Involvement of cyclic AMP and nitric oxide in immunoglobulin E dependent activation of Fc epsilon RII/CD23+ normal human keratinocytes

RETRACTED

View retraction

Amendment history:

Retraction

(January 1995)

Involvement of cyclic AMP and nitric oxide in

immunoglobulin E-dependent activation of Fc

epsilon RII/CD23+ normal human keratinocytes.

P A Bécherel, … , E Kilchherr, J J Guillosson

J Clin Invest.

1994;

93(5)

:2275-2279.

https://doi.org/10.1172/JCI117227

.

Retraction

Rapid

Publication

Involvement of Cyclic AMP and Nitric Oxide

in

Immunoglobulin

E-dependent

Activation of

FcERII/CD23+ Normal Human Keratinocytes

Pierre-Andre B1cherel,* _{M. Djavad Mossalayi,}*Fateh Ouaaz,*Liliane LeGoff,*BernardDugas,INathalie Paul-Eugene,§ CamilleFrances,*OlivierChosidow,t ErichKilchherr,11 Jean-Jacques Guillosson,' PatriceDebr6,*and Michel Arock *

*MolecularImmuno-Hematology Group (CNRS URA 625) and $Dermatology Division, Pitie-Salpetriere Hospital, 75013 Paris;

OINSERM

CJF92-10, Arnaud de Villeneuve Hospital, 34059Montpellier; I"Ciba-Geigy, Basel, Switzerland; and 'Laboratory of Hematology, Faculty ofPharmacy, 75006 Paris, France

Abstract

Epidermal keratinocytes (EK) are exposed to multiple inflam-matory stimuli and paracrine factors secreted by various der-mal cells(lymphocytes, mast cells, macrophages, fibroblasts)

during wounding, cutaneous allergy, andinfections. We have previously demonstrated that after stimulation with interleukin 4orinterferon-y,human EK express thelow-affinityreceptor for IgE (FceRII/CD23) on their surface. In the present study, weshowed that the ligation of CD23 byIgE/anti-IgE immune complexes orspecificmonoclonalantibodyinduces a dose-de-pendent release of interleukin 6 and tumor necrosis factor-a from EK. CD23-ligation activates the nitric oxide-dependent pathway,asdemonstrated by thehigh levels ofnitrites released in cell supernatants, and the accumulation of intracellular cy-clic nucleotides in EK. These second messengers are required

for IgE-dependent stimulation of cytokineproduction by these

cells, inasmuchasthis is completely abolished by the use of cAMPornitricoxide synthase antagonists. Human epithelial keratinocytes may thus participate in IgE-mediated immune responses, through their ability to express functional CD23

antigen. (J.Clin.Invest.1994. 2275-2279.)Key words: human keratinocytes-CD23 * nitric oxide-immunoglobulin E * cyclic

adenosine monophosphate

Introduction

Epidermalkeratinocytes

(EK),'

whichhavelongbeen

consid-eredas

simply

a

physical

barrier,

are an

important

sourceof

proinflammatory cytokines

including

IL-1,

IL-3, IL-6,

IL-7, GM-CSF, IFN-a and IFN-,B,

TNF-a,

TGF-,B,

and PDGF

(1,

2). Mostofthesefactorsare not

constitutively produced,

but

canbeinduced bya

variety

of

nonspecific

stimuli

(1),

includ-Address correspondenceto Dr.Michel Arock, Molecular Immuno-He-matologyGroup (CNRS URA 625),Pitie-Salp&riereHospital,47/83 Boulevard del'H6pital,75013Paris,France.

Receivedforpublication27October 1993 and inrevisedform 18

February1994.

1. Abbreviations used in this paper: EK, epidermal keratinocyte; LNMMA, NG-monomethyl-L-arginine; NOS, nitric oxide synthase; SNP, sodiumnitroprusside.

ing chemicals that induceirritantdermatitis, ultraviolet

irra-diation, tumor-promotingagents,epithelialtrauma, and other types ofinjury. EK may thus be involved in many different

inflammatoryandimmunologicalskin reactions.

Recently, wehave shown that EK(3), like their tumoral

counterparts(4),aretheonlynonhematopoieticcells ableto

expressFcERII/CD23.This antigen isa45-kDtype 2

glycopro-teinexhibiting substantial homologywithseveralC-type ani-mallectins,andisexpressed on avariety ofhematopoieticcells,

includingTand B lymphocytes, epidermal Langerhans cells, monocytes/macrophages, eosinophils, and platelets (5). Hu-manCD23 have twoisoforms, a and b, differing in their

cyto-plasmic amino-terminal tail, whereas only CD23ais detected inmice(6). The ligation of CD23a, expressed only in B lym-phocytes,triggers progression ofBcellsinthe cell cycle through

anintracellular Ca++increase(5).CD23bisoform, expressed on B and otherhematopoieticcells,wasshowntoinducean intracellular accumulation ofcAMP in human monocytes/ macrophages (7). CD23b istheform expressed by EK,after their activationby IL-4 and

IFN-'y

(3), twocytokines secreted duringtheearly phases oftheimmuneresponse(8).

In the present study, we have assayed the role ofCD23

expression inEKand theintracellularsignals induced afterthe

ligation ofthis antigenbyIgE/anti-IgEimmunecomplexes or

aspecificmAb. WeevidencedaroleforCD23 inIgE-mediated activation of EK-derived IL-6 and TNF-a release. We also

showedtheinvolvementof

cyclic

nucleotidesandnitric oxide intracellular transduction pathways inthisphenomenon.

Methods

Cells.Human EKwereobtainedfrom neonatalforeskins, expanded by

exvivoculturesasdetailed elsewhere(3).Cytokineproduction byEK required specialcultureconditions. The cellsmustbe grown to

con-fluence,andthus be inalowproliferativestate.Theyweretherefore incubated inDMEmedium(GibcoBRL,Cergy-Pontoise, France) for CD23induction andcytokinerelease.Infact,wehadpreviouslyshown thatEGF andhydrocortisone downregulatedCD23transcription (3).

EGFwas alsoshown toinhibit nitric oxide production byEK(9),an

important mediator forcytokineproduction (see below).

InductionofCD23 expressioninEKbyIL-4.ExvivoexpandedEK weretransferredtoeight-chamber glass slides(Nunc,Roskilde, DK),

at 105 cells/ml(400Ml/well). Half of thewellsweretreated with 25 ng/mlrhIL-4(recombinanthuman IL-4; Immugenex, LosAngeles, CA) for 48h.The cellswerethenfixed withacetoneat room

tempera-turebeforestainingwith thefollowingmonoclonal antibodies: anti-cy-tokeratin (Immunotech, Marseille, France); anti-vimentin (Amer-sham,LesUlis, France);andanti-CD23 mAb(clone135,Ciba-Geigy, Basel,Switzerland).Reactivity ofthese mAbswasvisualizedbythe

J.Clin.Invest.

©TheAmerican Societyfor ClinicalInvestigation,Inc.

0021-9738/94/05/2275/05 $2.00

immunoperoxidase method (Vector Laboratories, Burlingame, CA)

and asubsequentMayer's hemalum solutionstaining.

In situ hybridization

for

CD23 mRNA. EK cultures were trypsin-ized, suspended, and cytocentrifuged at 5 X 105 per slide. The slides werethen stained with a3"S-labeledcDNA probe as previously detailed (3). They were alsostained with Giemsa thereafter for better cell defini-tion, and examined by light microscopy. As a negative control, EK werealso hybridized with the murine IL-3 probe.

Cytokineproduction byEKafterCD23ligation. Confluentcultures ofEKweretrypsinizedandtransferred to 24-well plates (Becton Dick-inson, Plymouth, UK), at 8X 104cells/ml per well. They were stimu-latedby rhIL-4 (25 ng/ml), for 48 h. CD23+ EK were then treated with the anti-CD23 mAb (0-25

_gg/ml).

EK cultures were also supple-mentedwith IgE for 1 h (10,g/ml, Stallergene,Paris, France), then humananti-IgE (5-50

,gg/ml,

Nordic, Tilburg, The Netherlands) in the presence orthe absence of Fab fragments of anti-CD23 mAb (clone 135, 10 ug/ml; Ciba-Geigy),acAMP analogue

(dibutyrate-cAMP, 1 mM,Bioblock,Illkirch,France) oracAMPantagonist (Rp-cAMP, 0.2mM,Bioblock).Cell supernatantswerecollected48 h later and IL-6 and TNF-a levels assayed by specific ELISA (Medgenix,

Fleurus,Belgium).

DeterminationofcAMPand cGMP content inEK.CD23+EK (pre-viously treatedwith 25 ng/mlIL-4for 48 h)wereincubated in Eppen-dorfmicrotubes (4XI05cells/tube)in the presence ofIgE(10Ig/ml)

for 1h,followed byanti-IgE (50,g/ml),orwith anti-CD23 mAb(20

,qg/ml).

Thereactionwasstoppedatdifferenttimes (0-20 min) by

coolingthetubesat-170°C for 5 min, and thenincubatingthem 3 min

at 100°C.The tubeswerespunat10,000g for 5min, and the

superna-tants werekeptfrozenat-80°Cuntil cAMP and cGMPmeasurement

byradioimmunoassay(RIAkit, Amersham).

Nitric oxidesynthase(NOS)pathway. CD23+cells were cultured with/without NOS antagonist,

N0-monomethyl-L-arginine

(1 mM, LNMMA, SigmaChemicalCo.,St. Louis, MO),orwith NOS sub-strate,L-argininehydrochloride (2 mM, L-arg, SigmaChemicalCo.)

for 1 h beforetreatment with IgE/anti-IgE. The levels of IL-6 and

TNF-aweredetermined in 48hcell supernatantsbyELISAas

men-tioned above. Nitritesweredetermined in thesamesupernatantsbythe Griess reactionasdescribed elsewhere(10).EKwerealsostimulated withSodiumnitroprusside (SNP,SigmaChemical Co.),aNO donator, usedat 100gMfinaldilution,andthesubsequentsecretion of cyto-kines and nitriteswasthen evaluated.

Results

CD23

ligation

induces

cytokine

production by CD23+

EK. To

inducesurface expression of CD23,EK werefirsttreatedwith rhIL-4.Fig. 1 Ademonstrates theabsenceofvimentin-positive

cells(fibroblasts) in thesecultures, whereas allthecellswere cytokeratin positive with diffuse labeling (Fig. 1 B).Fig. 1 also

showsthatsurface CD23 (Cand D) and intracellular

CD23-mRNA (E and F) are presentonlyin EK treatedwithIL-4(D

andF). To assay the function of CD23, both CD23+ and

CD23-EK werethentreatedwith IgE/anti-IgEoranti-CD23

mAbs andtheirculture supernatants assayedforTNF-a and IL-6 levels. Data in Fig. 2 clearly show that theligationofCD23

inducesadose-dependent increaseincytokineproduction by EK.TheIgE-mediated cytokinereleaseis specific of the CD23

ligation,asnocytokineinductionis observed in CD23-_cells.

Inaddition, preincubationof EK with Fab fragments of

anti-CD23 mAb, abletoblockIgE-binding domain ofCD23 (11),1

hbeforeEKstimulation byIgE/anti-IgE, stronglyabolishes

cytokine release. IgEisthen ableto activate EKthroughthe

ligation of surfaceCD23.

Involvement

ofcyclic nucleotidesinCD23-mediated

stimu-lation

ofcytokine

releasein EK. AsCD23b

ligation

inducedthe

generation of cyclic nucleotides in monocytes(7), we asked whetherCD23 could transduce such signal in EK. After CD23 ligation by anti-CD23 mAborIgE/anti-IgE,asubstantial

ac-cumulation of cAMP and cGMPwasobserved (Fig. 3). While

cGMPlevelspeakedat10 minand then decreased, cAMP lev-els still increasedat20mninboth culture conditions (Fig. 3 A). Toinvestigate the role of cAMP in cytokine production by EK,wehave treated them with Rp-cAMP and But-cAMP, re-spectively inhibitor and active analogue of cAMP, before CD23 ligation. Data in Fig. 3 B show that Rp-cAMP

signifi-cantly inhibited cytokine production by IgE activated cells, whereaspretreatmentof EK with But-cAMP increased the

cy-tokine levels. On the other hand, But-cAMP alone induced cytokine production byIL-4-primed EK.These datatogether indicate that cyclic nucleotidesareinvolvedinCD23-mediated activation of EK functions.

Involvementofthe NOS pathway in the IgE stimulation of

CD23'

EK. The rapid increase in cGMP after CD23-ligation led us to investigate the involvement of NOS transduction pathway in EK. In fact, this pathway leadsto anaccumulation

of cGMP, and EK werepreviously reportedto possessNOS

(9).Data in Fig. 4 suggest the involvement ofaNOS pathway in IgE-mediated EK activation,aspretreatmentof CD23 + EK

with 1 mmolLNMMA, anantagonist of NOS, strongly

inhib-ited the IgE-induced cytokine production bythese cells.

L-ar-ginine, the substrate of NOS, was ableto overcomethis

inhibi-tory effect (Fig. 4 A). We did not overcome the inhibitory effect of LNMMA by D-arginine (not shown).The addition of

SNP, a NOgenerating chemical, alone toEK,induced the re-lease of significant levelsofcytokines,which furthersupports the role of NOin the above phenomenon (Fig.4A). In

addi-tion, we have quantified the effect of CD23 ligation on the

release of nitrites by EK. Data on Fig. 4 B demonstrate the

increased nitrites release from IgE/anti-IgE-stimulated EK, their inhibition bythe Fabanti-CD23,and further confirm the

involvement of NOS pathway in CD23-mediated EK

activa-tion.

Relationship between cAMP accumulation and NOS

path-way. InvolvementofbothmediatorsduringIgE-induced cyto-kine release from EK, led us to investigate the relationship between thesetwosignals. Fig.5 indicates that thesetwo media-tors are interdependent, as inhibitors of cAMP (Rp-cAMP) and of NO(LNMMA), respectively,decreasethe levels of

ni-trites and cAMPgenerated inIgE-activated cells. Inaddition,

But-cAMP induces nitrite production in EK, while SNP

in-creases intracellular levels of cAMPinthesecells. Meanwhile,

these levels remainlower than when the cells aredirectly

stimu-latedbyIgE/anti-IgE. Together thesedata suggestan

amplify-ingrole ofeachsignal in the increase of the otherone.

Discussion

Wehadpreviouslydemonstrated that EK could expressCD23

when stimulated with IL-4 (3). The present study indicates

that FcERII/CD23 expressed by EKis indeed functional and

constitutes the first well-defined human surface antigen that directly mediates the activation of the

NO-dependent

trans-ductionpathway inIL-4-primedEK. In contrasttomost

im-munoglobulin receptors, this 45-kD

glycoprotein

did not be-long tothe Igsuperfamily buttotheC-typeanimal lectins(5).

Afterproteolysis, solublefragments of CD23areproducedand

IgE-bind-B

Q0

0O44

4o

D -e .^^9't$'.' e- 'e;'42 Hi

_So~~~~~~~vi

vl.S a! A

AF

i'0

ti

E

Figure1. PureEKwereobtained under previously described conditions.Thiswasconfirmedbyanimmunoperoxidase labelingwithan

anti-vimentinmAbinA,which exhibitednostaining,andwithananti-cytokeratin mAbin B, which showedadiffusestaining.IL-4 induction of

CD23 in EKwasconfirmed eitherbyanimmunoperoxidase labelingwithananti-CD23 mAb before(C),and aftera48-hIL-4pretreatment (D),orbyinsitu hybridization for CD23mRNAexpression, before (E)and aftera48-h IL-4pretreatment(F).

ing site, CD23 and sCD23wereshowntodisplay other func-tional domains(11-14), includingalectinregion andasite for binding of CD21 antigen ( 15 ). The transcription and surface expression of the human b isoform of CD23 aredirectly in-ducedbyIL-4(8). Thepresentstudy shows first evidence ofa functional role ofCD23asIgEreceptorsina nonhematopoie-tic cell lineage. The activation of EK by IgE is mediated through CD23 ligationassuggested by the following findings: (a) no cytokine induction was observed in IgE/anti-IgE-treated CD23-_{EK; (b) the ligation of CD23 by specific mAb}

induced similarcytokine levelsasdid

IgE/anti-IgE,

while such an effect was not observed with anti-CD19 mAb (isotype-matchedcontroltoanti-CD23mAb;notshown); and (c) the Fabfragments of the anti-CD23 mAb completely inhibited IgE-mediated stimulation ofcytokine release from EK. Theuse

of Fabfragmentsallows theoccupancyof surface CD23 anti-gens without cell activation. EK are also able to produce sCD23, which hastheabilitytostimulatemastcell/basophil differentiation and/or histamine release by these cells (14). Therefore,inabove activationconditions, CD23 expression by EKmayhavedual functions:(a) production of inflammatory mediatorsby these cells through IgE binding; and(b)

induc-tion ofmast cell/basophilfunctionsthroughthe secretion of

solubleforms.

Cyclic nucleotides and NOS are involved in

CD23-me-diated stimulation ofEK, and givenourresults,thesetwo path-ways seem to collaborate in this stimulation. It is now well

established that NOS transductionpathwayleadstothe

genera-tion ofcGMPinvariouscellularmodels( 16-18).InEK, cyclic GMPincreaseprecededthe accumulation of cAMP. It is

there-...=...sS...v..^

Ci:, a

S a: ':

q;.b u An.*

s

.;a5*-S

_.

.1.

A.

:. '.2I

W'', _.:

...., i00 :I-N

.A

300

-_ None

200- _{Anti-CD23 F .}

200-100

+ None

1---

Anti-CD23

Fab 0.5

0 5 10 20 50 0 10 20 3'

Anti-IgE

_(gg/ml)

A

0 Anti-CD23

(jig/ml)

Figure 2. Stimulationof IL-6 and TNF-aproductionby CD23+EK after CD23ligation,either with the anti-CD23 mAb(0-25 gg/ml;

leftgraphs),orwithIgE/anti-IgE (IgE,10,ug/ml;anti-IgE,5-50

jg/ml;

rightgraphs).Effect of the Fab anti-CD23(10

IAg/ml)

on the

cytokine release(leftgraphs).Mean±standard deviation is shown for each value. *Per2x 105cells.

a 0 WC

E

a.

8

I 00

E_C.

Time (mn)

1.2

100 TNFa (pg/mi)' IL-6 (gm)

75-0.8

1so1

IgE/anti-IgE + + + + + +

But-cAMP - - + - - + +

Rp-cAMP - + - - - + -

-Figure 3.(A) Involvement of cGMP andcAMPin the CD23-me-diatedactivation of CD23+ EK. (B) Effect ofacAMPantagonist (Rp-cAMP; 0,2 mM) or of a cAMP analogue(dibutyrate-cAMP,

But-cAMP, 1 mM) on the cytokine release by these cells. *Per 2x 105 cells.

0

E

I

z

.E

Io

0. I₀₀

I. 50

z

0 1.2

0.5 .j 0.4

0

MT

6--

_bEAnCt23

_Fab

6

0

4-2

i

5 1 20 50o 10 20

Anti-IgE

_(jgg/mi)

AntI-CD23

(jigfmi)

Figure4.L-arginine is involved inIgE-mediatedactivation ofCD23+ EK(A) Effect ofaNOSantagonist (LNMMA;1mM)of the enzyme substrate(L-arginine;2mM)andof the NO donator(sodium

nitro-prusside, SNP, 100,uM)onthecytokinerelease.(B)Nitritesare

re-leased after CD23ligationby both anti-CD23(0-25ug/ml)and IgE/anti-IgE (IgE, 10,ug/ml; anti-IgE,5-50,ug/ml),andthis

secre-tion is strongly inhibitedbypreincubationwith the Fab anti-CD23 (10 sg/ml).Cellswereplatedat105/well( 1 ml in eachwell),before theirstimulation with IL-4. CD23 stimulationexperimentswere per-formed thereafter. Themeannumberof cells present after this stim-ulationwas near2X I05per well. *Per 2X I05cells.

forepossiblethat cAMPaccumulation isaresultof

phospho-diesteraseconsumption byexcesscGMP( 19-22).Thismight explain the decrease of cAMP accumulation in EK after

LNMMAadditionto EKcultures.Conversely, cAMPseems to

berequiredforthe increase of nitriteproduction. It

might

be due tothe

ability

ofcAMP tostimulate

synthesis

of inducible

NOSenzyme(23).Whatever themechanism, cAMP,cGMP,

andNOS were allreportedtobeinvolvedin the induction of TNF-atranscriptioninhematopoietic cells ( 16, 17, 24).More

recently, evidence for inducible NOS hasalsobeenpresented

inpatientsreceivingimmunotherapyorduring sepsis (25, 26). Implication ofCD23ligationandits relatedtransduction

sig-nals may thusexplaininvivo persistence of inflammatory

me-diatorsinthe

skin,

whereas theNO-dependent

pathway

could be initiated

during

the

development

of CD23-mediated pro-cesses.

Inhumanmonocytes/macrophages, earlyreportsindicated

thatCD23

ligation

by

IgE/anti-IgE

inducedanaccumulation ofintracellular cGMP and cAMP

(7);

on the other

hand,

E

U-J

U~--I

A

A

None IgE/antl-gE _ Rp-cAMP IgE/anti-IgE+Rp-cAMP But-cAMP

B

None IgE/anti-IgE LNMMA IgE/anti-lgE+LNMMA SNP 0 2 Nitrites

_(pmol/mi)*

*pwr 2.10cells

0 2 4 6

CAMP

(pmol/4lOcells)

Figure5. Interactionsbetween the NO and the cAMPpathways.(A) Nitritesarereleased after stimulationof EK withBut-cAMPalone,

but less than with IgE/anti-IgEalone. Rp-cAMP partiallyinhibits the releaseofnitritesby IgE/anti-IgE-stimulatedEK.(B) There is anintracellular accumulationofcAMP afterstimulationof EK by

SNP alone, butless than withIgE/anti-IgEalone. LNMMA partially

inhibits this intracellularincrease of cAMP in IgE/anti-IgE-stimu-latedEK.

cGMP analogues directly stimulate TNF-a production by

these cells. After CD23 ligation, monocytes produced higher

TNF-a levels ( 17), but comparable IL-6 amounts (7). The

similarities between the effect ofIgE on monocytes (27,

28)

and on EK suggest thatCD23b-relatedtransduction signals are

identicalin these two cell lineages.

Our studyalsounderlines theroleofIL-4as animportant early mediator in immune responses, inthat this cytokine

in-duces both CD23anditsbiologicalligand, IgE (8).Subsequent ligation of CD23 by IgE or otherpotential ligands can thus

stimulate the release ofinflammation cytokines from EK or

otherCD23+cells.Thishypothesis is supported bythe

simulta-neous invivo increase ofCD23 expression, serum IgE levels, and IL-4 inallergic patients (7).IgE and IL-4 are also detected in avariety of anti-microbial and anti-tumoral responses (7,

8). It will thus be of interesttoinvestigate the in situ CD23

expression

and its role ininflammed epithelial cells encoun-teredinthesediseases.

Acknowledgments

This workwassupportedin partbyagrant from AssistancePublique

desH6pitauxdeParis(Contrat de Recherche No. 922201).

References

1.Bos,J.D.,and M. L.Kapsenberg. 1993.Theskin immune system: progress incutaneousbiology. Immunol. Today. 14:75-80.

2. Luger, T. A.,and T. Schwarz. 1990.TNFa andskin.In SkinImmune

System.J. D.Bos, editor. CRC Press, Inc., Boca Raton,FL.257-291.

3.Dalloul,A.H.,M. Arock,C. Fourcade, J.Y.Beranger,P.Jaffray,P.Debre,

and M.D.Mossalayi.1992.Epidermal keratinocyte-derived basophilpromoting

activity. J.Clin.Invest.90:1242-1247.

4.Billaud,M., P.Buisson,D.Huang, N.Mueller-Lantzch,G.Rousselet,0.

Pavlitch, H.Wakasuki,J. M.Seigneurin,T.Tursz,and G. Lenoir. 1989. EBV

containing nasopharyngealcarcinoma cells express CD23andlowlevelsofthe EBV receptor CR2.J. Virol.63:4121-4128.

5. Delespesse, G., U. Sutter, M. D.Mossalayi,B.Bettler, M.Sarfati,H. Hoff-stetter, E.Kilchherr,P.DebrW,andA.H.Dalloul. 1991.Expression,structureand function of the CD23 antigen.Adv. Immunol.49:149-170.

6. Yokota, A.,H.Kikutani,T.Tanaka, R. Sato, E. L.Barsumian,M. Sue-mura, and T.Kishimoto. 1988. Two speciesof CD23:tissuspecificand IL-4

specific regulationof geneexpression. Cell.55:611-618.

7. Paul-Eugene, N., J. P. Kolb,A.Abadie, J.Gordon, G. Delespesse,M.

Sarfati, J. M.Mencia-Huerta,P.Braquet, andB.Dugas.1992.Ligation of CD23 triggers cAMPgeneration and release ofinflammatorymediators in human monocytes.J. Immunol. 149:3066-3071.

8.Paul, W. E. 1991.Interleukin4: aprototypic immunoregulatory

lympho-kine.Blood.77:1859-1870.

9. Heck,D.E., D. L. Laskin, C.R.Gardner, and J.D.Laskin. 1992. EGF suppresses nitric oxide and hydrogenperoxide productionbykeratinocytes. J. Biol.Chem. 267:21277-21281.

10. Green, L. C., D. A. Wagner, P. L. Glogowski, J. Skipper, J. S. Wishnock, andS. R. Tannenbaum. 1982. Analysis of nitrate and nitrite in biologicalfluids. Anal. Biochem. 126:131-136.

11. Mossalayi, M. D., M. Arock, G. Delespesse, H. Hoffstetter, B. Bettler,

A.H.Dalloul, E. Kilchherr, F. Ouaaz,P.Debr6,and M. Sarfati. 1992. Cytokine effectsofCD23aremediatedby an epitopedistinctfrom the IgEbindingsite. EMBO (Eur. Mol. Biol. Organ.)J. 11:3423-3428.

12.Gordon, J.,J.A.Cairns, Y. J. Liu, L. Flores-Romo,I.C. M. MacLennan, K.U. Jansen, and J. Y. Bonnefoy. 1991. Roleofmembrane and soluble CD23 in lymphocyte physiology. Monogr.Allergy.29:156-168.

13. Mossalayi, M. D., M. Arock, J.M.Bertho, C. Blanc,A.H.Dalloul, H. Hofstetter, M. Sarfati, G. Delespesse, and P.Debr671990.Proliferation of early human myeloid precursors induced byIL-I and recombinant soluble CD23.

Blood.75:1924-1927.

14.Arock, M., L. Michel, A. H. Dalloul, J. J. Guillosson, P. Debre, and M. D.

Mossalayi. 1990. Soluble CD23 increases IL-3 induction of histamine synthesis by human bonemarrowcells. Int.Arch.AllergyAppl.Immunol.21:2633-2637. 15.Aubry, J. P., S. Pochon, P. Graber, K. U. Jansen, and J. Y. Bonnefoy. 1992. CD21isaligandfor CD23 and regulates IgEproduction.Nature(Lond.).

358:505-508.

16. Billiar, T. R.,R.D.Curran, B. G. Harbrecht, J. Stadler, D. L. Williams, J.B.Ochoa, M. DiSilvio,R. L.Simmons, and S. A. Murray. 1992. Association betweensynthesisand releaseofcGMP and nitric oxide biosynthesis by hepato-cytes.Am.J.Physiol. 262:C1077-C1080.

17.Borish, L., J. J. Mascali, and L. J. Rosenwasser. 1991.IgE-dependent

cytokineproductionby human peripheralbloodmononuclear phagocytes. J.

Immunol. 146:63-67.

18. Nathan, C. F., and J.B.Hibbs. 1991. Roleofnitric oxide synthase in macrophage antimicrobial activity. Curr.- Opin.Immunol.3:65-70.

19. Beavo, J.A.,and D. H. Reifsnyder. 1990. Primary sequence of cyclic nucleotidephosphodiesterase isozymes and the design of selective inhibitors. TIPS(Trends Pharmacol.Sci.). 11:150-155.

20.Stamler, J.S.,D. J.Singel, and J. Loscalzo. 1992. Nitric oxide biochemis-try. Science(Wash. DC).258:1898-1902.

21.Stuehr, D. J., and0.W. Griffith. 1992. Mammalian nitricoxide

syn-thases.Adv.Enzymol.65:287-292.

22.Moncada, S., R.M.J. Palmer, andE.A.Higgs. 1991. Nitric oxide:

physiol-ogy,pathophysiologyandpharmacology. Pharmacol.Rev. 43:109-142. 23.Koide, M., Y. Kawahara,I.Nakayama,T.Tsuda, and M.Yokoyama.

1993.CyclicAMP-elevatingagents induce an inducible typeofnitric oxide syn-thase incultured vascular smooth muscle cells.J.Biol.Chem. 268:24959-24966.

24.Gong,J.H.,H.Renz,H.Sprenger,M.Nain, andD.Gemsa. 1990. En-hancement oftumornecrosisfactor-ageneexpressionby low doses ofprostaglan-din E2 andcyclicGMP.Immunobiology. 182:44-60.

25.Ochoa, J. B.,B.Curti,A. B.Peitzman,R. L.Simmons,T. R.Billiar,R. Hoffman,R.Rault,D.L.Longo,W. L.Urba, andA.C. Ochoa. 1992. Increased

circulating nitrogenoxidesderivativesafter humanderivatives immunotherapy: correlationwithhemodynamic changes. J. Natl.Cancer Inst.84:864-867.

26.Hibbs, J. B., C.Westenfelder,R. R.Taintor,Z.Vavrin, C.Klablitz,R. L. Baranowski,J.H.Ward,R. L.Menlove,M. P.MacMurry, J.B.Kushner,etal.

1992.Evidence forcytokine-induciblenitric oxidesynthesisfromL-argininein

patients receivingIL-2therapy.J.Clin.Invest.89:867-877.

27.Paul-Eugene,N.,J. P.Kolb,A. M.Vignola,J.Bousquet,M. D.Mossalayi,

M.Arock,P.Debre,J.Wietzerbin,J.C.Drapier,andB.Dugas. 1994.IgE acti-vates theL-argininedependantpathway in normalhumanmonocytes.J. Leuko-cyte Biol.Inpress.

28.Dessaint,J.P.,B.H.Waksman,H.Metzger, andA.Capron.1980. Cyto-philic bindingof IgEtothemacrophage:involvement of cGMPandcalcium in

macrophageactivationby dimericoraggregated ratmyelomaIgE. Cell. Im-munol.51:280-292.

-I

- H