AIDS RESEARCH AND HUMAN RETROVIRUSESVolume 13, Number 14, 1997Mary Ann Liebert, Inc.

High-Level Ability of Secretory IgA to Block HIV Type 1Transcytosis: Contrasting Secretory IgA and IgG Responses to

Glycoprotein 160

HAKIM HOCINI,1,2 LAURENT BÉLEC,1,2 SYLVIO ISCAKI,3 BENOIT GARIN,4 JACQUES PILLOT,3PIERRE BECQUART,1,2 and MORGANE BOMSEL5

ABSTRACT

The IgG and secretory IgA (S-IgA) responses to the HTV-1 envelope (gpl60 antigen) were analyzed in thecolostrum (Col) and in the cervicovaginal fluid (CVF) of HIV-1-infected women. We show IgG antibodies(Abs) to the recombinant gpl60 to be predominant as compared with the corresponding S-IgA isotype. Thelow level of the S-IgA response cannot be related to a general disturbance of the mucosal-associated lymphoidtissue (MALT) because the level of a current Ab to a caries-associated antigen from Streptococcus sobrinuswas in the normal range in these secretions. The major subclass of IgA to gpl60 was of the a\ isotype bothin Col and in CVF. However, the specific activities of S-IgA 1 and S-IgA2 were different when expressed as

the ratio of the anti-gpl60 related to total Ig of each subclass. Indeed, the specific activity of the S-IgA2 was

predominant over S-IgAl in the Col, whereas the reciprocal results were found in CVF, showing a subcom-partmentalization of these secretions. The ability of S-IgA and IgG to block one of the pathways involved inthe HTV-1 penetration across mucosa, i.e., transcytosis through epithelial cells, was evaluated using a func-tional in vitro assay. Both S-IgA and IgG Abs impaired virus transcytosis, irrespective of the level of anti-gpl60 specific activities. However, specific S-IgA was more efficient than IgG. These features suggest that mu-

cosal specific S-IgA to HIV-1 could be relevant in decreasing infectivity of HTV-1 in corporal fluids.

INTRODUCTION Little is known about humoral mucosal responses in immunedeficiency states, and the protective role of S-IgA and of IgG

The major route of HIV transmission occurs through ex- in the prevention of infections remains to be elucidated, withposure of mucosal surfaces to HTV-infected fluids, such as regard to development of immunization strategies.9 Secretory

semen, cervicovaginal fluid (CVF), and colostrum (Col).1 Most IgA and IgG Abs to HIV have been reported in nearly all ex-

immunoglobulins (Igs) in secretory sites consist of IgA poly- ternal secretions.10-13 Variable results have been presented as

mers actively transported across the epithelial barrier after bind- both unaltered and decreased S-IgA-specific responses were ob-ing with the transmembrane poly(Ig) receptor.2 Secretory IgA served in the secretions of HIV-infected subjects.14"17 We have(S-IgA) is released in the lumen, where it is involved in the first previously shown that both salivary and CVF humoral immuneimmune barrier against the entry of pathogens into the body.3"5 responses to HIV antigens are mainly associated with the IgGSecretory IgA is not the sole mucosal Ig because variable pro- isotype.18,19 In systemic humoral responses, the low IgA ac-

portions of IgG are also found in Col6 and in CVF.7 IgG in tivity was found to be restricted to the IgAj subclass.20 It hasthese secretions is believed to be serum derived but could also been proposed that a profound alteration of the regulation oforiginate from local synthesis.7,8 the mucosal S-IgA response can occur during HIV infec-

'Unité INSERM U430 (Immunopathologie Humaine), Hôpital Broussais, 75674 Paris, France.2Laboratoire de Microbiologie, Unité de Virologie, Hôpital Broussais, 75674 Paris, France.3Unité d'Immunologie Microbienne, Institut Pasteur, 75015 Paris, France.4Institut Pasteur de Bangui, Bangui, BP923 République Centrafricaine.'Institut Cochin de Génétique Moléculaire (ICGM) U332, 75014 Paris, France.

1179

1180 HOCINI ET AL.

tion.21-23 This alteration could be the major cause of the low-level specific activity of S-IgA against HTV antigens.

Among the common secretory immune responses, Abs to

Streptococcus sobrinus can be easily investigated. Indeed, thiscariogenic species induces a widespread chronic stimulation ofthe mucosal-associated lymphoid tissue (MALT). The Absagainst the cell wall carbohydrates (CHO) of S. sobrinus are

well designed to evaluate the mucosal immune disturbances inHIV-infected women. In this study, the S-IgA and IgG re-

sponses to the HIV-1 recombinant envelope (gpl60 antigen)are compared with those to CHO both in the Col and CVF. Themain antigen of the HIV-1 virus is the gpl60 antigen, which isknown to mediate the attachment of the virus to the cell sur-

face and allows cell infection. We have previously shown thatone of the mechanisms of virus penetration involves transcyto-sis of endosome-internalized HIV-1 particles generated by thecontact of HIV-1-infected cells (but not free HIV-1) and theapical surface of an epithelial cell line.24 We have also shownthat F105, an IgG against a conformational epitope on HIV-1envelope, inhibited virus transcytosis in an in vitro assay.24Therefore, we evaluate comparatively the functional role of theCol S-IgA and IgG Abs to HTV-1 by investigating their abilityto block the virus transcytosis through an epithelial cell line.

MATERIALS AND METHODS

SamplesFive of 10 Col samples and 4 of 10 CVF samples were se-

lected for their high anti-HTV activity, allowing sufficientamounts of Abs to be purified. These HIV-1 (clade A)-infectedwomen, from the Institut Pasteur de Bangui (Bangui, CentralAfrican Republic), ranged in age from 25 to 40 years. The CVFsamples were collected by washing the vagina with 3 ml ofphosphate-buffered saline (PBS) and then centrifuged to re-

move cells and insoluble mucus. Colostrum was collected bymanual expression and centrifuged; the resulting fat layer was

discarded and the supernatant was stored at -20°C. All sub-jects were at non-AIDS stages according to the 1986 classifi-cation of the Centers for Disease Control (CDC, Atlanta, GA).Control specimens consisted to two pools of 10 Col and of 20CVF from clinically and serologically healthy volunteers.

AntigensRecombinant gpl60 (MN/LAI VV.TG.9150) was kindly

provided by D. Labert (Pasteur-Mérieux Sérum et Vaccins,Lyon, France). In this construct, the env cleavage site was mu-

tated to avoid degradation into gpl20 and gp41, and the hy-drophobic transmembrane domain was removed to obtain a sol-uble glycoprotein. Cell wall carbohydrates were prepared bynitrous extraction from a cariogenic strain of 5. sobrinus as al-ready described.25

Purification of IgG and secretory IgAColostrum or CVF samples were fractionated by gel perme-

ation with three serial 1.5 X 100 cm columns of Sephacryl S300(Pharmacia, Uppsala, Sweden) in a PBS-azide solution con-

taining 0.5 M NaCl. Secretory IgA and IgG were absorbed twice

with anti-Fcr or anti-Fca beads, respectively. The final purityof each preparation was controlled by enzyme-linked im-munosorbent assay (ELISA), which showed no cross-contami-nations.

ELISA captureIgA levels were measured by a modified symmetrical sand-

wich ELISA on Nunc plates (Roskilde, Denmark), with sheephorseradish peroxidase (HRP)-labeled Abs to the human a

chain, raised and labeled in our laboratory.25,26 The plates were

coated overnight at 4°C with a 3-/xg/ml concentration of the a

chain Abs (100 /¿1/well), then blocked with 2% skim milk pow-der in PBS for 2 hr at 37°C, and washed three times with PBScontaining 0.05% (v/v) Tween 20. Three twofold dilutions ofeach sample, in duplicate, were applied for 1 hr at 37°C. Afterwashes, a-chain HRP-Abs were added for 1 hr at 37°C. TheHRP activity was revealed with o-phenylenediamine (Sigma,St. Louis, MO) and read at 492 nm with a Titertek Multiskanspectrophotometer (Flow Laboratories, Glasgow, UK). The re-

sults were drawn by comparison with a standard curve of a poolof human whey containing S-IgA (460 pg/ml). Colostrum andCVF IgG were also measured by symmetrical sandwich ELISAwith rabbit HRP-Abs to human Fcr A large pool of 250 nor-

mal human sera was taken as a standard containing IgG (14mg/ml).

Level of total secretory IgA subclassesLevels of S-IgA2 were measured by using a new capture

ELISA. Briefly, plates were coated overnight at 4°C with 3 pg/mlof Abs to human secretory component (SC),26 then overcoatingand washed. Three twofold dilutions of purified S-IgA were ap-plied for 1 hr at 37°C. After washes, 0.1 pg/m\ of IgAi protease(Boehringer GmbH, Mannheim, Germany) was added (100/tl/well) in the assay wells and incubated for 1 hr at 37°C. Afterwashes, the rabbit labeled Abs to the human (Fab')2 fragmentswere added (2 ¿ig/ml) for 1 hr at 37°C, followed by the subse-quent ELISA steps. The concentration of the IgA! protease al-ways largely exceeded that needed to cleave all S-IgAi. The con-

trol wells were incubated with the buffer without enzyme. Themean optical density (OD) of the wells in the assay correspondedto that of the uncleaved S-IgA2. The results of the S-IgAi sub-class were determined by the difference in OD between the pro-tease-treated and nontreated samples. These results were drawnfrom a standard curve with a pure colostral S-IgA preparationand expressed as a percentage of total S-IgA. This procedure was

used instead of merely subtracting OD values.

ELISA for gpl60 and CHO-specific antibodiesELISA plates were coated overnight at 4°C with 100 /tl/well

of either recombinant gpl60 (2.5 pglm\), or of CHO (0.1pglmV) diluted in PBS. After blocking with 3% skim milk andwashes, purified S-IgA or IgG was incubated for 1 hr at 37°C.After washes, the rabbit HRP-labeled Abs to human (Fab')2fragments were added, followed by the subsequent ELISAsteps. The Abs standard consisted of a human whey sample, se-

lected for its high level of Abs to gpl60 and to CHO. The spe-cific activities of S-IgA and of IgG were expressed as arbitraryunits per milligram of Igs.

MUCOSAL S-IgA AND IgG IN HIV-1 INFECTION 1181

ELISA for gpl60 subclass-specific antibodies

ELISA plates were coated overnight at 4°C with gpl60 (2.6pglmX). After overcoating, purified S-IgA or IgG was incubatedfor 1 hr. IgA] protease (0.1 /xg/ml) was added in the wells for1 hr while control wells were incubated as described above withPBS only. After washes, the rabbit HRP-Abs to SC were addedand followed by the subsequent ELISA steps of subclass de-termination. The untreated control corresponded to 100% of theOD. The results were expressed both as percentages of eachsubclass bound to gpl60 and as specific activities, i.e., the ra-

tio of the S-IgAi or S-IgA2 Abs to total S-IgAi or S-IgA2.

Apical HIV-1 transcytosis neutralization by secretoryIgA and IgG

The model used was described in Ref. 24. A carcinoma en-

dometrial cell line HEC-1 (American Type Culture Collection[ATCC], Rockville, MD) (2 X 105 cells) is grown as a tightand polarized monolayer,27,28 in RPMI 1640, supplementedwith 10% heat-inactivated fetal calf serum (FCS) and 1 mMglutamine, for 7 days on a permeable filter (polycarbonate, 12-mm diameter, 0.45-¿im pore size; Costar, Cambridge, MA). Pu-rified Col S-IgA and IgG were added to the apical chamber andincubated for 10 min at 37°C. To initiate virus transcytosis, 106CD4+ T lymphoid cells (CEM) chronically infected with HIV-1 NDK clade D29 were added to the apical chamber. Contactbetween CEM-NDK and HEC-1 cells resulted in rapid buddingof the HTV virions, followed by their transcytosis from apicalto basolateral HEC-1 cell sides.24 After 2 hr, inhibition of trans-

cytosis was estimated by p24 detection in the basolateralmedium using a p24 ELISA test (Mediators; provided by theNIH AIDS reagent program). After 2 hr of viral transcytosisthe level of p24 was estimated to be 150 pg/ml in the absenceof any Ab. The results are expressed as the mean percentage ofthe transcytosis inhibition obtained by three independent assays,as compared to the control, which consisted of nonspecific Col

S-IgA and IgG. The interassay variations were estimated to bebelow 7%.

RESULTS

Secretory IgA and IgG antibodies to gp!60 andto CHO

To compare S-IgA and IgG Abs, we evaluated the specificactivities (Ab/total Ig) to both gpl60 and CHO antigens in thefive Col samples presented in Table 1. Both the specific activ-ities of IgG and of S-IgA varied highly according to the sub-ject and to the antigen. The specific activities of IgG to gpl60always greatly exceeded those of S-IgA. In contrast, the S-IgAactivity to CHO was about 3- to 20-fold higher than the IgGactivity in Col and in CVF (Table 1). The Abs to gpl60 were

mainly of the IgG isotype whereas the Abs to CHO were mainlylocated within the S-IgA fraction. The normal control of Coldisplayed an S-IgA/IgG ratio about 20-fold that of the CHOantigen, showing that there was no change in the secretion ofboth anti-CHO isotypes in HIV-1-infected subjects. The same

pattern was observed for CVF (Table 1). Indeed, most of theanti-gpl60 Abs were of the IgG isotype whereas Abs to CHOwere located in the S-IgA fraction both in healthy subjects andin HIV-1-infected patients.

Distribution of the anti-gp160 secretory IgA subclassesThe percentages of S-IgA 1 and S-IgA2 that bound to gpl60

showed similar patterns in Col and in CVF (Fig. 1A). In thefour CVF specimens, as well as in three of five Col specimens,more than 80% of S-IgA to gpl60 was S-IgA h This percent-age was 65% for Col-64 and dropped to less than 40% forCol-84.

In the CVF samples, the specific activity to gpl60 was

largely associated with the S-IgAi subclass, exceeding twice

Table 1. Activities and Specific Activities of Secretory IgA and of IgG to gpl60and to CHO in Colostra and in Cervicovaginal Fluids from HIV-1-Infected Patients

Antibodies

Specimen HTV-1

Anti-gpl60S-IgA IgG

Anti-CHO

S-IgA IgGCol-20Col-48Col-63Col-64Col-84Pool of normal Col

CVF-1CVF-2CVF-3CVF-4Pool of normal CVF

+++++

++++

3.3a (6)b36 (120)14 (14)

139 (320)21 (20)

0

192 (35)18 (0.2)

875 (5)100 (0.8)

0

100(6)1,400 (1,300)

320 (120)654 (340)625 (1,500)

0

2,600 (8,600)1,714 (36)

10,238 (5,200)2,040 (200)

0

Nd112(50)115 (20)776 (200)368 (30)772 (600)

130 (3.5)109 (1.2)137 (5.5)95(5)

187 (3)

Nd19(2)35(2)

211 (10)18(2)35 (20)

5.2 (1)33 (0.7)

2.5 (2)8(2)

60 (0.9)

"Specific activity expressed as: Abs in arbitrary units/total S-IgA or total IgG (in mg).bAbs in arbitrary units/ml.Abbreviations: CHO, S. sobrinus cell wall carbohydrates; Nd, not determined; Col, colostrum; CVF, cervicovaginal fluid.

1182 HOCINI ET AL.

100

B

<O)

có

o.CO

CVF-1 CVF-2 CVF3 CVF4 CVF-N Cbl-20 QJ48 Cd-63 CM-« CM-S4 Cd-N

D S-lgA1 S-I9A2

MLCVF-1 CVF-2 CVF-3 CVF4 CVFN Cbl-20 Col48 Cbl-63 Cd-64 Cd-8Í Cd-N

FIG. 1. Anti-gpl60 S-IgA subclasses in the Col and in theCVF. (A) S-IgAl and S-IgA2 Abs to gpl60 expressed as per-centages of the total S-IgA anti-gpl60. (B) Specific activitiesof S-IgAl and of S-IgA2 to gpl60 expressed as the ratio of theS-IgAl or S-IgA2 Abs/Total S-IgAl or S-IgA2. Col,Colostrum; CVF, cervicovaginal fluids.

that of S-IgA2 (Fig. IB). Nevertheless, in the presented Col, thespecific activity of S-IgA2 was generally higher than that of S-IgA! (Fig. IB).

Apical HIV-1 transcytosis neutralization by colostrumsecretory IgA and IgG

The ability of Col S-IgA and IgG to inhibit the penetrationof HIV-1 through endometrial HEC-1 cells was evaluated invitro using a functional assay (Fig. 2). Both S-IgA and IgG Absimpaired HIV-1 transcytosis; however, S-IgA was more effi-cient than IgG in the five presented Col samples (Fig. 3A). Se-cretory IgA and IgG inhibited viral transcytosis, irrespective ofthe level of anti-gpl60-specific activities (Fig. 3B). In spite oftheir low activities to gpl60, the S-IgA from 3 Col specimenswas able to inhibit more than 80% of viral transcytosis. In con-

trast, the corresponding IgG displaying high specific activitiesinduced only 60% of inhibition or even less (Fig. 3B).

DISCUSSION

Our results demonstrate that Abs to gpl60 are mainly of theIgG isotype in the secretions of HIV-1-infected subjects, con-

trasting with the predominant S-IgA response to common mu-cosal pathogens in these secretions (Table 1). The analysis ofAbs to gpl60, in both Col and CVF, showed a low proportionof S-IgA as compared with the IgG response. On the other hand,S-IgA to Cho from oral bacteria, which induce local and MALTchronic stimulation, was much higher than the IgG response.The CHO from S. sobrinus was selected because normal adultsusually display a high mucosal response to this antigen.25 Thepredominance of specific S-IgA over IgG already observed withmany bacterial antigens18 suggested that this response could bethe same in HIV-infected patients as in healthy individuals. Ourresults confirm that the responses to CHO were not significantlydifferent in the secretions of HIV-1-infected and healthy sub-jects. The weak S-IgA anti-gpl60 response is probably not thedirect result of a disorder of the mucosal immune system inspite of the well-known depletion of CD4+ lymphocytes influ-encing antibody production during the progression of HIV in-

HEC-1 cells grownon a permeable support

SIgA or IgG HIV-1 infected lymphocytes (CEM-NDK)

Transcytosis inhibition estimated byp24 ELISA determination

Contact between the HIV-1 infected lymphocytes(CEM-NDK) and HEC-I cells inducing virionbudding, followed by transcytosis of the virus

FIG. 2. Experimental system used to measure the ability of Col S-IgA and IgG Abs to neutralize HIV-1 transcytosis throughHEC-1 epithelial cells barrier. AP, Apical side; BL, basolateral side; HEC-1, human endometrial epithelial cells.

MUCOSAL S-IgA AND IgG IN HIV-1 INFECTION 1183

B S-IgA (20ng)D IgG (20 (ig)

Col-63 OdI-64

OS-lgA (20 M)• IgG (20 ng>

0.01 oí i io 100

Anti-gp160 specific activity (AU/mg Ig)

FIG. 3. (A) Inhibition of HIV-1 transcytosis by Col S-IgAand IgG. Twenty micrograms of purified S-IgA or IgG was in-troduced in the apical chamber 10 min prior to addition of in-fected lymphocytes (CEM), as the source of viruses. The re-sults are expressed as the percentage of the transcytosisinhibition compared to the control, which consisted of nonspe-cific Col S-IgA and IgG. (B) the ability of S-IgA (20 pg) andIgG (20 pg) to block HTV-1 transcytosis related to their anti-gpl60 specific activities expressed as Abs in AU/mg of Ig. Col,Colostrum.

fection.30 It is more likely that other immune factors can con-

tribute to favor the production of IgG Abs. These results sup-port the idea that the mucosal immune system remains func-tional for the T cell-independent bacterial CHO antigen and thuscan be efficient for mucosal vaccinations.

The distribution of the S-IgAi and S-IgA2 Abs to gpl60 hasbeen examined and similar patterns were observed with Col andCVF. The S-IgA i represented the major subclass of Abs. Theseresults reflect the contribution of each subclass to the interac-tion with the gpl60. The biological significance of the respec-tive role of the IgA subclasses has not been fully elucidated. Ithas been demonstrated that secretory S-IgA i is selectively sus-

ceptible to cleavage by bacterial proteases on mucosal surfaces,

which are frequently colonized with IgAj protease-producingbacteria. This cleavage may interfere with the effector functionsof S-IgA], as described for salivary Abs to streptococcal anti-gens.31,32

The variations of specific activities of each subclass to

gpl60 in Col and CVF demonstrated a subcompartmentaliza-tion of these secretions (Fig. IB). These data indicate that HIVinfection has remarkably different effects on the IgA subclassdistribution of Abs in the mucosal compartments. The com-

parison of the IgA subclass-specific activities reflects the levelof activation of the plasmocytes producing each subclass inthe corresponding tissues. In addition, it is likely that the pre-cursors of IgA-producing cells in the mammary gland and inthe vagina do not originate from the same inductive sites inthe MALT.

The secretions, such as Col and CVF from HIV patients con-

tain both HIV-infected mononuclear cells and free virus parti-cles, including many defective forms. As the passage across

mucosa is the major route of HIV-1 transmission, we investi-gated possible interferences of the S-IgA and IgG Abs with thevirus transcytosis. The ability of S-IgA and IgG to inhibit, on

the apical side, the penetration of HIV-1 through epithelial cellswas evaluated in vitro using a functional assay, which does notinvolve the polymeric immunoglobulin receptor. The contactbetween chronically HIV-1-infected lymphocytes and the en-

dometrial cells (HEC-1, CD4") resulted in a rapid budding ofHIV virions, followed by their transcytosis from apical to ba-solateral HEC-1 sides.33 The transcytosis of highly infectiousHIV-1 NDK across the epithelial cells was decreased both byS-IgA and by IgG Abs. However, S-IgA were more efficientthan equal amounts of IgG. Converse results were reportedfor monomeric serum IgA and IgG Abs to neutralize cell-freeHIV-1 particles on CEM T cell lines (CD4+).34 These varia-tions are likely related to the different mechanisms involved inthe interaction of the virus with the CEM T cells, which bearthe CD4 receptor, and with CD4-negative HEC-1 epithelialcells. Both S-IgA and IgG were able to impair the transcytosisof virus irrespective of the level of their anti-gpl60 specific ac-

tivities. Interestingly, S-IgA, with its low specific activities, in-duced more than 80% of inhibition of HIV-1 transcytosis. Con-versely, IgG, with its high activity, was less efficient in blockingtranscytosis. Secretory IgA is believed to be more efficient thanIgG in the immune exclusion at mucosal surfaces, because itpossesses a dimeric structure with four Ab attachment sites.35Furthermore, we have reported the transcytosis inhibition in-duced by Abs to HIV-l/gpl20 and to galactosyl ceramide.24Galactosyl ceramide is widespread at the apical pole of HEC-1 and of various epithelial cells.36 Thus, the high-level abilityof the specific S-IgA to block HIV-1 transcytosis could be dueto the higher activities to HIV-l/gpl20 epitopes that interactwith galactosyl ceramide.

Whether the observed transcytosis inhibition induced by Absto HIV-1 in the present model corresponds to a true neutral-ization activity, as could be evidenced in a traditional neutral-ization assay, or to nonneutralizing anti-HIV Ab blocking re-

sponses, is currently unknown. Furthermore, the identities ofS-IgA-recognized HIV-1 neutralization epitopes remain to bedetermined. Indeed, the fine mapping of these epitopes wouldprovide additional information crucial to the design of an effi-cient local humoral immune vaccine against AIDS.

1184 HOCINI ET AL.

ACKNOWLEDGMENTS

We thank Dr. J.-P. Bouvet for helpful discussions and forcorrecting the manuscript. This work was supported by SIDAC-TION and by the Agence Nationale de Recherche sur le SIDA,Grants 93004, 94006, 94650 (to J. Pillot), and FF005B (to M.Bomsel). SIDACTION fellowship to H. Hocini.

REFERENCES

1. Dunn T, Newell M-L, Ades AB, and Peckham CS: Risk of humanimmunodeficiency virus type 1 transmission through breastfeed-ing. Lancet 1992;340:585-588.

2. Mostov KE, Kraehenbuhl J-P, and Blobel G: Receptor-mediatedtranscellular transport of immunoglobulin: Synthesis of secretorycomponent as multiple and larger membrane forms. Proc Nati AcadSei USA 1980;77:7257-7261.

3. Mestecky J and Mcghee JR: Immunoglobulin A (IgA): Molecularand cellular interactions involved in IgA biosynthesis and immuneresponse. Adv Immunol 1987;40:153-245.

4. Kerr MA: The structure and function of human IgA. Biochem J1990;271:285-296.

5. Kraehenbuhl JP and Neutra MR: Molecular and cellular basis ofimmune protection of mucosal surfaces. Physiol Rev. 1992;72:853-879.

6. Ogra PL and Dayton DH: Colostrum-derived immunity and ma-

ternal-neonatal interaction. Ann NY Acad Sei 1983;409:82-95.7. Hocini H, Barra A, Bélec L, Iscaki S, Preud'homme J-L, Pillot

J, and Bouvet J-P: Systemic and secretory humoral immunityin normal human vaginal tract. Scand J Immunol 1995;42:269-274.

8. Bouvet JP, Bélec L, Pires R, and Pillot J: Immunoglobulin G an-

tibodies in human vaginal secretions after parenteral vaccination.Infect Immun 1994;62:3957-3961.

9. Mestecky J, Kutteh WH, and Jackson S: Mucosal immunity in thefemale genital tract: Relevance to vaccination efforts against thehuman immunodeficiency virus. AIDS Res Hum Retroviruses1994;10:S11-S20.

10. Archibald DW, Witt DJ, Craven DE, Vogt MW, Hirsch MS, andEssex M: Antibodies to human immunodeficiency virus in cervi-cal secretions from women at risk for AIDS. J Infect Dis1987;156:240-241.

11. Holmstrom P, Syrjanen L, Laine S, Valle PSL, and Suni J: HIVantibodies in whole saliva detected by ELISA and Western-blot as-

says. J Med Virol 1990;30:245-248.12. Bélec L, Dupré T, Prazuck T, Tévi-Bénissan C, Kanga JM, Pathey

O, Lu XS, and Pillot J: Cervicovaginal overproduction of specificIgG to human immunodeficiency virus (HIV) contrasts with nor-

mal or impaired IgA local response in HIV infection. J Infect Dis1995;172:691-697.

13. Artenstein AW, VanCott TC, Sitz KV, Robb ML, Wagner KF, VeitSCD, Rogers AF, Garner RP, Byron JW, Burnett PR, and Birx DL:Mucosal immune responses in four distinct compartments ofwomen infected with human immunodeficiency virus type 1: Acomparison by site and correlation with clinical information. J In-fect Dis 1997;175:265-271.

14. Coogan MM, Sweet SP, and Challacombe SJ: Immunoglobulin A(IgA), and IgA2 antibodies to Candida albicans in whole parotidsaliva in human immunodeficiency virus infection and AIDS. In-fect Immun 1994;62:892-896.

15. Wray D, Felix DH, and Cumming CG: Alteration of humoral re-

sponses to Candida in HIV infection. BrDentJ 1990;168:326-329.16. Tylenda CA, Larsen J, Yeh C-K, Lane HC, and Fox PC: High lev-

els of oral yeasts in early HIV-1 infection. J Oral Pathol1989;18:520-524.

17. Mandel ID, Barr CE, and Turgeon L: Longitudinal study ofparotid saliva in HIV-1 infection. J Oral Pathol Med 1992;24:209-213.

18. Lu XS, Delfraissy JF, Grangeot-Keros L, Rannou MT, and PillotJ: Rapid and constant detection of HIV antibody response in salivaof HIV-infected patients; selective distribution of anti-HIV activ-ity in the IgG isotype. Res Virol 1994;145:369-377.

19. Bélec L, Georges AJ, Steenman G, and Martin PMV: Antibodiesto human immunodeficiency virus in vaginal secretion of hetero-sexual women. J Infect Dis 1989;160:385-391.

20. Kozlowski PA and Jackson S: Serum IgA subclasses and molecu-lar forms in HIV infection: Selective increases in monomer and ap-parent restriction of the antibody response to IgAl antibodiesmainly directed at env glycoproteins. AIDS Res Hum Retroviruses1992;8:1773-1780.

21. Quesnel A, Moja P, Blanche S, Griscelli C, and Genin C: Earlyimpairment of gut mucosal immunity in HIV-1-infected children.Clin Exp Immunol 1994;97:380-385.

22. Janoff EN, Jackson S, Wahl MS, Thomas K, Peterman JH, andSmith PD: Intestinal mucosal immunoglobulins during human im-munodeficiency virus type 1 infection. J Infect Dis 1994;170:299-307.

23. Trajman A, Belo M, Oliveira A, Madi K, and Elia C: Quantitativeand qualitative changes in intestinal immunoglobulin-bearingplasma cells in jejunal mucosa of 52 HIV-infected patients. BrazJ Med Biol Res 1994;27:1215-1224.

24. Bomsel M: Transcytosis of infectious human immunodeficiencyvirus across a tight human epithelial cell line barrier. Nature Med1997;3:42^17.

25. Hocini H, Iscaki S, Bouvet J-P, and Pillot J: Unexpected highlevels of some presumably protective secretory immunoglobulinA antibodies to dental plaque bacteria in salivas of both caries-resistant and caries-susceptible subjects. Infect Immun 1993;61:3597-3604.

26. Iscaki S, Geneste C, and Pillot J: Human secretory component. I.Evidence for a new antigenic specificity. Immunochemistry1978;15:401^W8.

27. Ball JM, Moldoveanu Z, Meisen LR, Kozlowski PA, Jackson S,Mulligan MJ, Mestecky JF, and Compans RW: A polarized humanendometrial cell line that binds and transports polymeric IgA. InVitro Cell Dev Biol 1995;31:196-206.

28. Kuramoto H, Tamura S, and Yukio N: Establishment of a cell lineof human endometrial adenocarcinoma in vitro. Am J Obstet Gy-necol 1972;114:1012-1019.

29. Spire B, Sire J, Zachar V, Rey F, Barré-Sinoussi F, Galibert F,Hampe A, and Chermann JC: Nucleotide sequence of HIVl NDK,a highly cytopathic strain of the human immunodeficiency virusHTV1. Gene 1989;81:275-284.

30. Müller F, Froland SS, Hvatum M, Radl J, and Brandtzaeg P: BothIgA subclasses are reduced in parotid saliva from patients withAIDS. Clin Exp Immunol 1991;83:203-209.

31. Kilian M, Mestecky J, and Russell MW: Defense mechanisms in-volving Fc-dependent functions of immunoglobulin A and theirsubversion by bacterial immunoglobulin A proteases. MicrobiolRev 1988;52:296-303.

32. Hajishengallis G, Nilolova E, and Russell MW: Inhibition of Strep-tococcus mutans adherence to saliva-coated hydroxyapatite by hu-man secretory immunoglobulin A (S-IgA) antibodies to cell sur-face protein antigen I/II: Reversal by IgAj protease cleavage. InfectImmun 1992;60:5057-5064.

33. Phillips DM: The role of cell-to-cell transmission in HIV infection.AIDS 1994;8:719-731.

34. Kozlowski PA, Chen D, Eldridge JH, and Jackson S: Contrasting

MUCOSAL S-IgA AND IgG IN HTV-1 INFECTION

IgA and IgG neutralization capacities and responses to HTV type1 gp 120 V3 loop in HIV-infected individuals. AIDS Res HumRetroviruses 1994;10:813-822.

35. Taylor HP and Dimmock NJ: Mechanisms of neutralization of in-fluenza virus by secretory IgA is different from that of monomericIgA or IgG. J Exp Med 1985;161:198-209.

36. van der Bijl P, Lopes-Cardoza M, and van Meer G: Sorting ofnewly synthesized galactosphingolipids to the two surface domainsof epithelial cells. J Cell Biol 1996;132:813-821.

1185

Address reprint requests to:H. Hocini

Laboratoire de MicrobiologieUnité de VirologieHôpital Broussais

96 rue DidotF-75674 Paris 14, France

•-mail: hakim.hocini@brs.ap-hop-paris.fr