br-cAMP
(1 mg/2.5 ml) was in distilled water, while cPKA (10 U/2.5 ml),
pseudo3RACK (1 mg/2.5 ml) and MEK1C (1 U/2.5 ml) were in
5 (2005) 191–2030.9% NaCl to a concentration of 10 mg/ml and stored at K20 8C.

probed for the a1, a3, or a5 integrin protein using SDS-PAGE
dorsal root ganglia (L1–L6) were dissected from pentobarbitol-
homogenization buffer, re-pelleted (1 h, 55,000 rpm, 4 8C), and
finally resuspended in 40 ml 4! Laemmli buffer.
ain 11anesthetized rats, desheathed, incubated 1 h in 0.125% collagenase
at 37 8C, followed by 7-min in 0.25% trypsin at 37 8C, and
triturated. Debris was removed by straining through a 40-mm mesh,
followed by centrifugation (3 min, 500 g). Cells were resuspended
in 15 ml Neurobasal A/B27-media plus 1 mM ethanol (Gibco),
plated onto polyornithine/laminin coated dishes and incubated 12 h
at 37 8C in 95% O2/5% CO2.
Cells were solubilized by scraping into homogenization bufferusing Western blot protocols previously described (Dina et al.,
2004). Briefly, to cause accumulation of detectable amounts of
integrin protein, on the first day of ODN injection the saphenous
nerves were ligated 1 cm proximal to the knee-level bifurcation in
anaesthetized rats (Parada et al., 2003). Three days after the last
ODN administration, a 5-mm section of nerve just proximal to the
ligation was removed and sonicated in radioimmunoprecipitation
buffer (100 mM tris–HCl, 10 mM EDTA, 150 mM NaCl, 1%
Nonidet-P40, 1% sodium deoxycholate, 0.5 mM phenylmethyl-
sulfonyl fluoride, 1 mg/ml leupeptin, 1 mg/ml aprotinin, 1 mg/ml
pepstatin). The samples were homogenized, on ice, in radio-
immunoprecipitation buffer and incubated on ice for 30 min before
centrifugation at 14,000 rpm for 30 min at 4 8C. The supernatant
was decanted and assessed for the amount of protein using the
BCA Protein Assay (Pierce Biotechnology, Rockford, IL).
Equal amounts of protein were loaded and separated on a 5%
polyacrylamide gel and transferred electrophoretically to a
polyvinylidenedifluoride membrane (Perkin–Elmer Life Sciences,
Boston, MA). The membrane was probed with an affinity-purified
goat polyclonal anti-a1 integrin (rat origin), anti-a3 (human
origin), anti-a5 (human origin) IgG (Santa Cruz Biotechnology,
Santa Cruz, CA) diluted 1:100, followed by incubation with
horseradish peroxidase-conjugated rabbit anti-goat IgG diluted
1:2000. Protein content was normalized using an antibody against
PKC3 (human origin; Santa Cruz Biotechnology). The same
immunoblot was stripped by antibodies for 30 min at 60 A˚C using
a stripping buffer (62.5 mM tris–HCl, 2% SDS, pH 6.7 containing
0.7% b-mercaptoethanol), treated with anti-PKC3 rabbit IgG and
goat anti-rabbit IgG (1:500 and 1:10,000, respectively, Santa Cruz
Biotechnology). Specific bands were visualized by ECL Western
Blotting Detection Kit (Pierce Biotechnology) and exposure to
X-ray film.
2.7. Detection of integrins in detergent-resistant membranes
Cultures of dissociated dorsal root ganglia were prepared from
male rats as described previously (Khasar et al., 1998b). Briefly,A dose of 40 mg of each ODN was intrathecally administered in a
volume of 20 ml once daily for 3 days. Rats were anesthetized with
2.5% isoflurane inhalation anesthetic (in O2); a 30-gauge needle
was inserted into the subarachnoid space on the midline between
L4 and L5 vertebrae; ODN was injected at 1 ml/s. The animals
regained consciousness approximately 1 min after discontinuation
of anesthesia.
2.6. Western blot analysis of integrin expression
To assess the efficacy of antisense ODN treatment in reducing
expression of integrin subunits, saphenous nerve preparations were
O.A. Dina et al. / P(25 mM Tris, 250 mM NaCl, 1! Complete Protease Inhibitor MixIntegrin subunits were detected in the membrane samples by
Western blot of sodium dodecyl sulfate (SDS) gels. Samples
(20 ml) were separated on a 4–15% gradient SDS minigel, semi-dry
blotted, blocked with 5% Tris-buffered saline Tween-20 (TBST) in
milk, and probed with anti-a1 integrin monoclonal antibody (goat,
1:500, 12 h, Santa Cruz), amplified with horseradish peroxidase-
conjugated secondary antibody (1 h, 1:5000, Pierce), and detected
by BioMax Light film (Kodak) using SuperSignal Pico ECL-
solution (Pierce). The blot-membrane was then stripped by
antibodies (100 mM b-mercaptoethanol, 2% SDS, 62.5 mM
Tris–HCl pH 6.8, 60 8C), washed with TBST, re-probed
with anti-a5 integrin monoclonal antibody (hamster, Pharmingen,
1:500), amplified with horseradish peroxidase-conjugated
secondary antibody (1 h, 1:5000, Santa Cruz), and detected using
SuperSignal Femto ECL solution (Pierce).
2.8. Statistical analyses
Group data are presented as meanGstandard error of the mean,
and comparisons between groups used Student’s t-test or analysis
of variance (ANOVA) followed by Tukey’s multiple comparison
post hoc test, as appropriate. A probability of P!0.05 was
considered significant.
3. Results
3.1. Second messenger-induced hyperalgesia
To investigate the dependence of known hyperalgesic
second messenger cascades on the function of specific
integrin subunits, we elicited hyperalgesia using compounds
that act as agonists at various stages in the PKA, PKC3 or
MAPK/ERK signaling pathways. Hyperalgesia was induced
by intradermal injection of the catalytic subunit of PKA
(cPKA; 10 U/2.5 ml), pseudo3RACK (1 mg/2.5 ml), or
MEKC (10 U/2.5 ml) in the dorsal rat hind paw. In addition,
because production of cAMP by adenylate-cylase is
required in the PKA-dependent hyperalgesic pathway, we
also administered 1 mg/2.5 ml forskolin (an activator of(Boehringer), pH 7.4). The homogenate was processed in three
different ways. (1) A total membrane sample was collected by
centrifugation of the homogenate (55,000 rpm, TS-55 Beckman
rotor, 1 h, 4 8C), and the pellet resuspended in 4! Laemmli buffer,
and stored at K80 8C. (2) A detergent-resistant membrane sample
was prepared by solubilization in homogenization buffer plus 1%
Triton X-100 (1 h, on ice). (3) A raft-depleted membrane sample
was prepared by solubilization in homogenization buffer plus 1%
Triton X-100 and 60 mM b-n-octylglucoside (1 h, on ice).
To collect detergent-treated membranes, samples were adjusted
to 40% sucrose content (in 0.7 ml) and overlayed with 1.3 ml
homogenization buffer containing 30% sucrose and 0.3 ml 5%
sucrose buffer. Tubes were then centrifuged (18 h, 34,000 rpm,
4 8C). Unsolubilized membranes (which form vesicles enclosing
homogenization buffer that rise through the denser sucrose buffer)
were harvested at the 30/5% sucrose interface, diluted in 2.3 ml
5 (2005) 191–203 193adenylate cyclase), or the cAMP analog 8-Br-cAMP

(1 mg/2.5 ml). As shown in Fig. 1 (first bar in each graph),
these agents caused significant decreases in nociceptive
threshold, measured 15 min after injection. The magnitude
of hyperalgesia (30–40% decrease in nociceptive threshold)
was comparable to that induced by the inflammatory
mediators PGE2 and epinephrine, reported previously
(Aley and Levine, 1999; Aley et al., 2001a; Dina et al.,
2003; Khasar et al., 1999b).
Fig. 1. Function-blocking monoclonal antibodies against individual integrin subunits. Graphs display mean percent reduction in paw-withdrawal threshold
after injection of antibody (1 mg) followed 30 min later by forskolin, 8-Br-cAMP, cPKA, pseudo3RACK, or MEK1C(all 1 mg). (A) Anti-a1, -a3 and -b1
antibodies significantly blocked forskolin-induced hyperalgesia (nZ6 for each antibody; all P!0.001) while the anti-a5 antibody had no significant effect
C; nZ
antib
sted b
show
ent b
O.A. Dina et al. / Pain 115 (2005) 191–203194(nZ6). Hyperalgesia induced by either 8-Br-cAMP (B; nZ6) or by cPKA (
antibodies each blocked pseudo3RACK-induced hyperalgesia (nZ6; for each
anti-a1 or -a3 antibody (nZ6 for each antibody). (E) All the antibodies te
0.001). The pre-treatment baseline paw threshold for all antibody-treated rats
(D) 112.4G1.6 g and (E) 119.0G1.3 g (each nZ24). The mean pre-treatm
115.8G0.9 g. None of the antibodies had a significant effect on paw-withdrawal6) was unaffected by any of the antibodies tested. (D) The anti-a5 and -b1
ody; P!0.001). However, pseudo3RACK hyperalgesia was not reduced by
locked hyperalgesia induced by MEK1C(nZ6 for each antibody; all P!
n in each graph was: (A) 116.7G1.2 g, (B) 115.6G1.3 g, (C) 115.1G1.0 g,
aseline paw-withdrawal threshold for all untreated rats in this figure was
threshold when injected alone (nZ30 for each monoclonal antibody).

3.2. Anti-integrin monoclonal antibodies
Function-blocking anti-integrin monoclonal antibodies
have been used to probe integrin function in many systems
(Humphries, 2000). We recently reported the efficacy of
intradermally injected anti-a1, -a3, -a5 and -b1 antibodies
for evaluating the role of integrin signaling in inflammatory
mediator-induced hyperalgesia (Dina et al., 2004). To
investigate potential interactions of integrins with second
messenger signaling pathways that mediate peripheral
hyperalgesia, we tested whether intradermal injection
of antibodies against a1, a3, a5 or b1 integrin subunits
(each 1 mg/10 ml) could attenuate mechanical hyperalgesia
induced by forskolin, 8-Br-cAMP, cPKA, pseudo3RACK or
MEK1C.
We found that hyperalgesia induced by forskolin, an
activator of adenylate cyclase, was abolished by antibodies
directed against the a1, a3 or b1 integrin subunits, but not by
antibody against a5 (Fig. 1A; P!0.05). In contrast,
hyperalgesia induced by 8-Br-cAMP or cPKA was not
significantly affected by monoclonal antibodies against a1,
a3, a5 or b1 subunits (Fig. 1B and C). Hyperalgesia induced
by pseudo3RACK was significantly inhibited only by
monoclonal antibodies against a5 and b1 subunits (Fig. 1D),
mism
confir
41.0G
abolis
peralg
ralges
mism
O.A. Dina et al. / Pain 115 (2005) 191–203 195Fig. 2. Selective knockdown of a1 integrin subunit expression. Antisense or
hind paw daily for 3 days. (A) Knockdown of a1 integrin expression was
47.9G5.8; arbitrary units normalized to the reference protein; decrease of
t-test). (B–F) Hyperalgesia induced by forskolin (B) or MEK1C (F) was
compared to mismatch). Following recovery from antisense treatment, hy
from that in mismatch-treated controls. Anti-a1 ODN did not inhibit hype
each agonist). When administered alone, anti-a1 ODN or its corresponding
each ODN).atch ODN against the a1 integrin subunit were injected intradermally in the
med by Western blot of saphenous nerve (mismatch, 81.2G8.1; antisense,
7.9% compared to mismatch-treated controls; nZ4, P!0.009, one-tailed
hed in rats treated with antisense ODN (nZ8 for each agonist; P!0.001
esia induced by forskolin or by MEK1C was not significantly different
ia induced by 8-Br-cAMP (C), cPKA (D) or pseudo3RACK (E) (nZ8 for
atch ODN had no significant effect on paw-withdrawal threshold (nZ8 for