Beha io al/Sys ems/Cogni i e
Widesp ead Exp ession o he AMPA Recep o GluR2
Subuni a Glu ama e gic Synapses in he Ra Spinal Co d
and Phospho yla ion o GluR1 in Response o Noxious
S imula ion Re ealed wi h an An igen-Unmasking Me hod
Ge gely G. Nagy,
1
Muna Al-Ayyan,
1
Da id And ew,
1
Masahi o Fukaya,
2
Masahiko Wa anabe,
2
and And ew J. Todd
1
1
Spinal Co d G oup, Ins i u e o Biomedical and Li e Sciences, Uni e si y o Glasgow, Glasgow G12 8QQ, Uni ed Kingdom, and
2
Depa men o Ana omy,
Hokkaido Uni e si y School o Medicine, Sappo o 060-8638, Japan
Glu ama e, hep incipalexci a o yneu o ansmi e in hespinalco d,ac sp ima ily h oughAMPA ecep o s.Al houghall ou AMPA
subuni s a e exp essed by spinal neu ons, we know li le abou hei dis ibu ion a glu ama e gic synapses. We used an an igen-
unmasking echnique o e eal he synap ic dis ibu ion o glu ama e ecep o (GluR) 1–4 subuni s wi h con ocal mic oscopy. A e
pepsin ea men , punc a e s aining was seen wi h an ibodies agains each subuni : GluR2-immuno eac i e punc a we e dis ibu ed
h oughou he g ay ma e , whe eas GluR1-immuno eac i e punc a we e es ic ed o he do sal ho n and we e mos nume ous in
laminas I–II. Punc a e s aining o GluR3 and GluR4 was ound in all laminas bu was weak in supe icial do sal ho n. Colocaliza ion
s udies showed ha GluR2 was p esen a i ually all (98%) punc a ha we e GluR1, GluR3, o GluR4 immuno eac i e and ha mos
(⬎90%) immuno eac i e punc a in laminas IV, V, and IX showed GluR2, GluR3, and GluR4 immuno eac i i y.
E idence ha hesepunc a ep esen ed synap ic ecep o s was ob ainedwi helec onmic oscopyand by examining he associa iono
GluR2- and GluR1-immuno eac i e punc a wi h glu ama e gic bou ons (iden i ied wi h esicula glu ama e anspo e s o ma ke s o
unmyelina ed a e en s). The g ea majo i y (96%) o hese bou ons we e associa ed wi h GluR2-immuno eac i e punc a. Ou indings
sugges ha GluR2 is almos uni e sally p esen a AMPA-con aining synapses, whe eas GluR1 is p e e en ially associa ed wi h p ima y
a e en e minals.
We also ound a subs an ial, apid inc ease in s aining o synap ic GluR1 subuni s phospho yla ed on he S845 esidue in he
ipsila e al do sal ho n a e pe iphe al noxious s imula ion. This inding demons a es plas ic changes, p esumably con ibu ing o
cen al sensi iza ion, a he synap ic le el.
Key wo ds: immunocy ochemis y; VGLUT1; VGLUT2; CGRP; IB4; capsaicin
In oduc ion
Glu ama e is he main exci a o y neu o ansmi e in he spinal
co d and is used by p ima y a e en s, exci a o y in e neu ons,
p ojec ion neu ons, and some descending axons (B oman, 1994).
Glu ama e ac s on iono opic glu ama e ecep o s (iGluRs) and
me abo opic glu ama e ecep o s, which a e widely exp essed by
spinal neu ons (Coggeshall and Ca l on, 1997). iGluRs can be
di ided in o h ee classes on he basis o hei p e e ence o
AMPA, kaina e, o NMDA. AMPA ecep o s media e as EPSPs
in spinal neu ons (Yoshimu a and Jessell, 1990; Yoshimu a and
Nishi, 1992) and a e esponsible o ansmission o nocicep i e
signals pe cei ed as acu e pain (Dickenson e al., 1997). AMPA
ecep o s ha e also been implica ed in exci o oxici y and degen-
e a ion o mo oneu ons in amyo ophic la e al scle osis ( an
Damme e al., 2002).
mRNAs o all ou subuni s o he AMPA ecep o
(GluR1–4) ha e been iden i ied in he spinal co d (Fu uyama e
al., 1993; Henley e al., 1993; To¨lle e al., 1993, 1995; Jakowec e
al., 1995b; Shiba a e al., 1999). Al hough con en ional immuno-
cy ochemical s udies ha e been pe o med o de ec AMPA sub-
uni s (Tachibana e al., 1994; Jakowec e al., 1995a; Pop a ilo e
al., 1996, 1998; Mo ison e al., 1998; Spike e al., 1998), i is
unlikely ha hese e ealed ecep o s a glu ama e gic synapses,
because hese a e hough o be inaccessible due o ex ensi e
c oss-linking o he p o ein meshwo k o he synap ic cle and
pos synap icdensi y ha esul s omaldehyde ixa ion(Baudee al.,
1995; O e sen and Landsend, 1997; F i schy e al., 1998; Wa anabe
e al., 1998). Pos embedding immunogold-labeling has been used o
in es iga e synap ic AMPA ecep o s in he spinal co d (Pop a ilo
e al., 1996, 1998; Mo ison e al., 1998; Ragna son e al., 2003);
howe e , despi e his, he pa e n o AMPA subuni exp ession
a indi idual synapses emains la gely elusi e.
Recei ed Dec. 24, 2003; e ised May 4, 2004; accep ed May 10, 2004.
Thiswo kwassuppo edby heBio echnologyandBiologicalSciencesResea chCouncil,WellcomeT us ,B i ish
Council,andFacul yo MedicalandHeal hSciencesa heUni edA abEmi a esUni e si y.We hankR.Ke ,C.Wa ,
and M. McGill o expe echnical assis ance and P o . D. J. Maxwell o help ul discussion and ad ice.
Co espondence should be add essed o D . A. J. Todd, Spinal Co d G oup, Wes Medical Building, Uni e si y o
Glasgow, Uni e si y A enue, Glasgow G12 8QQ, UK. E-mail: [email p o ec ed].
DOI:10.1523/JNEUROSCI.1237-04.2004
Copy igh © 2004 Socie y o Neu oscience 0270-6474/04/245766-12$15.00/0
5766 •The Jou nal o Neu oscience, June 23, 2004 •24(25):5766–5777
The e a e majo di e ences be ween AMPA subuni s in hei
physiology, pha macology, and in ol emen in synap ic plas ic-
i y (Dingledine e al., 1999). Fo example, he subuni s di e in
hei esponse o phospho yla ion (B ed and Nicoll, 2003), a
p ocess ha is hough o con ibu e o cen al sensi iza ion o
do sal ho n neu ons in ch onic pain s a es (Sandku¨hle , 2000;
Fang e al., 2003; Ji e al., 2003). In addi ion, absence o he GluR2
subuni esul s in Ca
2⫹
pe meabili y, which has been implica ed
in ce ain pa hological condi ions (So kin e al., 1999, 2001; Van-
denbe ghe e al., 2001). Knowledge o he subuni composi ion a
synapses in speci ic neu onal ci cui s is he e o e o undamen al
impo ance o ou unde s anding o synap ic unc ion and
plas ici y.
In his s udy, we examined synap ic AMPA subuni s wi h con-
ocal and elec on mic oscopy by adop ing a p o eoly ic an igen-
unmasking me hod ha allows an ibodies o gain access o syn-
ap ic ecep o s (Wa anabe e al., 1998). We p o ide he i s
demons a ion o he lamina dis ibu ion o all ou AMPA sub-
uni s a synapses in he spinal co d, and o GluR1 and GluR2
subuni s, we in es iga ed hei ela ionship o di e en ypes o
glu ama e gic axon. We also show ha his me hod can be used in
unc ional s udies, because we we e able o demons a e phos-
pho yla ion o GluR1 a synapses in he supe icial do sal ho n in
esponse o noxious s imula ion.
Ma e ials and Me hods
GluR an ibodies. A abbi an ibody agains he C- e minal amino acid
esidues 830–862 o mouse GluR3 (GenBank accession numbe
AB022342) and a guinea pig an ibody agains esidues 245–273 o mouse
GluR4 (GenBank accession numbe AB022913) we e gene a ed as de-
sc ibed p e iously (Fukaya and Wa anabe, 2000). Using he pGEX4T-2
plasmid ec o (Ame sham Biosciences, Bucks, UK), p o eins we e ex-
p essed as a glu a hione S- ans e ase (GST) usion p o ein and pu i ied
using glu a hione-Sepha ose 4B (Ame sham Biosciences). A e h om-
bin diges ion, an igen polypep ides (R3C and R4N) we e sepa a ed om
GST by e e se-phase HPLC. Pu i ied polypep ides we e injec ed in o
emale abbi s and guinea pigs a in e als o 2 weeks. An ibodies o
GluR3 (GluR3C) and GluR4 (GluR4N) we e a ini y-pu i ied using GST
usion p o ein-coupled cyanogen b omide-ac i a ed Sepha ose 4B (Am-
e sham Biosciences).
To es he speci ici y o hese an ibodies, spinal co ds o adul
C57BL/6 mice we e homogenized in 8–10 olumes o ice-cold bu e
con aining 0.32 Msuc ose, 1 mMEDTA, 1 mMEGTA, 10 mMT is-HCl,
pH 7.2, and 0.4 mMphenylme hylsul onyl luo ide (homogeniza ion
bu e ) using a Po e homogenize wi h 15 s okes a 800 pm. To ob ain
a pos synap ic densi y (PSD) ac ion, he homogena e was cen i uged
a 1000 ⫻g o 10 min o emo e nuclei and la ge deb is. The supe na an
was cen i uged a 10,000 ⫻g o 20 min o ob ain a c ude synap osomal
ac ion and subsequen ly lysed hypo-osmo ically and cen i uged a
25,000 ⫻g o 30 min o pelle a synap osomal ac ion. The pelle was
suspended wi h homogeniza ion bu e con aining 0.5% T i on X-100
o 15 min and cen i uged a 111,000 ⫻g o 1 h o pelle a PSD ac ion.
The p o ein concen a ion was de e mined by Low y’s me hod. A e
SDS-PAGE, ac iona ed samples in he gel we e elec oblo ed on o ni-
ocellulose memb anes (BioT aceNT; Pall Gelman Labo a o y, Ann A -
bo , MI). Memb anes we e incuba ed wi h 5% skimmed milk in T is-
bu e ed saline (TBS) con aining 0.1% Tween 20 (TBST), pH 7.5, o 1
h , ollowed by incuba ion wi h p ima y an ibodies (1
␮
g/ml) in TBST
o 2 h . Immuno eac ion was isualized wi h he ECL chemilumines-
cence de ec ion sys em (Ame sham Biosciences). Fo p eabso p ion ex-
pe imen s, pep ides (0.2
␮
g/ml) we e added o he p ima y an ibodies.
To check he c oss-immuno eac i i y o GluR3C an ibody o he GluR2
subuni , a do blo assay was pe o med using C- e minal pep ides o
GluR2 and GluR3 subuni s (amino acid esidues 826–858 o GluR2,
R2C; 830–862 o GluR3, R3C). A e he pep ides we e do ed on o
ni ocellulose memb anes, immuno eac ion was pe o med as abo e.
The a ini y-pu i ied abbi an ibodies agains GluR1 and GluR2 used
in his s udy (Chemicon In e na ional, Ha low, UK) show no c oss-
eac i i y wi h o he GluR subuni s (speci ica ion o manu ac u e ). The
mouse monoclonal an i-GluR2 (Chemicon; clone 6C4) has been ex en-
si ely cha ac e ized and shown no o de ec any o he AMPA o kaina e
subuni s (Vissa ajjhala e al., 1996).
Animals o immunocy ochemis y. Expe imen s we e app o ed by he
Uni e si y o Glasgow E hical Re iew P ocess Applica ions Panel and
pe o med in acco dance wi h he UK Animals (Scien i ic P ocedu es)
Ac 1986. Twel e adul male Wis a a s (220–390 gm; Ha lan, Lough-
bo ough, UK) we e deeply anes he ized wi h pen oba bi one (300 mg,
i.p.) and pe used ansca dially wi h ixa i e con aining 4% eshly de-
polyme ized o maldehyde (10 a s) o 4% o maldehyde/0.1% glu a al-
dehyde ( wo a s). Th ee addi ional a s (280–330 gm) we e used o
in es iga e phospho yla ion o he GluR1 subuni in esponse o noxious
s imula ion. These we e anes he ized wi h ke amine and xylazine (73.3
and 7.3 mg/kg, i.p., espec i ely) and ecei ed an injec ion o 250
␮
go
capsaicin (Sigma, Poole, UK; dissol ed in 25
␮
l o 7% Tween 80 in saline)
in o he plan a su ace o he le hindpaw. They we e main ained unde
gene al anes hesia and pe used wi h 4% o maldehyde unde e minal
pen oba bi one anes hesia 10 min a e he capsaicin injec ion.
Lumba spinal co d segmen s om all animals we e emo ed, s o ed
in ixa i e o 8–24 h , and cu in o ans e se 60
␮
m Vib a ome sec ions,
which we e imme sed in 50% e hanol o 30 min o enhance an ibody
pene a ion.
Immuno luo escen de ec ion o GluR1–4. To expose AMPA ecep o
subuni s a synap ic si es, sec ions we e p ocessed acco ding o an
an igen-unmasking me hod in ol ing pepsin diges ion (Wa anabe e al.,
1998). In some expe imen s, glu ama e gic axons we e iden i ied wi h
an ibodies agains he esicula glu ama e anspo e s VGLUT1 and
VGLUT2, which a e p ima ily associa ed wi h myelina ed p ima y a e -
en s and exci a o y in e neu ons, espec i ely (Todd e al., 2003). Unmy-
elina ed (C) p ima y a e en s gene ally do no ha e de ec able le els o
hese anspo e s (Todd e al., 2003), and we he e o e iden i ied pep i-
de gic and nonpep ide gic C a e en s wi h an an ibody agains calci o-
nin gene- ela ed pep ide (CGRP) and binding o Bandei aea simplici olia
isolec in B4 (IB4), espec i ely. A e pepsin ea men , s aining o hese
axonal ma ke s disappea ed nea he su aces o he Vib a ome sec ion,
whe e punc a e s aining o he GluRs was op imal. De ec ion o hese
compounds was he e o e pe o med be o e pepsin ea men wi h he
y amide signal ampli ica ion (TSA) me hod, which esul ed in deposi-
ion o co alen ly bound luo opho e ha was esis an o p o eolysis.
Pepsin ea men was pe o med by incuba ing sec ions a 37°C o 30
min in PBS ollowed by 10 min in 0.2 MHCl con aining 1 mg/ml pepsin
(Dako, Glos up, Denma k). Fo sec ions p ocessed o con ocal mic os-
copy, he PBS con ained 0.3% T i on X100.
To in es iga e he ela ionship be ween GluR1, GluR2, and a ious
ypes o glu ama e gic bou on, sec ions we e incuba ed o 72 h a 4°Cin
one o he ollowing: (1) guinea pig an i-VGLUT1 (Chemicon; 1:200,000
o 1:500,000), (2) guinea pig an i-VGLUT2 (Chemicon; 1:50,000), (3) a
mix u e o bo h guinea pig VGLUT an ibodies (a he co esponding
concen a ions), (4) guinea pig an i-CGRP (Bachem, Me seyside, UK;
1:100,000), o (5) bio inyla ed IB4 (Sigma; 2
␮
g/ml). They we e hen
incuba ed o 2 h a oom empe a u e in bio inyla ed donkey an i-
guinea pig IgG (Jackson ImmunoResea ch, Wes G o e, PA; 1:500)
(excep o hose ea ed wi h IB4) and p ocessed wi h a TSA ki ( e a-
me hyl hodamine; Pe kinElme Li e Sciences, Bos on, MA) acco ding o
he ins uc ions o he manu ac u e . Sec ions we e insed, pepsin
ea ed, and incuba ed o 24 h a 4°C in a mix u e o abbi an i-GluR1
(1:500) and mouse an i-GluR2 (3.9
␮
g/ml), ollowed by 2–24 h in spe-
cies-speci ic donkey seconda y an ibodies: an i-mouse IgG conjuga ed o
Alexa 488 (Molecula P obes, Eugene, OR; 1:500) and an i- abbi IgG
conjuga ed o Cyanine 5.18 (Cy5; Jackson ImmunoResea ch; 1:100).
They we e moun ed wi h an i- ade medium and s o ed a ⫺20°C. Sec-
ions om h ee a s we e p ocessed wi h each o hese an ibody
combina ions.
We pe o med iple-immuno luo escence labeling o examine he
ela ionship be ween GluR2, GluR3, and GluR4. Sec ions om h ee a s
we e pepsin ea ed and incuba ed o 72 h a 4°C in a mix u e o mouse
Nagy e al. •Synap ic AMPA Recep o s in Spinal Co d J. Neu osci., June 23, 2004 •24(25):5766–5777 • 5767
an i-GluR2 (3.9
␮
g/ml), abbi an i-GluR3 (GluR3C) (0.41
␮
g/ml), and
guinea pig an i-GluR4 (GluR4N) (0.76
␮
g/ml), ollowed by species-
speci ic donkey seconda y an ibodies (an i-mouse and guinea pig IgG
conjuga ed o Cy5 o Rhodamine Red, 1:100, Jackson ImmunoResea ch;
an i- abbi IgG conjuga ed o Alexa 488, 1:500, Molecula P obes).
P eabso p ion con ols o he GluR3 and GluR4 an ibodies we e pe -
o med by adding R4N pep ide o he GluR4 an ibody and R2C o R3C
pep ide o he GluR3 an ibody (10
␮
g/ml pep ide added o 50
␮
g/ml
an ibody in each case) and hen dilu ing he an ibody o he concen a-
ions abo e. To compa e s aining ob ained wi h abbi and mouse GluR2
an ibodies, some sec ions we e ea ed wi h pepsin and incuba ed in a
mix u e o hese wo p ima y an ibodies, ollowed by app op ia e luo-
escen seconda y an ibodies.
To assess he e ec o pepsin ea men , nonpepsin- ea ed sec ions
we e incuba ed in abbi an i-GluR1 (1:500), mouse an i-GluR2 (3.9
␮
g/ml), abbi an i-GluR3 (0.41
␮
g/ml), o guinea pig an i-GluR4 (0.76
␮
g/ml), ollowed by luo escen seconda y an ibodies.
Con ocal mic oscopy and analysis o GluR1–4. Sec ions eac ed wi h
an ibodies agains GluR2, GluR3, and GluR4 we e analyzed o de e mine
he ex en o colocaliza ion o hese subuni s in immuno eac i e punc a.
One sec ion om each o h ee animals was scanned h ough a 60⫻
oil-imme sion lens wi h a Bio-Rad (Hemel Hemps ead, UK) Radiance
2100 con ocal lase -scanning mic oscope. Z-se ies we e scanned a 0.3
␮
m z-sepa a ion h ough he op 10–15
␮
m o he sec ion, and i e
egions we e analyzed in each case: laminas I/II, III, IV, V, and IX. Con-
ocal image s acks we e analyzed wi h Me aMo ph so wa e (Uni e sal
Imaging, Downing on, PA). The h ee-colo channels we e me ged and
combined in o a monoch ome image, so ha he subuni combina ion in
each punc um could no be de e mined. One hund ed immuno eac i e
punc a we e selec ed om each egion in each animal. To a oid bias, a
100-squa e g id was placed on he image, and he punc um closes o
bo om igh co ne o each g id squa e was selec ed. Selec ed punc a
we e hen examined in he h ee-colo image s ack o de e mine which
ypes o subuni immuno eac i i y hey exp essed.
F om each o h ee a s, a single sec ion eac ed wi h an i-GluR1 and
an i-GluR2, oge he wi h an i-VGLUT1, an i-VGLUT2, an i-CGRP, o
bio inyla ed IB4, was analyzed. These sec ions we e scanned h ough a
60⫻oil-imme sion lens as desc ibed abo e. VGLUT1- and VGLUT2-
labeled bou ons we e analyzed in laminas I, II, III, IV, V, and IX. Mos
VGLUT1 e minals in he inne hal o lamina II (IIi) belong o myelin-
a ed p ima y a e en s, whe eas hose in ou e lamina II (IIo) a e no o
p ima y a e en o igin (Todd e al., 2003). We he e o e analyzed lami-
nas IIi and IIo sepa a ely o bo h VGLUTs. CGRP- and IB4-labeled
axons we e analyzed in he egions in which hey we e mos nume ous:
laminas I/IIo (CGRP) and lamina II (IB4). Con ocal image s acks we e
iewed wi h Me aMo ph such ha only he axonal ma ke was isible,
and 100 labeled axons we e andomly selec ed om each egion in each
animal. The o he wo con ocal channels we e hen iewed, and he
numbe o GluR1- and/o GluR2-immuno eac i e punc a in con ac
wi h each bou on was de e mined.
VGLUT1 and VGLUT2 a e hough o label he g ea majo i y o glu-
ama e gic e minals in he spinal co d, apa om hose belonging o C
a e en s. To in es iga e he p opo ion o GluR2 punc a ha was in
con ac wi h glu ama e gic e minals, and hus likely o ep esen synap-
ic ecep o s, we he e o e analyzed sec ions eac ed wi h an ibodies
agains bo h anspo e s and GluR2. One sec ion om each o h ee
animals was scanned h ough a 60⫻oil-imme sion lens, and h ee e-
gions we e analyzed in each case: laminas I/II, III/IV, and IX. Image
s acks we e ini ially iewed such ha only GluR2 s aining was isible, and
100 immuno eac i e punc a we e selec ed om each egion in each an-
imal using a g id o a oid bias. The VGLUT s aining was hen examined,
and he p opo ion o selec ed punc a in con ac wi h a VGLUT-
immuno eac i e bou on was de e mined.
Elec on mic oscopy. Al hough AMPA subuni s ha e been ound in
p ima y a e en e minals in he spinal co d (Lu e al., 2002), we did no
see GluR immunos aining ha appea ed o be loca ed inside p ima y
a e en bou ons a e pepsin ea men (see Resul s). Howe e , we could
no ule ou he possibili y ha some o he punc a e labeling apposed o
p ima y a e en e minals co esponded o ecep o s in he p esynap ic
memb ane. We he e o e pe o med elec on mic oscopy o con i m ha
he punc a e labeling seen wi h GluR1 and GluR2 an ibodies ep esen ed
pos synap ic ecep o s. Sec ions om he wo a s ixed wi h o maldehyde/
glu a aldehyde we e ea ed wi h 1% sodium bo ohyd ide o 30 min ol-
lowed by pepsin and incuba ed in abbi an i-GluR1 (1:2,000) o abbi
an i-GluR2 (1:500) o 72 h and in bio inyla ed donkey an i- abbi IgG
(Jackson ImmunoResea ch; 1:500) ollowed by Ex a idin-pe oxidase con-
juga e (Sigma; 1:1000), each o 24 h . Pe oxidase was e ealed wi h 3,3⬘-
diaminobenzidine, and he sec ions we e osmica ed, dehyd a ed in ace one,
and la -embedded in Du cupan. Ul a hin sec ions we e cu on o Fo m a -
coa ed single-slo g ids, s ained wi h lead ci a e, and iewed wi h a Philips
CM100 elec on mic oscope. Fo compa ison, nonpepsin- ea ed Vi-
b a ome sec ions we e p ocessed in pa allel.
Fo he pepsin- ea ed ma e ial, one ul a hin sec ion om each ani-
mal eac ed wi h GluR1 and one eac ed wi h GluR2 we e examined. The
egion co esponding o laminas I–III was sys ema ically scanned un il
50 immuno eac i e synapses had been iden i ied on each sec ion, and he
ela ionship o eac ion p oduc o synap ic memb anes was no ed in
each case. We also no ed any immuno eac i e esicle-con aining p o-
iles. Ul a hin sec ions om issue eac ed wi h GluR1 o GluR2 wi hou
pepsin ea men we e hen iewed (one om each a o each an i-
body), and an a ea equi alen o ha scanned in he pepsin- ea ed sec-
ions was examined.
De ec ion o phospho yla ed GluR1 subuni s. To de e mine whe he
immunocy ochemically de ec able changes in ol ing synap ic AMPA e-
cep o s occu ed a e noxious s imula ion, immunolabeling was pe -
o med o e eal GluR1 subuni s phospho yla ed a he Se ine 845 si e
(GluR1-pS845) (Roche e al., 1996) on sec ions om h ee a s ha e-
cei ed injec ions o capsaicin in o he le hindpaw 10 min be o e pe u-
sion. Pepsin- ea ed L4 sec ions we e incuba ed in a mix u e o abbi
an i-GluR1-pS845 (Co ance, Be keley, CA; 0.5
␮
g/ml) and mouse an i-
GluR2 (3.9
␮
g/ml) o 48 h , ollowed by donkey seconda y an ibodies
(an i-mouse-Alexa 488, 1:500, Molecula P obes; bio inyla ed an i-
abbi , 1:500, Jackson ImmunoResea ch) o 2 h . GluR1-pS845 was e-
ealed wi h he TSA ki . The an ibody agains GluR2 was used o dem-
ons a e ha punc a e s aining seen wi h he GluR1-pS845 an ibody
co esponded o AMPA-con aining synapses, because GluR2 was ound
in i ually all punc a ha con ained GluR1 (see Resul s).
One sec ion om each a was scanned h ough a 60⫻oil-imme sion
lens wi h he con ocal mic oscope. F om hese sec ions, h ee adjacen
o e lapping ields (each 155 ⫻155
␮
m) co e ing he medial hal o
laminas I and II o each do sal ho n we e scanned sequen ially a 1
␮
m
z-sepa a ion o e eal GluR1-pS845 and GluR2. The medial hal o he
supe icial do sal ho n was analyzed because nocicep i e p ima y a e -
en s om he hindpaw e mina e in his a ea. Fo each do sal ho n, he
h ee image s acks we e s i ched oge he wi h Adobe Pho oshop 7.0
(Adobe Sys ems, San Jose, CA) o p oduce a single op ical sec ion ha
co e ed he medial hal o he do sal ho n, and he a ea co esponding o
laminas I and II was ou lined o subsequen analysis. The ed channel
(co esponding o GluR1-pS845) was swi ched o while his was pe -
o med o a oid bias in he selec ion o he op ical sec ion o he a ea ha
was o be analyzed. GluR1-pS845 immuno eac i i y was analyzed in each
o he op ical sec ions by se ing a h eshold o pixel luminance alue
ha excluded mos o he “basal”immunos aining seen in he con ala -
e al do sal ho n. Punc a in each do sal ho n ha had a leas one pixel
exceeding his h eshold alue we e iden i ied and coun ed, and he p es-
ence o absence o GluR2 immunos aining in hese punc a was
de e mined.
The GluR1-pS845 an ibody was aised agains a syn he ic phos-
phopep ide co esponding o amino acids su ounding he Se 845 esi-
due o he a GluR1 subuni and pu i ied by sequen ial ch oma og aphy
on p o ein A and phospho- and dephospho-pep ide a ini y columns. On
Wes e n blo s o a hippocampal homogena e, he an ibody labels a
single band co esponding o a molecula weigh o ⬃100 kDa (speci i-
ca ion o he manu ac u e ). P eabso b ion con ols o immunocy o-
chemis y we e pe o med by adding he phosphopep ide used o aise
he an ibody o he co esponding dephosphopep ide (0.1
␮
g/ml in each
case) o he an ibody 24 h be o e use.
5768 •J. Neu osci., June 23, 2004 •24(25):5766–5777 Nagy e al. •Synap ic AMPA Recep o s in Spinal Co d
Resul s
GluR3 and GluR4 an ibodies
The sequence used o gene a e he GluR3 an ibody has a high
le el o homology wi h he co esponding pa o he GluR2 sub-
uni , and we he e o e in es iga ed whe he his an ibody c oss-
eac ed wi h GluR2. On do blo s, he GluR3 an ibody de ec ed
he R3C bu no he R2C pep ide, indica ing ha i did no c oss-
eac wi h he C- e minal po ion o he GluR2 subuni . In Wes -
e n blo s o he PSD ac ion om mouse spinal co d, he GluR3
and GluR4 an ibodies each de ec ed a single p o ein band co e-
sponding o a molecula weigh o ⬃98 kDa. S aining wi h he
GluR3 an ibody was blocked by p eabso p ion wi h R3C bu no
R2C pep ide, and he GluR4 s aining was blocked wi h he R4N
pep ide. On sec ions o pepsin- ea ed spinal co d, punc a e s aining
wi h GluR3 an ibody was blocked by p eabso p ion wi h R3C bu
no R2C pep ide, whe eas ha wi h GluR4 an ibody was blocked
wi h R4N pep ide. Cha ac e iza ion da a o hese an ibodies is
shown in supplemen al Figu e 1 (a ailable a www.jneu osci.o g).
Gene al appea ance o immuno luo escence wi h GluR1–4
an ibodies a e an igen unmasking
On pepsin- ea ed spinal co d sec ions, punc a e s aining was
seen nea he su ace o he sec ion (supe icial 10–15
␮
m) wi h
each GluR an ibody (Fig. 1). This became weake deepe in he
sec ion, and o he GluR2 an ibodies was eplaced by cell body
labeling iden ical o ha epo ed wi hou pepsin ea men
(Pop a ilo e al., 1996, 1998). The loss o cy oplasmic labeling
sugges s ha ex asynap ic ecep o s a e damaged du ing p o eo-
ly ic ea men , whe eas synap ic ecep o s a e p esumably p o-
ec ed by he su ounding p o ein meshwo k. The abbi and
mouse GluR2 an ibodies s ained iden ical s uc u es. Because
punc a e s aining was seen wi h an ibodies di ec ed agains bo h
C- e minal (in acellula ) (GluR1, GluR3, polyclonal GluR2)
and N- e minal (ex acellula ) (GluR4, monoclonal GluR2)
epi opes, p o eoly ic diges ion p esumably allows access o bo h
pos synap ic densi y and synap ic cle . Lu e al. (2002) ound ha
a e weak ixa ion, GluR2/3, GluR2/4, and GluR4 an ibodies la-
beled cen al e minals o p ima y a e en s in he do sal ho n. In
pepsin- ea ed issue iewed wi h con ocal mic oscopy, we did
no see a simila s aining pa e n wi h any o he GluR an ibodies
used in his s udy, sugges ing ha epi opes on p esynap ic ecep-
o s a e ei he damaged o emain inaccessible a e pepsin
ea men .
Lamina dis ibu ion o AMPA subuni s
P e ious immunohis ochemical and in si u hyb idiza ion s udies
ha e sugges ed ha GluR1 and GluR2 subuni s a e p esen
h oughou he do sal ho n, wi h highes le els in laminas I and
II, ha GluR2 is also p esen in he en al ho n, and ha GluR3
and GluR4 a e highly exp essed in en al ho n, wi h mode a e
le els in deep do sal ho n and limi ed exp ession in he supe i-
cial laminas (Fu uyama e al., 1993; To¨lle e al., 1993; Tachibana
e al., 1994; Jakowec e al., 1995a,b; Ha is e al., 1996; Pop a ilo
e al., 1996, 1998; Mo ison e al., 1998; Spike e al., 1998; En-
gelman e al., 1999; Shiba a e al., 1999). Ou indings a e pepsin
ea men a e consis en wi h hese epo s (Fig. 1). GluR1
Figu e1. Immunos aining o GluR1–4in hespinalco da e pepsin ea men .a–dshow hemedialpa o hesupe icialdo salho n(co esponding olaminasI–III).e–hinclude hemedial
pa o laminasIVand V.i–la e omlaminaIX.GluR1-immuno eac i epunc aa enume ousinlaminaII,p esen a lowe densi yino he pa so hedo salho n, anda e i uallyabsen inlamina
IX.GluR2-immuno eac i epunc a a ep esen inallpa s o hedo salho n, bu helabelingiss onges inlaminaII.Punc a elabeling wi hbo hGluR3and GluR4 an ibodiesisseen h oughou he
g ayma e bu isleas densein hesupe icialdo salho n.Imageso GluR1andGluR2a e aken omonesec ion,and hose o GluR3and GluR4 a e aken om ano he . Each image was ob ained
om a p ojec ion o h ee op ical sec ions a 0.5
␮
m z-sepa a ion. Scale ba , 100
␮
m.
Nagy e al. •Synap ic AMPA Recep o s in Spinal Co d J. Neu osci., June 23, 2004 •24(25):5766–5777 • 5769
showed he mos es ic ed dis ibu ion,
wi h immuno eac i e punc a being e-
quen in lamina II, p esen a lowe densi y
in o he do sal ho n laminas bu a ely
seen in he en al ho n. Nume ous
GluR2-immuno eac i e punc a we e
p esen h oughou he g ay ma e , wi h
he s onges labeling in laminas I–II.
GluR3- and GluR4-immuno eac i e
punc a we e densely dis ibu ed h ough-
ou he en al ho n and laminas III-VI,
wi h much weake labeling in laminas I
and II, al hough a ew s ongly labeled
punc a we e seen in his egion, pa icu-
la ly in lamina I. I has been epo ed ha
mo oneu ons exp ess GluR1 (Pelleg ini-
Giampie o e al., 1994; Vi go e al., 1996;
Temkin e al., 1997; Williams e al., 1997;
Ba -Peled e al., 1999; Shiba a e al., 1999)
o lack GluR2 subuni s (Williams e al.,
1997; Ba -Peled e al., 1999; Del Can˜oe
al., 1999; Shaw e al., 1999). Ou esul s do
no suppo hese sugges ions, because we
ound ha in lamina IX, he e was i ually
no punc a e s aining o GluR1, whe eas
GluR2 was exp essed a i ually all punc a
ha con ained any o he AMPA subuni s
(see below).
In nonpepsin- ea ed sec ions, labeling
o cell bodies and small p o iles ha e-
sembled dend i es was seen wi h GluR1
and GluR2 an ibodies. Nume ous GluR2-
immuno eac i e cell bodies we e obse ed
h oughou laminas I–III, whe eas sca -
e ed GluR1-labeled cell bodies we e
p esen p ima ily in laminas I–II. Howe e ,
he immuno luo escen labeling wi h hese
an ibodies in sec ions ha had no unde -
gone pepsin ea men was much weake
han ha in pepsin- ea ed sec ions and
punc a o he ype seen a e pepsin ea -
men we e no isible (supplemen al Fig. 2,
a ailable a www.jneu osci.o g). The GluR3
and GluR4 an ibodies did no s ain cell bod-
ies o gi e punc a e labeling in he absence o
pepsin ea men . This sugges s ha con en-
ional immunocy ochemis y wi hou an i-
gen unmasking does no gi e signi ican la-
beling o synap ic AMPA ecep o s.
Colocaliza ion o AMPA subuni s
A speci ic ques ion add essed in his pa o he s udy was whe he
any AMPA-con aining glu ama e gic synapses lacked GluR2
subuni s and would he e o e ha e exclusi ely Ca
2⫹
-pe meable
AMPA ecep o s. We ound ha i ually all GluR1-
immuno eac i e punc a h oughou he do sal ho n we e also
GluR2 immuno eac i e (Fig. 2a–c), al hough he ela i e in en-
si y o labeling o he wo subuni s a ied conside ably be ween
punc a. The ex en o colocaliza ion o hese wo subuni s was
in es iga ed in he analysis o di e en ypes o glu ama e gic
bou ons (see below). Colocaliza ion o GluR2, GluR3, and GluR4
was examined in iple-labeled sec ions (Fig. 2d–o; Table 1).
The e was ex ensi e colocaliza ion o hese subuni s in all pa s o
he g ay ma e en al o lamina III, wi h mos immuno eac i e
punc a being iple labeled. The ela ionship was analyzed quan-
i a i ely in laminas I/II, III–V, and IX (Table 1). In laminas IV, V,
and IX, he g ea majo i y (⬎90%) o labeled punc a we e immu-
no eac i e wi h all h ee an ibodies. The p opo ions o punc a
ha we e iple labeled in lamina III and laminas I–II we e 82 and
36%, espec i ely. Be ween 98 and 99% o he punc a ha we e
analyzed showed GluR2 immuno eac i i y in all a eas. The ela-
i e in ensi y o labeling o he h ee subuni s a ied be ween
punc a (Fig. 2d–o). Labeling o punc a wi h he GluR3 and GluR4
an ibodies was gene ally much weake in laminas I–II han in
o he laminas, al hough a ew punc a wi h s ong GluR3 o
Figu e 2. Colocaliza ion o AMPA subuni s. a–cshow he same ield om lamina II scanned o e eal GluR1 ( ed) and GluR2
(g een). No e ha all o he GluR1-immuno eac i e punc a a e also labeled wi h he GluR2 an ibody, and ha some GluR2-
immuno eac i epunc aa eno s ainedwi h heGluR1an ibody(3o hesea eindica edwi ha ows).d–oshow he ela ionship
be weenimmunos aining o GluR2(blue),GluR3(g een),andGluR4( ed)inlaminaI(d–g),laminaIV(h–k),andlaminaIX(l–o).
Ineachcase,ame gedimageisshownin he igh ( g,k,o).Nea lyallo helabeledpunc ainlaminasIVandIXshowall h ee ypes
o immuno eac i i y.Al houghsomecolocaliza iono GluR2wi hGluR3and GluR4immuno eac i i ycanbeseenind–g, someo
he punc a a e only labeled wi h he GluR2 an ibody (2 a e indica ed wi h a ows). All images show single op ical sec ions. Scale
ba : a–o,2
␮
m.
Table 1. Colocaliza ion o GluR2, GluR3, and GluR4
Subuni combina ion
Lamina 2/3/4 2/3 2/4 3/4 2 3 4 GluR2⫹(%)
I/II 36 32.7 7.3 1 22.3 0.7 0 98.3
(31–39) (28–35) (2–10) (0–2) (22–23) (0–1)
III 82 6.7 6.3 1.3 3.3 0.3 0 98.3
(77–89) (4–9) (4–9) (0–3) (1–8) (0–1)
IV 92 0.7 5 0 0.7 1.3 0.3 98.4
(91–93) (0–1) (3–7) (0–1) (1–3) (0–1)
V 94.3 2 2 0.3 0.3 0.7 0.3 98.6
(92–96) (0–3) (1–3) (0–1) (0–1) (0–1) (0–1)
IX 96 0 1.3 1.3 0.7 0.7 0 98
(95–98) (1–3) (0–2) (0–2) (0–1)
Meannumbe s o punc ashowing di e en pa e ns o colocaliza ion o immunos aining o GluR2, GluR3, andGluR4. One hund ed immuno eac i epunc a
we e selec ed om each egion in each o h ee a s. Ranges a e gi en in pa en heses.
5770 •J. Neu osci., June 23, 2004 •24(25):5766–5777 Nagy e al. •Synap ic AMPA Recep o s in Spinal Co d

GluR4 immuno eac i i y we e seen he e (Fig. 2d–g). In con as ,
he in ensi y o GluR2 immuno eac i i y in indi idual punc a
was gene ally highe in laminas I and II han in o he laminas.
Rela ionship o GluR1 and GluR2 o glu ama e gic axons
De ec ion o VGLUT1, VGLUT2, and CGRP immuno eac i i y
and IB4 binding wi h he TSA me hod be o e pepsin ea men
esul ed in labeling pa e ns iden ical o hose epo ed p e i-
ously o hese ma ke s in he a spinal co d (Sakamo o e al.,
1999; Va oqui e al., 2002; Todd e al., 2003). This shows ha he
TSA eac ion can be used o p ese e immunos aining o labile
epi opes be o e an igen unmasking wi h pepsin. Mo e han 90%
o he labeled bou ons in each o hese popula ions we e in con-
ac wi h one o mo e GluR2-immuno eac i e punc a (Figs. 3-5;
Table 2), and o he 4800 glu ama e gic bou ons analyzed in his
pa o he s udy, 96.2% had a leas one con ac .
In all egions examined, each VGLUT1-immuno eac i e bou-
on was gene ally associa ed wi h mo e han one GluR2 punc um
(Table 2). In many cases, pa icula ly in laminas IIi and III, clus-
e s o GluR2 punc a we e seen su ounding VGLUT1-
immuno eac i e bou ons wi h an appea ance ha was consis en
wi h a glome ula a angemen (Fig. 3a–c). In laminas IIi and III,
he mean numbe s o GluR2 punc a associa ed wi h each
VGLUT1 bou on we e 4.2 and 3.7, espec i ely, whe eas o lam-
ina IX, his igu e was 5.5 (Fig. 3, Table 2). The mean numbe s o
GluR2 punc a associa ed wi h VGLUT1 bou ons in he o he
laminas analyzed a ied om 1.9 o 2.8 (Table 2). VGLUT2 bou-
ons we e associa ed wi h signi ican ly ewe GluR2 punc a (Fig.
4), wi h he mean numbe o punc a pe bou on a ying om 1.4
o 1.9 in di e en laminas (Table 2) (gene al linea model wi h
Tukey’spos hoc es ; p⬍0.05 o lamina I; p⬍0.001 o all o he
egions examined). Fo IB4-labeled axons in lamina II, he mean
numbe o GluR2 punc a pe bou on was 4.8 (Fig. 5a–c), whe eas
o CGRP-immuno eac i e bou ons in laminas I–IIo, he co e-
sponding alue was 2.6 (Fig. 5d– ) (Table 3).
Many GluR1-immuno eac i e punc a we e obse ed in he
do sal ho n, and i ually all o hese we e also GluR2 immuno-
eac i e (Figs. 3–5, Table 2). Only 18 (0.14%) o he 12,493
punc a iden i ied on he 4800 bou ons examined in his pa o
he s udy showed GluR1 bu no GluR2 immuno eac i i y. Al-
hough punc a ha we e bo h GluR1 and GluR2 immuno eac i e
we e associa ed wi h each ype o axonal bou on in he do sal
ho n, his a angemen was no uni o m. In laminas I and II,
be ween 30 and 40% o he GluR2 punc a associa ed wi h
VGLUT1 bou ons was also GluR1 immuno eac i e, whe eas o
VGLUT2 bou ons, he p opo ion was much lowe , be ween 13
and 17% (Table 2). Simila ly in laminas III and IV, 19% o punc a
adjacen o VGLUT1 bou ons was GluR1 immuno eac i e, com-
pa ed wi h 2–4% o VGLUT2 bou ons (Table 2). The p opo -
ion o punc a showing bo h GluR1 and GluR2 immuno eac i i y
was signi ican ly highe o VGLUT1 han o VGLUT2 bou ons
in each o laminas I–IV (gene al linea model wi h Tukey’spos
hoc es ; p⬍0.05 o laminas I and III; p⬍0.01 o laminas IIo,
Iii, and IV). Fo he IB4 and CGRP bou ons, he p opo ions o
GluR2 punc a ha we e also GluR1 immuno eac i e we e 34 and
31%, espec i ely (Table 3), and hese alues we e signi ican ly
di e en om hose o VGLUT2 bou ons in he co esponding
laminas (gene al linea model wi h Tukey’spos hoc es ; p⬍
0.05). These esul s sugges ha GluR1-con aining synapses a e
p e e en ially associa ed wi h p ima y a e en e minals.
To de e mine whe he a signi ican p opo ion o he GluR2-
immuno eac i e punc a seen a e pepsin ea men ep esen ed
nonsynap ic (e.g., cy oplasmic) ecep o s, we analyzed sec ions
eac ed wi h GluR2, VGLUT1, and VGLUT2 an ibodies. Be ween
49 and 60% (mean, 55.7) o GluR2 punc a in laminas I–II we e
associa ed wi h a bou on ha was VGLUT1 o VGLUT2 immu-
no eac i e. The co esponding alues o laminas III–IV we e
84–89% (mean, 86.3), and o lamina IX hey we e 83–87%
(mean, 84.7). This sugges s ha o he deep do sal ho n and
en al ho n, a leas 85% o punc a a e associa ed wi h glu ama-
e gic synapses, and because VGLUT1 and VGLUT2 an ibodies
do no s ain all glu ama e gic axons, his igu e is p obably an
unde es ima e. Because bou ons belonging o C ibe s a e o en
no labeled wi h VGLUT1 o VGLUT2 an ibodies (Todd e al.,
2003), hese a e likely o accoun o many o he GluR2 punc a
ha we e no in con ac wi h a VGLUT1- o VGLUT2-con aining
bou on in laminas I and II (Fig. 5).
Elec on mic oscopy
Al hough he ul as uc u e was comp omised by pepsin ea -
men , i was equen ly possible o iden i y immuno eac i e syn-
apses wi h bo h GluR1 and GluR2 an ibodies, and in all cases,
Figu e 3. GluR1 and GluR2 punc a associa ed wi h VGLUT1-immuno eac i e bou ons. In
eachcase,GluR2immunos ainingisshown ing eenin hele panel,GluR1inbluein hecen e
panel,and hese ha ebeenme gedwi hVGLUT1 s aining( ed)in he igh panel.a–ca e om
lamina IIi. d– a e om lamina IV. g–ia e om lamina IX. In all cases, he VGLUT1 bou ons
con ac GluR2-immuno eac i epunc a.In hedo salho n,someo hesepunc aa ealsoGluR1
immuno eac i e. In lamina IIi, hese punc a o en su ound he VGLUT1 bou on, wi h an ap-
pea ance ha issugges i eo aglome ula a angemen (a ows).No e ha occasionalo e lap
be ween GluR-immuno eac i e punc a and axonal s aining is likely o be a ibu able o ei he
in agina iono adend i ein o a bou on o obliqui y o he synapse wi h espec o heplaneo
sec ion. All images we e ob ained om single op ical sec ions. Scale ba , 2
␮
m.
Nagy e al. •Synap ic AMPA Recep o s in Spinal Co d J. Neu osci., June 23, 2004 •24(25):5766–5777 • 5771
eac ion p oduc was es ic ed o he pos synap ic p o ile and
associa ed wi h he pos synap ic densi y (Fig. 6). This is consis-
en wi h a cy oplasmic loca ion o he C e minal o hese sub-
uni s and con i ms ha punc a seen adjacen o glu ama e gic
bou ons wi h con ocal mic oscopy co espond o pos synap ic
ecep o s.
Wi h bo h an ibodies, eac ion p oduc was occasionally seen
in cell bodies, nonsynap ic egions o dend i es, and esicle-
con aining p o iles. While scanning an a ea ha con ained 50
immuno eac i e synapses on each sec ion, we ound h ee im-
muno eac i e esicle-con aining p o iles in each o he wo
GluR1 sec ions analyzed and one in one o he GluR2 sec ions.
One o hese p o iles ( ha was ound in he GluR2 sec ion) was
p esynap ic a an asymme ical synapse and was he e o e p e-
sumably an axonal bou on. The eac ion p oduc in his p o ile
was no associa ed wi h he p esynap ic memb ane bu was lo-
ca ed in he cen e o he bou on. The o he six esicle-con aining
p o iles did no o m synapses in hese sec ions, and i was no
possible o de e mine whe he hese we e axons o esicle-
con aining dend i es. In o maldehyde- ixed issue ea ed wi h
pepsin and iewed wi h con ocal mic oscopy, we ne e saw
GluR1-immuno eac i e cell bodies, and GluR2-immuno eac i e
cell bodies we e only seen in deepe pa s o he sec ion. The
inding o occasional GluR1- and GluR2-immuno eac i e cell
bodies nea he su ace o pepsin- ea ed sec ions p ocessed o
elec on mic oscopy sugges s ha glu a aldehyde can p o ec
nonsynap ic AMPA ecep o subuni s om loss du ing pepsin
diges ion. I glu a aldehyde also p ese es some p esynap ic e-
cep o s du ing pepsin ea men , his may accoun o some o
he labeled esicle-con aining p o iles seen wi h EM.
Nume ous synapses we e seen in he sec ions ha had been
eac ed wi h GluR1 and GluR2 an ibodies wi hou pepsin ea -
men ; howe e , none o hese was immuno eac i e wi h ei he
an ibody. This is consis en wi h ou ailu e o see punc a e s ain-
ing o GluR1 and GluR2 wi h con ocal mic oscopy on sec ions
ha had no been pepsin ea ed.
Ou indings wi h con ocal and elec on mic oscopy, aken
oge he wi h s udies ha ha e e ealed p esynap ic ecep o s on
ligh ly ixed ma e ial (Lu e al., 2002), sugges ha ecep o sub-
uni s a di e en subcellula loca ions a e a ec ed di e en ly by
ixa ion and pepsin ea men . I is likely ha no single me hod
can p o ide simul aneous labeling o ecep o subuni s in all
loca ions.
De ec ion o GluR1-pS845 a e capsaicin injec ion
S ong punc a e immunos aining co esponding o GluR1-
pS845 was obse ed in sec ions om he a s ha ecei ed injec-
ions o capsaicin 10 min be o e ixa ion. This was p ima ily e-
s ic ed o he medial hal o laminas I–II on he ipsila e al side
(Figs. 7, 8a– ), al hough a ew mode a ely immuno eac i e
punc a we e seen on he con ala e al side. The dis ibu ion o
s ongly labeled punc a in he medial pa o he ipsila e al supe -
icial do sal ho n ma ches ha o nocicep i e a e en s ha inne a e
he plan a su ace o he oo . The GluR1-pS845-immuno eac i e
punc a we e almos in a iably also labeled wi h he GluR2 an i-
body (Fig. 8a–c; Table 4) bu we e g ea ly ou numbe ed by
GluR2immuno eac i e punc a ha lacked GluR1-pS845 immuno-
eac i i y. Be ween 67 and 126 punc a wi h pixels exceeding he
h eshold luminance alue we e seen ipsila e ally in medial laminas
I–II in a single op ical sec ion (1.66–3.3/1000
␮
m
2
) compa ed wi h
be ween4and14punc a(0.1–0.37/1000
␮
m
2
)in heco esponding
pa o he con ala e al do sal ho n (Table 4). This di e ence was
signi ican (Mann–Whi ney, one- ailed U es ; p⬍0.05). S aining
Figu e 4. GluR1 and GluR2 punc a associa ed wi h VGLUT2-immuno eac i e bou ons. In
eachcase,GluR2immunos ainingisshown ing eenin hele panel,GluR1inbluein hecen e
panel,and hese ha ebeenme gedwi hVGLUT2 s aining( ed)in he igh panel.a–ca e om
lamina I. d– a e om lamina V. g–ia e om lamina IX. No e ha VGLUT2 bou ons a e associ-
a edwi h ewGluR2-immuno eac i epunc aand ha someo hepunc aincon ac wi h hese
bou ons in he do sal ho n a e also GluR1 immuno eac i e. All images we e ob ained om
single op ical sec ions. Scale ba , 2
␮
m.
Figu e 5. GluR1 and GluR2 punc a associa ed wi h e minals o ine a e en s in lamina II.
a–cshowGluR2(g een)andGluR1(blue)immuno eac i i y, oge he wi hbindingo IB4( ed).
The IB4-labeled bou ons a e in con ac wi h GluR2-immuno eac i e punc a, some o which a e
also GluR1 immuno eac i e. In many cases, he IB4 bou ons we e su ounded by immuno eac-
i epunc a wi hanappea ance ha wassugges i eo aglome ula a angemen (a ow).d– ,
GluR2 (g een), GluR1 (blue), and CGRP ( ed) immuno eac i i y. The CGRP bou ons a e associ-
a ed wi h GluR2-immuno eac i e punc a, some o which a e also GluR1 immuno eac i e. All
images we e ob ained om single op ical sec ions. Scale ba , 2
␮
m.
5772 •J. Neu osci., June 23, 2004 •24(25):5766–5777 Nagy e al. •Synap ic AMPA Recep o s in Spinal Co d
was abolished by p eabso bing he an ibody wi h he phosphopep-
ide bu no he dephosphopep ide (Fig. 8g–i), indica ing ha i is
speci ic o GluR1-pS845.
In addi ion o he s ongly labeled punc a, nume ous e y
weakly immuno eac i e punc a we e seen
h oughou he supe icial do sal ho n on
bo h sides wi h he GluR1-pS845 an i-
body, and again hese we e gene ally also
GluR2 immuno eac i e. This weak s ain-
ing is likely o e lec a low basal le el o
phospho yla ion o GluR1 subuni s a glu-
ama e gic synapses, whe eas he subs an-
ial inc ease in s aining in he medial pa
o he supe icial do sal ho n ipsila e ally
p esumably esul s om inc eased phos-
pho yla ion o GluR1 a S845 in esponse
o capsaicin injec ion (Fang e al., 2003).
Discussion
The main inding o his s udy was ha
an igen unmasking wi h pepsin e ealed
punc a e labeling o all ou AMPA e-
cep o subuni s in he spinal co d. These
punc a ep esen synap ic ecep o s, as
judged by hei ela ionship wi h glu ama-
e gic bou ons and he esul s o elec on
mic oscopy. The e was ex ensi e colocal-
iza ion be ween he subuni s, wi h GluR2
being p esen a nea ly all immuno eac-
i e punc a h oughou he g ay ma e ,
which sugges s ha GluR2 is almos uni e sally exp essed a
AMPA-con aining synapses h oughou he spinal co d. In addi-
ion, we showed ha noxious s imula ion leads o phospho yla-
ion o synap ic GluR1 subuni s a he S845 esidue.
Figu e6. Elec onmic oscopicappea anceo GluR1andGluR2in hesupe icialdo sal ho n a e pepsin ea men . a–cshow
GluR1 immunolabeling. d– show GluR2 immunolabeling. Immunope oxidase eac ion p oduc is associa ed wi h he pos syn-
ap ic aspec o synapses (a ows). Scale ba , 0.5
␮
m.
Table 2. Associa ion o GluR1 and GluR2 punc a wi h VGLUT-immuno eac i e bou ons
Lamina
VGLUT1 VGLUT2
Pe cen age o
bou ons wi h ⱖ1
GluR2 punc a
Pe cen age o
bou ons wi h ⱖ1
GluR1 punc a
Mean GluR2
punc a pe
bou on
Pe cen age o
GluR2 punc a
wi h GluR1
Pe cen age o
bou ons wi h ⱖ1
GluR2 punc a
Pe cen age o
bou ons wi h ⱖ1
GluR1 punc a
Mean GluR2
punc a pe
bou on
Pe cen age o
GluR2 punc a
wi h GluR1
I 91.7 40.3 1.9 29.8** 92 16.7 1.4 12.9
(89–94) (34–44) (1.6–2.3) (23.1–35.0) (88–100) (9–24) (1.1–1.7) (8.2–17.2)
IIo 95.0 64.3 2.8 40.2 97.3 24.7 1.6 17.0*
(91–99) (53–71) (2.6–3.4) (30.7–55.0) (96–100) (19–28) (1.4–1.8) (15.4–19.6)
IIi 98.3 76.3 4.2 38.1** 96.7 24 1.6 16.9
(97–100) (73–82) (3.4–4.7) (32.7–41.0) (92–100) (16–29) (1.4–1.7) (11.6–20.6)
III 96.0 44.3 3.7 19.0*** 95 5 1.4 3.6
(95–97) (30–60) (3.0–4.2) (15.5–24.0) (90–100) (4–7) (1.3–1.5) (2.7–4.9)
IV 93 31.7 2.6 19.4** 96.7 4.3 1.6 1.9***
(89–95) (29–33) (2.2–3.2) (14.7–24.9) (96–98) (0–11) (1.4–1.9) (0–4.3)
V 94 21 2.6 10.4** 97.7 1.7 1.5 0.8*
(91–99) (16–26) (2.4–2.9) (7.3–14.0) (95–100) (0–4) (1.2–1.7) (0–1.9)
IX 97.3 4.3 5.5 0.8** 97.7 0 1.9 0
(97–98) (0–11) (3.9–6.4) (0–1.8) (97–99) (1.6–2.2)
Thep opo iono VGLUT1- andVGLUT2-immuno eac i ebou ons ha we eadjacen o GluR1 andGluR2punc a, oge he wi h he mean numbe o GluR2punc a ha con ac edeach bou on ineachg oup,and hepe cen ageo hese ha
we e also GluR1 immuno eac i e a e shown. One hund ed bou ons wi h each ype o VGLUT immuno eac i i y we e examined in each egion in h ee a s. Ranges a e gi en in pa en heses. Ve y ew punc a we e GluR1 bu no GluR2
immuno eac i e. The numbe o hese in he sample o 300 VGLUT-immuno eac i e bou ons in each g oup is indica ed by as e isks as ollows: *one; ** wo; *** h ee.
Table 3. Associa ion o GluR1 and GluR2 punc a wi h CGRP- and IR4-labeled bou ons
Lamina
CGRP IB4
Pe cen age o
bou ons wi h ⱖ1
GluR2 punc a
Pe cen age o
bou ons wi h ⱖ1
GluR1 punc a
Mean GluR2
punc a pe
bou on
Pe cen age o
GluR2 punc a
wi h GluR1
Pe cen age o
bou ons wi h ⱖ1
GluR2 punc a
Pe cen age o
bou ons wi h ⱖ1
GluR1 punc a
Mean GluR2
punc a pe
bou on
Pe cen age o
GluR2 punc a
wi h GluR1
I 93.3 50.3 2.6 34.4
IIo (92–94) (38–58) (2.4–2.9) (27.0–42.3) 100 77.3 4.8 30.7
IIi (69–93) (4.5–5.2) (24.9–35.4)
Thep opo iono CGRP-immuno eac i eandIB4-labeledbou ons ha we eadjacen oGluR1 andGluR2punc a, oge he wi h hemeannumbe o GluR2 punc a ha con ac edeach bou onineachg oup,and hepe cen ageo hese ha
we e also GluR1 immuno eac i e a e shown. One hund ed bou ons o each ype we e examined om laminae I–IIo (CGRP) o lamina II (IB4) in h ee a s. Ranges a e gi en in pa en heses. In hese popula ions, all GluR1-immuno eac i e
punc a we e also GluR2 posi i e.
Nagy e al. •Synap ic AMPA Recep o s in Spinal Co d J. Neu osci., June 23, 2004 •24(25):5766–5777 • 5773
Lamina dis ibu ion o AMPA subuni s and ela ionship o
glu ama e gic axons
Al hough s udies wi h in si u hyb idiza ion and con en ional im-
munocy ochemis y ha e p o ided in o ma ion abou he lami-
na dis ibu ion o neu ons wi h di e en subuni s, hey do no
allow us o de e mine he p opo ions o neu ons ha exp ess a
pa icula subuni o he ex en o colocaliza ion o subuni s
wi hin neu ons o a indi idual synapses. Based on he assump-
ion ha all spinal neu ons possess AMPA ecep o s, ou esul s
sugges ha a midlumba le els, all (o i ually all) do sal ho n
neu ons and mo oneu ons exp ess he GluR2 subuni , ha
GluR1 is es ic ed o ce ain do sal ho n cells (pa icula ly hose
in he supe icial laminas) and is no exp essed by mo oneu ons,
and ha GluR3 and GluR4 a e exp essed by all (o i ually all)
mo oneu ons, by he g ea majo i y o neu ons in he deep do sal
ho n (laminas III–V), and by some o hose in he supe icial
laminas. The ex ensi e colocaliza ion o GluR2, GluR3, and
GluR4 a punc a in laminas IV, V, and IX sugges s ha neu ons in
hese laminas ha e all h ee subuni s a he g ea majo i y o
synapses ha possess AMPA ecep o s.
Because ou indings sugges ha he GluR2 subuni is almos
uni e sally p esen a AMPA-con aining synapses in midlumba
spinal co d, we examined he ela ionship o GluR2 o se e al
ypes o glu ama e gic axon. Mos VGLUT1-immuno eac i e
bou ons om lamina IIi–IX a e cen al e minals o myelina ed
p ima y a e en s, whe eas he majo i y o VGLUT2-con aining
axons a e p obably de i ed om exci a o y in e neu ons (Todd
e al., 2003). We ound ha VGLUT1 bou ons we e adjacen o
signi ican ly mo e punc a han hose wi h VGLUT2 in all laminas
examined. Punc a seen wi h con ocal mic oscopy do no neces-
sa ily ep esen indi idual synapses bu could co espond o ac-
i e si es wi hin a synapse. Many VGLUT1-immuno eac i e p o-
iles in lamina IX a e e minals o Ia muscle-spindle a e en s
(Todd e al., 2003). Pie ce and Mendell (1993) epo ed ha mos
Ia a e en bou ons in he ca we e p esynap ic o a single elemen ,
bu ha he mean numbe o ac i e si es a hese synapses was 6.1.
The mean numbe o GluR2 punc a ha we ound on VGLUT1
bou ons in lamina IX was 5.5, and hese punc a he e o e p e-
sumably co espond o ac i e si es wi hin indi idual synapses.
VGLUT1-immuno eac i e bou ons in laminas IIi and III p oba-
bly include A
␦
down-hai a e en s, which o m cen al axons o
ype II synap ic glome uli (Re´ helyi e al., 1982; Ribei o-da-Sil a
and Coimb a, 1982), and we o en obse ed clus e s o GluR2
punc a su ounding VGLUT1 axons in his egion. A
␤
a e en s
e mina e widely in laminas III–V and o m simple synap ic
a angemen s on a ew di e en pos synap ic elemen s (Maxwell
and Re´ helyi, 1987). Consis en wi h his, he numbe o punc a
associa ed wi h VGLUT1 bou ons was lowe in laminas IV and V.
Less is known abou synap ic connec ions o med by axons o
exci a o y in e neu ons, bu he numbe o punc a associa ed
wi h VGLUT2-immuno eac i e bou ons (1.4–1.9) sugges s ha
hese ha e simple a angemen s, in many cases wi h only a single
pos synap ic elemen . Many o he IB4-labeled e minals co e-
spond o ype I synap ic glome uli (Ribei o-da-Sil a and Coim-
b a, 1982), and hese we e also equen ly su ounded by clus e s
o GluR2 punc a (Fig. 5).
In e es ingly, he p opo ions o GluR2 punc a ha we e also
GluR1 immuno eac i e we e signi ican ly highe o VGLUT1
bou ons han o VGLUT2 bou ons in laminas I–IV, and in he
supe icial laminas IB4- and CGRP-labeled bou ons we e also
associa ed wi h mo e GluR1 punc a han we e VGLUT2 bou ons.
This sugges s ha GluR1 is mo e o en p esen a synapses
o med by p ima y a e en s han a hose o med by exci a o y
in e neu ons. The e a e wo possible explana ions o his: ei he
neu ons ha exp ess GluR1 ecei e ela i ely mo e p ima y a -
e en synapses han o he neu ons, o al e na i ely GluR1 is se-
lec i ely a ge ed o synapses o med by p ima y a e en s. Selec-
i e a ge ing o AMPA subuni s has been epo ed p e iously in
he CNS (Rubio and Wen hold, 1997). The p esence o GluR1 a
synapses o med by nocicep i e a e en s is impo an because
GluR1 (and GluR4-)-con aining ecep o s a e inse ed in o glu-
ama e gic synapses in esponse o synap ic ac i i y, unlike hose
ha con ain only GluR2 o GluR3, which a e cons i u i ely in-
se ed (B ed and Nicoll, 2003). This sugges s ha inse ion o
new ecep o subuni s could con ibu e o ac i i y-dependen
plas ici y a synapses o med by hese a e en s.
Baba e al. (2000) epo ed ha silen synapses ( hose wi h
NMDA bu no AMPA ecep o s) we e ex emely a e in laminas
II–III o he adul a spinal co d. Al hough we canno de e mine
he p opo ion o glu ama e gic synapses a which AMPA ecep-
o s we e p esen , ou inding ha ewe han 5% o 3000 glu a-
ma e gic bou ons analyzed in laminas I–III lacked GluR2 punc a
is consis en wi h his obse a ion.
Ca
2ⴙ
-pe meable AMPA ecep o s
One o he majo in e es s in AMPA ecep o s in he spinal co d
has been he p esence o Ca
2⫹
-pe meable ecep o s. These ha e
been iden i ied on bo h do sal ho n neu ons (Engelman e al.,
1999; S an a e al., 2000) and mo oneu ons (Vandenbe ghe e al.,
2001; an Damme e al., 2002). Ca
2⫹
-pe meable AMPA ecep-
o s in he do sal ho n a e hough o be esponsible o ac ile
allodynia a e a noxious he mal s imulus o ca ageenan-
induced in lamma ion (So kin e al., 1999, 2001). Those on mo-
oneu ons appea o be in ol ed in he exci o oxici y ha is
hough o unde lie amyo ophic la e al scle osis. Ca
2⫹
pe me-
abili y is p e en ed by he p esence o GluR2 subuni s in which
Figu e 7. Immunos aining o GluR1-pS845 in a sec ion om L4 in a a ha ecei ed an
in aplan a capsaicininjec ion.ashows hemedialpa o hesupe icialdo salho non hele
side (ipsila e al o he injec ion), and bshows he co esponding pa o he igh do sal ho n.
The do ed line indica es he bo de be ween g ay and whi e ma e . Nume ous immuno eac-
i epunc aa e isiblein hesupe icial laminas on he ipsila e al side, bu hese a e a ely seen
on hecon ala e alside.Eachimagewasob ained omap ojec iono 11op icalsec ionsa 0.5
␮
m z-sepa a ion. Scale ba , 100
␮
m.
5774 •J. Neu osci., June 23, 2004 •24(25):5766–5777 Nagy e al. •Synap ic AMPA Recep o s in Spinal Co d