Nsu

SLa

b

c

Md

Ae

F

a

ARR1AA

KQIUSIH

1

wpsteh

h0

Vaccine 34 (2016) 1162–1171

Contents lists available at ScienceDirect

Vaccine

j o ur na l ho me page: www.elsev ier .com/ locate /vacc ine

ovel ISCOMs from Quillaja brasiliensis saponins induce mucosal andystemic antibody production, T-cell responses and improved antigenptake

amuel Paulo Cibulskia,b, Gustavo Mourglia-Ettlinc, Thais Fumaco Teixeiraa,enora Quiricid, Paulo Michel Roeheb, Fernando Ferreirae, Fernando Silveirad,∗

FEPAGRO Saúde Animal, Instituto de Pesquisas Veterinárias Desidério Finamor, Laboratório de Virologia, Eldorado do Sul, RS, BrazilDepartamento de Microbiologia Imunologia e Parasitologia, Laboratório de Virologia, Universidade Federal do Rio Grande do Sul, Porto Alegre, RS, BrazilCátedra de Inmunología, Departamento de Biociencias, Facultad de Ciencias/Química, Universidad de la República (UdelaR), Av. Alfredo Navarro 3051,ontevideo CP. 11600, UruguayLaboratorio de Carbohidratos y Glicoconjugados, Departamento de Desarrollo Biotecnológico, Facultad de Medicina. Universidad de la República (UdelaR),v. Alfredo Navarro 3051, Montevideo CP. 11600, UruguayLaboratorio de Carbohidratos y Glicoconjugados, Departamento de Desarrollo Biotecnológico, Facultad de Medicina, Departamento de Química Orgánica,acultad de Química, Universidad de la República (UdelaR), Av. Alfredo Navarro 3051, Montevideo CP. 11600, Uruguay

r t i c l e i n f o

rticle history:eceived 15 October 2015eceived in revised form5 December 2015ccepted 17 January 2016vailable online 28 January 2016

eywords:uillaja brasiliensis

SCOMptake

a b s t r a c t

In the last decades, significant efforts have been dedicated to the search for novel vaccine adjuvants. Inthis regard, saponins and its formulations as “immunostimulating complexes” (ISCOMs) have shown to becapable of stimulating potent humoral and cellular immune responses, enhanced cytokine production andactivation of cytotoxic T cells. The immunological activity of ISCOMs formulated with a saponin fractionextracted from Quillaja brasiliensis (QB-90 fraction) as an alternative to classical ISCOMs based on Quil A®

(IQA) is presented here. The ISCOMs prepared with QB-90, named IQB-90, typically consist of 40–50 nm,spherical, cage-like particles, built up by QB-90, cholesterol, phospholipids and antigen (ovalbumin, OVA).These nanoparticles were efficiently uptaken in vitro by murine bone marrow-derived dendritic cells.Subcutaneously inoculated IQB-90 induced strong serum antibody responses encompassing specific IgG1and IgG2a, robust DTH reactions, significant T cell proliferation and increases in Th1 (IFN-� and IL-2)

ubcutaneouslyntranasally deliveredumoral and cell responses

cytokine responses. Intranasally delivered IQB-90 elicited serum IgG and IgG1, and mucosal IgA responsesat distal systemic sites (nasal passages, large intestine and vaginal lumen). These results indicate that IQB-90 is a promising alternative to classic ISCOMs as vaccine adjuvants, capable of enhancing humoral andcellular immunity to levels comparable to those induced by ISCOMs manufactured with Quillaja saponariasaponins.

© 2016 Elsevier Ltd. All rights reserved.

. Introduction

Infectious diseases are major causes of morbidity and mortalityorldwide, especially in poor and developing countries. Amongst alethora of preventive measures available in attempting to reduceuch burden, vaccines stand out as highly efficacious and cost effec-ive tools, which have been successfully used in the control or

radication of some of the most impacting infectious diseases inumans and animals [1].

∗ Corresponding author. Tel.: +598 2 4871288x1124; fax: +598 2 4873073.E-mail address: fsilveir@higiene.edu.uy (F. Silveira).

ttp://dx.doi.org/10.1016/j.vaccine.2016.01.029264-410X/© 2016 Elsevier Ltd. All rights reserved.

Subunit vaccines, whose immunogenicity relies essentially on aparticular protein or peptide, are generally less immunogenic thanvaccines based on live attenuated or whole inactivated microor-ganisms. Therefore, subunit vaccines usually require addition ofadjuvants in order to improve its immunogenicity. These have beenused in order to either induce more rapid and robust immuneresponses that correlate with increased protection or to allow dosesparing in the context of antigens which can be in limited supplyor problematic to manufacture [2].

A crucial aspect addressing the challenges in vaccine develop-

ment is antigen delivery, which encompasses administration ofdrugs to specific sites of the body, as well as the delivery of theantigen provide the necessary signals for activation and maturationof relevant antigen presenting cells (APCs) [3,4]. Most pathogens

ccine 3

iri[

vaatnat

cplfiikcahsoewtAai

i(tdPtafahip

cfpis

2

2

PoweUtAtPw

S.P. Cibulski et al. / Va

nvade the host or establish infection at mucosal surfaces. In thisegard, antigen delivery in mucosal surfaces may mimicry naturalnfection and induce local and remote specific immune responses5–7].

Most vaccine adjuvants in clinical use for human and veterinaryaccines (mainly parenteral injections) of which alum compoundsre the main representatives, induce protection by enhancingntibody responses. However, for some infections, particularlyhose caused by intracellular pathogens, specific antibodies areot sufficient to induce protection. In such cases, stimulation ofntigen-specific CD4+ or CD8+ T-cell is not only required but essen-ial for eliciting protective responses [8].

Many natural products with potential for use as adjuvants areurrently under investigation [9,10]. In veterinary medicine, triter-enoid saponins extracted from Quillaja saponaria Molina have a

ong usage history as vaccine adjuvants. In fact, a partially puri-ed mixture of saponins from Q. saponaria, named Quil A® [11],

s the most widely used saponin-based vaccine adjuvant and it isnown to stimulate both humoral and cellular responses againsto-administered antigens, with the generation of T helper 1 (Th1)nd cytotoxic cells (CTLs) responses [12]. However, Quil A® use inuman vaccines has been restricted due to undesirable side effectsuch as local reactions, hemolytic activity and occasional eventsf systemic toxicity [11,13]. A similar saponin fraction has beenxtracted from Quillaja brasiliensis leaves, named QB-90 [14], whichas found to possess adjuvant potential into levels comparable

o those of Quil A®. In addition, QB-90 was less toxic than Quil® [15–17]. Both QB-90 and Quil A® showed similar patterns ofntibody induction (IgG and subclasses) and stimulation of cellularmmunity by generation of Th1 responses [15,17,18].

An improvement to the use of saponins as adjuvants wasntroduced by the development of immune stimulating complexesISCOMs). These have been used as antigen delivery systemshat proved to exert powerful immune stimulating activities, yetisplaying reduced toxicity in several animal models [7,19,20].hysicochemical properties of ISCOMs include the cage-like struc-ures with about 40 nm in diameter, composed by aggregates of anntigen (usually proteic), cholesterol, phospholipids and saponinsrom Q. saponaria. ISCOMs were shown to up-regulate both Th1-nd Th2-like immune responses as well as to stimulate strongumoral responses (IgG1, IgG2b and IgG2a) with cytotoxic T cell

nduction [19,21]. In view of these findings, ISCOMs are consideredromising innate immune cell-stimulating adjuvants [2,22].

In this study, for the very first time, ISCOM formulations wereonstructed by replacing the Quil A® component by QB-90. Thisormulation was assessed on its adjuvant capacity using a modelrotein antigen. Safety and efficiency analyses were performed and

ts use as an effective alternative to Quil A®-based ISCOMs has beenuccessfully developed.

. Material and methods

.1. Adjuvant and vaccine preparation

Q. brasiliensis (A. St.-Hil. et Tul) Mart. leaves were collected inarque Battle (Montevideo, Uruguay). Extraction and purificationf saponins were carried out as previously described [14]. ISCOMsith ovoalbumin (OVA-ISCOMs) were prepared by the modified

thanol injection technique [23]. Briefly, ovoalbumin (OVA, Sigma,SA) solution (1 mg/mL in TBS, pH 7.4) was added either to a mix-

ure of Q. brasiliensis saponins fraction (QB-90, 1 mg/mL) or to Quil

® (QA) (Brenntag, Denmark) (1 mg/mL). Ethanol-dissolved choles-

erol (Sigma, USA) and di-palmitoyl phosphatidyl choline (Avantiolar Lipids, USA) were immediately injected into the mixture,hich were finally stirred during 48 h at 4 ◦C. After that procedure,

4 (2016) 1162–1171 1163

ISCOMs derived from QB-90 (IQB-90) and ISCOMs from QA (IQA)were obtained.

2.2. Transmission electron microscopy (TEM)

An aliquot (10 �L) of an aqueous solution of ISCOMs was placedon formvar carbon grids (200 mesh) and negatively stained with 2%phosphotungstic acid (pH 7.2) for 2 min at room temperature anddried. The samples were examined with a JEOL (JEMM 10.10) trans-mission electron microscope (JEOL, Japan) operated at an 80 kVaccelerating voltage.

2.3. In vitro and in vivo toxicity assays

Hemolytic activities of QA, QB-90 and IQB-90 were assessedas previously described [24], over a range of 10–200 �g/mL andusing a concentration of 0.5% rabbit red blood cells. Saline and Q.saponaria saponins (250 �g/mL) were used for 0% and 100% hemol-ysis, respectively. Samples were tested in triplicates. Hemolyticactivity was expressed as the sample concentration producing 50%of the maximum hemolysis (HD50).

Cytotoxicity on VERO cells (ATCC CCL-81) was determinedthrough MTT assay and by determination of lactate dehydroge-nase (LDH) release into the supernatant media with samples ofQA and QB-90. Briefly, cells were cultured in Eagle’s minimalessential medium (E-MEM) supplemented with 10% fetal bovineserum (FBS, GIBCO) and antibiotics (penicillin 100 UI/mL; strepto-mycin 100 �g/mL) (E-MEM/FBS) at 4.0 × 104 cells/well on 96-wellmicroplates and incubated for 18 h at 37 ◦C in a humid atmospherewith 5% CO2. Afterwards, the medium was removed and cellswere further incubated for 24 h with 100 �L/well of E-MEM/FBScontaining different concentrations of either QA or QB-90. ForMTT assays, 50 �L/well of MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; Invitrogen, USA) at 2 mg/mL wasadded. Cells were incubated during 4 h at 37 ◦C. After centrifugation(1400 × g for 5 min), supernatants were removed and 100 �L/wellof dimethyl sulfoxide (DMSO) were added. Optical density (OD)was measured in a microplate reader (Dynex MMXII, USA) at570 nm. Results were expressed as the percentage of each cultureOD related to the OD of untreated cells. General cytotoxicity wasreported as the concentrations that decreased viability by 50%. Forthe LDH release assay, LDH released from damaged cells into cul-ture media was quantified with the aid of a LDH assay kit (Labtest,Brazil). Survival ratio was determined by comparing the absorbancein test wells with those of positive (complete cell destruction)and negative (spontaneous cell destruction) control wells. Valueswere expressed as the highest dilution of sample concentrationwhich caused 50% LDH activity release compared to positive con-trols (EC50). Cytotoxicity assays were not performed with IQB-90because ISCOMs formulation are well known formulations retainedthe adjuvant activity of the saponins, while increasing its stability,reducing its hemolytic activity, and producing less toxicity [20,25].

Acute toxicity was evaluated as previously described [26] withsome modifications. Briefly, groups of CD-1 male mice (8 weeks ofage, n = 5) were given a subcutaneous administration of 100 �L ofQB-90 or QA in PBS (31.25, 62.5 and 125 �g/dose) on the scapu-lar region. Mice were monitored during 3 days in search for signsof toxicity (lethality, local swelling, loss of hair, and piloerection).Control mice were inoculated with 100 �L of PBS.

2.4. Antigen uptake by murine bone marrow-derived dendriticcells

Bone marrow-derived dendritic cells (BMDCs) were obtainedby differentiation of bone marrow precursors from 8 to 10 weeksold C57Bl/6 mice [27]. Briefly, femurs and tibia were flushed out

1 ccine 3

amm(ror

a2oQplt2Pcro

2wwatcISwI

2

iP(atha

pSd(w(apt(

wFet(w2

tvw0

164 S.P. Cibulski et al. / Va

nd the cells were cultured on Petri plates in RPMI 1640 cultureedium (Gibco), complemented with 10% FBS (Gibco), 50 �M 2-ercaptoethanol (Sigma), 1% HEPES (Sigma), 1% sodium pyruvate

Gibco), 1% non-essential amino acids (Gibco) and 20 ng/mL ofecombinant mouse GM-CSF (PeproTech). Medium was replacedn day 3 and BMDCs were obtained on day 10. Phenotype wasoutinely checked and 90–95% cells were CD11c+.

Antigen (Ag) uptake was performed according to protocolslready reported [28]. Briefly, BMDCs (2 × 105) were incubated with

�g of FITC-labeled OVA (OVA:FITC, Molecular Probes®) in 200 �Lf RPMI media for 120 min at 37 ◦C in the presence of 10 �g/mL ofB-90, vehicle or IQB-90 formulated with 2 �g of OVA:FITC. Onearallel experiment was performed on ice to inhibit intracellu-

ar uptake (cell surface binding controls). Then, cells were washedwice with ice-cold PBS, and resuspended in FACS buffer (PBS with% FBS and 0.1% sodium azide). Cell staining was performed withE-Cy7-conjugated anti-mouse CD11c (clone HL3, BD) and flowytometry was performed on a FACS Canto II (BD) cytometer. Fluo-escence values were reported as mean fluorescence intensity (MFI)f FITC in CD11c+ cells.

For fluorescence microscopy, 1.0 × 106 BMDCs were plated on4-well plates (Costar) and after 18 h of culture, OVA:FITC alone,ith QB-90 or formulated as IQB-90 was added. After 120 min, cellsere washed with ice-cold PBS, fixed with 4% paraformaldehyde

nd then stained with DAPI (4′,6-diamidino-2-phenylindole, Dilac-ate, Invitrogen) for 10 min at 37 ◦C. Fluorescence microscopy wasarried out with a Spot insight camera (model no. 3.1.0; Diagnosticnstruments Inc, Sterling Heights, MI) mounted over an Axiovert100 microscope (Zeiss, Göttingen, Germany). Image acquisitionas performed with Meta Imaging Series 6.1 software (Universal

maging Corporation, Downington, PA).

.5. Immunizations and sample collection

Experimental work involving animals was carried out follow-ng international guides on use and care of laboratory animals.rotocols were performed in accordance with CHEA guidelinesComisión Honoraria de Experimentación Animal) and werepproved by the Uruguayan University Research Ethics Commit-ee (approval number 070153-000531-13). Animals were properlyoused under controlled temperature (22 ± 2 ◦C) and humidity in

12/12 h light/dark cycle, with food and water ad libitum.Female Rockfeller mice of the CF-1 breed (5–6 weeks old) were

urchased from Fundac ão Estadual de Produc ão e Pesquisa emaúde (FEPPS, Porto Alegre, RS, Brazil) and were immunized onays 0 and 14 with OVA through subcutaneous (s.c.) or intranasali.n.) routes. Subcutaneous immunizations (n = 6) (in the hind neck)ere performed with 10 �g/dose of IQB-90, IQA or OVA alone

unadjuvanted). For intranasal immunization, mice (n = 7) werenesthetized with ketamine–xylazine and received 2 �g of antigener dose. Bleedings were performed immediately before inocula-ions (days 0 and 14) and 2 weeks after the second immunizationday 28). Sera were stored at −20 ◦C.

The samples, including nasal and vaginal washes as well as feces,ere collected from euthanized animals on 28 day post priming.

ecal samples were obtained from 3 to 4 freshly voided pellets fromach animal, which were weighed and collected into a 15 mL conicalube. A 10 × volume (per gram of wet feces) of extraction bufferPBS with 5% FBS and 0.02% sodium azide) was added to each tube,hich were vigorously vortexed and centrifuged at 16,000 × g for

0 min. Supernatants were collected and stored at −80 ◦C.For nasal washes, a 1 cm incision was performed parallel to the

rachea through the skin, and a midline incision was made on theentral aspect of the trachea slightly superior to the thoracic inletith a scalpel. A 25G needle was tied at the top of the trachea and

.5 mL of PBS was slowly injected. Nasal washes were collected

4 (2016) 1162–1171

through the nostrils and stored at −80 ◦C. Vaginal washes wereperformed with a micropipette. PBS (75 �L) were flushed 10 timesthrough vaginas, and samples were centrifuged 5 min at 16.000 × g.Supernatants were stored at −80 ◦C.

2.6. Splenocyte proliferation assay

Six-weeks-old female Rockfeller mice were divided into 5groups, each consisting of six mice. Animals were immunized sub-cutaneously with OVA 10 �g alone or with OVA 10 �g dissolved insaline containing QB-90 (10 �g), Quil A® (10 �g), IQB-90 (10 �g)and IQA (10 �g) on day 0. Saline-treated animals were included ascontrols. A boosting injection was given 2 weeks later.

Spleens were collected 28 days after the second immunizationunder aseptic conditions, immersed in RPMI 1640 medium (Gibco),minced, and mechanically dissociated to obtain a homogeneouscell suspension. Erythrocytes were lysed with ACK (Ammonium-Chloride-Potassium) lysis buffer. After centrifugation (380 × g at4 ◦C for 10 min), pelleted cells were washed three times inRPMI 1640 and resuspended in the same medium supplementedwith 0.05 mM 2-mercaptoethanol, 100 IU/mL penicillin, 100 �g/mLstreptomycin, 2 mM l-glutamine, and 10% fetal bovine serum(RPMI complete media). By trypan blue dye exclusion, cells count-ing revealed >95% viability. Splenocytes were seeded at 2.5 × 106

cells/mL in 100 �L of RPMI complete medium into each well of a 96-well flat-bottom microtiter plate (Nunc). Subsequently, 100 �L OVA(10 �g/mL, Sigma) or medium only was added. Plates were thenincubated at 37 ◦C in a humid atmosphere with 5% CO2. After 68 h,50 �L of MTT (1-(4,5-dimethylthiazol-2-yl)-3,5-diphenylformazanSigma) solution (2 mg/mL) was added to each well and incubatedfor 4 h. The plates were centrifuged at 1400 × g for 5 min and theuntransformed MTT was removed carefully by pipetting. Next, aDMSO solution (192 �L of DMSO with 8 �L of 1 N HCl) was addedto wells in volumes of 100 �L. After 15 min of incubation, theabsorbance was measured in an ELISA reader at 550 nm with wave-length reference fixed at 620 nm. The stimulation index (SI) wascalculated as the absorbance ratio of mitogen-stimulated culturesand the non-mitogen-stimulated cultures.

2.7. Quantification of cytokine levels in spleen culturesupernatants

Spleens were aseptically removed, rinsed and mechanically dis-rupted in RPMI media and isolated spleen cells were pelleted andincubated in RBC lysis solution. After washing with RPMI, spleno-cytes (5 × 105 cells) were re-stimulated for 3 days with OVA antigen(10 �g/mL). The supernatants were harvested and IL-2 and IFN-� cytokines were measured by capture ELISA using commercialkits from Novex (USA) and following the manufacturers’ instruc-tions. Sensitivities were 8 pg/mL and 2 pg/mL for IL-2 and IFN-�kits, respectively.

2.8. Determination of antigen-specific antibodies

Anti-OVA IgG, IgG1 and IgG2a antibodies were determinedby ELISA as described [24]. Briefly, ELISA plates (Greiner Bio-One, Germany) were coated with OVA (5 �g/mL) in acetate buffer(100 �L/well, pH 5.0) overnight at 4 ◦C. Then, plates were washedthree times with PBS containing 0.05% Tween® 20 (Sigma, USA)(PBS-T20) and blocked with 1% Tween® 20 in PBS at 37 ◦C for 2 h.Appropriately diluted samples in PBS-T20 were added in dupli-cate (100 �L/well) and incubated for 1 h at 37 ◦C. After washing

three times with PBS-T20, HRP-conjugated anti-mouse IgG (Sigma,USA), IgG1 (Invitrogen, USA) and IgG2a (Invitrogen, USA) dilutedin PBS-T20 (1:5000, 1:5000 and 1:2000, respectively) were addedto each well (100 �L/well) and plates were incubated for 1 h at

ccine 34 (2016) 1162–1171 1165

3Sf3Ewa

iaefgfen

2

d1iist

2

Dw

F9Opp

TIos

S.P. Cibulski et al. / Va

7 ◦C. After five washings, 100 �L of OPD (ortho-phenylenediamine;igma, USA) with 0.003% H2O2 were added to wells, and plates wereurther incubated for 30 min at 25 ◦C. Reactions were stopped with0 �L/well of 1 N HCl. Optical densities (OD) were measured in anLISA plate reader (Anthos 2020) at 492 nm. A pool of positive seraas used as standard curve, and antibody titers were expressed in

rbitrary units per mL (AU/mL).Anti-OVA IgA determinations were similarly performed. After

ncubation of samples for 1 h at 37 ◦C, plates were washed five timesnd 100 �L/well of goat anti-mouse IgA antibodies were added toach well (1:4000 dilution; Sigma, USA). After incubation at 37 ◦Cor 1 h and five washings, 100 �L/well of HRP-conjugated anti-oat antibodies (1:5000, Zimed, USA) were added and incubatedor 1 h at 37 ◦C and reveled as above. OVA-specific IgA titers werexpressed as OD values for samples diluted 1:10 (fecal, vaginal andasal samples).

.9. Delayed-type hypersensitivity (DTH) assay

Delayed type hypersensitivity (DTH) responses were tested 28ays post-priming. Briefly, mice were intradermally injected with

�g of OVA in one footpad of the hind limb. Thickness of thenjected footpads was measured 24 h later with a caliper. Swellingn mice inoculated with saline revealed basal conditions. OVA-pecific DTH responses in each animal were determined as thehickness of injected footpad minus average basal swelling.

.10. Statistical analyses

Statistical significance was assessed by one-way-ANOVA withunnet’s post test correction (GraphPad Prism 5.01, GraphPad Soft-are, USA). Significance was assigned at P-value < 0.05.

ig. 1. Transmission electron microscopy (TEM) of IQB-90. Microphotography of IQB-0 prepared with purified fraction of saponins from Quillaja brasiliensis (QB-90) andVA by the ethanol injection technique. QB-90 or QA formulation with 3:2:5 relativeroportions of saponins:cholesterol:phosphatidylcholine rendered mostly ISCOMarticles with an average diameter of 47 nm (overall range of 40–50 nm).

able 1n vivo toxicity assays for QB-90 and QA. Results are expressed as the percentagef mortality in each group of mice inoculated subcutaneously with QB-90 or QAaponins within 72 h.

Sample Dose

125 �g 62.5 �g 31.25 �g

Saline (n = 5) 0% (0)* 0% (0) 0% (0)QB-90 (n = 5) 60% (3) 40% (2) 0% (0)Quil A® (n = 5) 100% (5) 80% (4) 40% (2)

* () Number of death animals in each group.

Fig. 2. In vitro toxicity assays. (A) Haemolytic activity of QB-90, QA and IQB-90.Haemolysis was expressed as percent referred to saline and Q. saponaria saponins(250 �g/mL), which were used as 0% and 100% of haemolysis, respectively. (B) Cyto-toxicity of QB-90 and QA on VERO cells. Cell viability was measured by MTT 24 hafter treatment with the indicated saponin concentrations. (C) LDH release. QB-90 and QA on VERO cells was measured 24 h after treatment with the indicatedsaponin concentrations. Results are presented as the mean value ± SD (A) and ± SEM(B and C).

1 ccine 3

3

3

Ocs

Foiw

166 S.P. Cibulski et al. / Va

. Results

.1. Formulation of ISCOMs with Quillaja brasiliensis saponins

Purified fraction of saponins from Q. brasiliensis (QB-90) andVA were combined with cholesterol and phospholipid underontrolled conditions to obtain IQB-90, which formed cage-liketructures. IQB-90 structures were confirmed by transmission

ig. 3. Uptake of model antigen (OVA) by BMDCs. (A) BMDCs were loaded with 2 �g of fluoref QB-90, vehicle or as IQB-90 and their uptake was detected by flow cytometry after 120 ms shown. (B) BMDCs were treated as in (A) and examined by fluorescence microscopy (6

as used to investigate significant statistical differences (*** P < 0.001).

4 (2016) 1162–1171

electron microscopy (TEM), where the average diameter of thecage-like structures was about 47 nm (overall range: 40–50 nm)(Fig. 1).

3.2. QB-90 showed lower toxicity than QA in vitro and in vivo

Hemolytic activities of QB-90 and QA showed HD50 valuesof 88.83 ± 0.16 �g/mL and 40.43 ± 0.10 �g/mL, respectively. No

scein isothiocyanate (FITC)-conjugated OVA (OVA:FITC) in the presence of 10 �g/mLin. The mean fluorescence increase (MFI) ± standard deviation of three experiments00×). Data is shown as mean ± SEM and one-way ANOVA with Dunnet’s post-test

S.P. Cibulski et al. / Vaccine 3

4 (2016) 1162–1171 1167

significant hemolytic activity was observed at the concentra-tion used in the vaccine formulations (i.e. 10 �g/mL) (Fig. 2A).Interestingly, no hemolytic activity was determined for ISCOMsformulated with QB-90 in any tested concentration (Fig. 2A).

Similar results were obtained from the cytotoxicityassays. Results shown in Fig. 2B revealed that QA was moretoxic to VERO cells (EC50 = 50.6 ± 0.38 �g/mL) than QB-90(EC50 = 70.8 ± 0.03 �g/mL). Indeed, at 50 �g/mL, more than90% of cells exposed to QB-90 were viable, whereas cell viability ofQA-exposed cells was only 55% (P ≤ 0.001).

The LDH assay (Fig. 2C) revealed that QA promoted a moreextensive cytoplasmic content release than QB-90 (EC50 valuesof 56.59 ± 2.37 �g/mL and 72.76 ± 3.93 �g/mL for QA and QB-90, respectively). In fact, VERO cells exposed to 50 �g/mL of QAreleased approximately 40% of LDH, whereas those exposed toQB-90 released similar levels of LDH as unexposed control cells(P ≤ 0.001).

Finally, acute toxicity assay at the lowest dose tested (31.25 �g)showed no lethality or signs of local toxicity (local swelling, lossof hair and piloerection) within the mice group inoculated withQB-90, but 40% lethality in the group treated with QA (Table 1).Moreover, at the highest dose tested (125 �g) lethality showed tobe 60% and 100% in mice inoculated with QB-90 and QA, respec-tively (Table 1). Summing up, the results presented here provideevidence that QB-90 is significantly less toxic than QA, both in vivoand in vitro, suggesting that QB-90 could be used as an alternativeto QA.

3.3. BMDCs take up IQB-90 more efficiently than OVA formulatedwith soluble QB-90

In order to determine whether IQB-90 enhances antigen inter-nalization, the uptake of OVA:FITC by BMDCs in vitro in the presenceof soluble QB-90 or IQB-90 was analyzed (Fig. 3). Results in Fig. 3Ashowed that BMDCs internalized OVA:FITC at the same extenteither in the presence or in absence of soluble QB-90 (10 �g/mL)(P ≥ 0.05). However, OVA:FITC as IQB-90 was more efficiently inter-nalized by BMDCs than OVA:FITC in the presence or absence ofsoluble QB-90 (P < 0.001). In order to confirm that detected flu-orescence in BMDCs was due to internalization of OVA:FITC andnot an artifact of OVA:FITC binding to cell surface, similar exper-iments were performed at 4 ◦C showing a FITC signal abrogationindependently of the test conditions. Interestingly, similar resultswere obtained by fluorescence microscopy (Fig. 3B). Therefore, theresults presented here showed that OVA uptake by BMDCs is largelymore efficient when formulated as IQB-90 than a mixture withsoluble QB-90.

3.4. Subcutaneous administration of QB-90 orIQB-90-adjuvanted vaccine significantly increases humoral andcellular responses in mice

Immune-stimulating activities of OVA formulated in ISCOMsfrom QB-90 or QA were studied in immunized mice, both at

the humoral (i.e. antibody) and cellular (i.e. DTH and prolifera-tion assay) levels. Administration of OVA formulated with solublesaponins was also analyzed. Modulation of antibody responses(Fig. 4A and B) show that anti-OVA IgG as well as IgG1 levels were

Fig. 4. OVA-specific antibodies response of subcutaneously immunized mice. Serumtiters of anti-OVA total IgG (A), IgG1 (B) and IgG2a (C) 2 weeks after the secondimmunization. Data is shown as geometric mean titers (n = 5–6). One-way ANOVAwith Dunnet’s post-test was used to investigate significant statistical differences (**P < 0.01 and *** P < 0.001).

1168 S.P. Cibulski et al. / Vaccine 34 (2016) 1162–1171

Fig. 5. Evaluation of delayed type hypersensitivity (DTH) and splenocyte proliferation responses in OVA-immunized mice. (A) DTH; (B) splenocyte proliferation assay. Data iss ctivels

sp(iI

lrnbwO((owgco

fSwdu

FwI(

hown as mean ± SEM (n = 3 or 6 mice, DTH reaction or cell proliferation assay, respetatistical differences (* P < 0.05, ** P < 0.01 and *** P < 0.001).

ignificantly enhanced in respect to control group (P < 0.001), inde-endently of the saponin (QB-90 or QA) or the formulation typeISCOMs or soluble). However, anti-OVA IgG2a levels were onlyncreased in mice immunized with ISCOM formulations with eitherQB-90 or IQA (P < 0.001) (Fig. 4C).

In order to determine immune-stimulating activity on cellu-ar immune responses, we evaluated the induction of in vivo DTHeactions and in vitro cell proliferation in response to OVA. Sig-ificant DTH reactions were observed in mice immunized withoth ISCOM formulations (IQB-90 and IQA) and with soluble QA,hile no DTH reactions were observed for soluble QB-90 (Fig. 5A).n the other hand, animals vaccinated with QB-90 (10 �g) or QA

10 �g) saponins significantly enhanced the proliferative responsesP < 0.05, P < 00.1, respectively). The highest proliferation was alsobserved in the groups vaccinated with IQB-90 and IQA (P < 0.01)hich was not significantly different between them (Fig. 5B). Alto-

ether, results showed that IQB-90 induces strong humoral andellular immune responses. Interestingly, Th1 and Th2 modulationf antibody responses seem to be induced by IQB-90.

The profile of major Th1 cytokines (IFN-� and IL-2) was obtainedrom antigen stimulated and unstimulated splenocytes cultures.

pleen cells from OVA, QB-90, IQB-90, QA and IQA-treated miceere cultured in vitro in presence of antigen and cytokine pro-uction was measured in the supernatant after 3 days of culturesing ELISA. Mice immunized with either IQB-90 or IQA presented a

ig. 6. Analysis of IL-2 and IFN-� patterns in immunized mice. Mice were immunized with Oas administered at day 14. Splenocytes were prepared 2 weeks after second antigen dos

FN-� and IL-2 using ELISA (n = 3). Data is shown as mean ± SEM and one-way ANOVA wi* P < 0.05, ** P < 0.01 and *** P < 0.001).

y). One-way ANOVA with Dunnet’s post-test was used to investigate any significant

similar significantly increase of IFN-� (P < 0.01 and P < 0.05, respec-tively) and IL-2 (P < 0.05) cytokines levels in the supernatant thanthose obtained for the control group (Fig. 6).

3.5. Intranasal immunization with IQB-90 induces serum andmucosal specific antibody responses

Results in Fig. 7 illustrate the serum levels of OVA-specific IgG,IgG1 and IgA in i.n. immunized mice determined 2 weeks afterthe last immunization. Delivery of IQB-90 and IQA by intranasalroute induced significant serum levels of OVA-specific IgG, IgG1 andIgA (P < 0.001). However, no significant increases in specific IgG2aantibodies and DTH reactions were observed (P ≥ 0.05) (data notshown). Interestingly, no antibody induction was observed after i.n.administration of OVA formulated with either soluble QB-90 or QA(Fig. 7). Results on production of specific antibodies at mucosal sites2 weeks after i.n. immunizations are shown in Fig. 8. While adminis-tration of IQA induced significant OVA-specific sIgA responses onlyat the nasal level, IQB-90 enhanced specific sIgA responses at sev-eral distal mucosal sites (Fig. 8). Once again, no antibody productionat mucosal sites was observed after i.n. administration of OVA for-mulated with either soluble QB-90 or QA. In summary, these results

indicate that intranasal delivery of IQB-90 induces significant spe-cific antibody responses both at the systemic and mucosal level,even at distal mucosal sites.

VA alone or adjuvanted with QB-90, IQB-90, QA or IQA and a booster immunizatione. The cell supernatant was analyzed after 72 h of stimulation with OVA antigen forth Dunnet’s post-test was used to investigate any significant statistical differences

S.P. Cibulski et al. / Vaccine 34 (2016) 1162–1171 1169

Fig. 7. OVA-specific serum antibodies response from intranasal immunized mice. Serumtiters of anti-OVA IgG (A), IgG1 (B) and IgA (C) 2 weeks after the second immunizationadministered either with no adjuvant or formulated with QB-90, IQB-90, QA or IQA.Data is shown as geometric mean (n = 6–7) and one-way ANOVA with Dunnet’spost-test was used to investigate significant statistical difference (*** P < 0.001).

Fig. 8. OVA-specific antibodies response of intranasal immunized mice in mucosal sites.Mucosal titers of anti-OVA sIgA in nasal, fecal and vaginal mucosae 2 weeks after thesecond immunization. Data is shown as media ±SEM (n = 6–7) and one-way ANOVA

with Dunnet’s post-test was used to investigate any significant statistical differences(* P < 0.05 and ** P < 0.01).

4. Discussion

Adjuvants are defined as any substance usually added to vaccineantigens in order to enhance and/or modulate their immuno-genicity. Remarkably, saponins have been shown to activate themammalian immune system, turning them into interesting sub-stances in the field of potential vaccine adjuvants [12]. Quil A®

have been widely used as adjuvants in veterinary vaccines [29] andhave the exclusive capacity of stimulating Th1 immune responsesas well as inducing production of CTLs against exogenous anti-gens [30,31]. This fact makes them ideal for subunit vaccines andvaccines directed against intracellular pathogens or cancer [13,32].

In the present study a saponin fraction from Q. brasiliensis knownas QB-90 [14] – which is structurally similar to Quil A – showedlower toxic effects in vitro and in vivo. In this regard, hemolyticactivity and cytotoxicity on VERO cells of QB-90 were lower thanQA saponins, a fact previously reported [15]. Here, complementarytoxicity assays were performed and showed that QB-90 induced-release of LDH is lower than QA. Consistent results were obtainedin vivo through acute toxicity assays, which showed no lethality orsigns of local toxicity in mice after s.c. administration of 31.25 �gof QB-90 while 40% lethality with the same dose of QA. Therefore,as reported elsewhere [14,15] this work reinforces that saponins inQB-90 show significantly less toxicity. This toxicity has been associ-ated with saponins high-affinity for cholesterol, and consequentlytend to form pores in mammalian cell membranes [33]. Concern-ing this matter, ISCOMs–as well as other micellar formulationsbuilt up by saponins and cholesterol–prevent interactions betweensaponins and membranes, thus preventing hemolysis [34]. In thiswork we were able to show that IQB-90 displays no hemolysis incomparison with soluble QA or QB-90.

Particulate formulations of saponins – like Matrix-MTM andISCOMs – are known to enhance cell trafficking and activate innateimmune cells [22,35]. BMDCs were evaluated for their capacityof uptaking FITC-labeled OVA formulated with IQB-90 or mixed

with soluble QB-90. OVA:FITC was taken up more efficiently byBMDCs as IQB-90 than mixed with soluble QB-90, supporting theidea that saponin-based formulations as ISCOMs represent the best

1 ccine 3

caTieracRtasoaptOTcIcdrlsrQpoapc

bttio(ifisIttliiuInw[(O

oshidcosb

[

[

[

[

[

[

[

[

[

[

[[

[

170 S.P. Cibulski et al. / Va

hoice for activation of innate immunity [20,36]. Likewise, ISCOMss well as Quil A® formulations are known to induce potent Th1 andh2 responses, to activate CTLs and to enhance antibody responsesncluding high levels of IgG1, IgG2b and IgG2a [20,30]. Here, theffects of IQB-90 and QB-90 on humoral and cellular immuneesponses were analyzed. The results showed that ISCOMs (IQB-90nd IQA) and soluble saponins (QB-90 and QA) promoted signifi-ant Ag-specific splenocyte proliferation and strong DTH reactions.egarding humoral responses, ISCOMs, as well as saponin formula-ions, efficiently enhanced the systemic production of Ag-specificntibodies after s.c. immunizations. Remarkably, ISCOMs – unlikeoluble saponin formulations – were able to induce significant titersf serum OVA-specific IgG2a. The results indicated that IQB-90 isn efficient and potent modulator of T and B cells functions, a factreviously reported for ISCOMs containing Quil A® [20,37,38]. Fur-hermore, the immune potentiation of Q. brasiliensis saponins in aVA antigen model are in agreement with other reports [15,17].he Th1 immune response is characterized by production of theytokines IL-2, TNF-� and IFN-�, and an enhanced production ofgG2a, IgG2b and IgG3 in mice. Moreover the Th2 subset producesytokines, such as IL-4, IL- 5 and IL-10, and stimulates the pro-uction of IgG1 and secretory IgA [18,20]. Despite potent antibodyesponses and the DTH reaction stimulated by saponins formu-ations (QB-90 and QA), the only re-stimulation effect was onlyignificant for IQA and IQB90 formulation (Fig. 6). However, in earlyeports we have shown that splenocytes from mice vaccinated withB-90 saponins formulations vaccine adjuvanted with bovine her-esvirus 5 or poliovirus antigen respond with increased productionf Th1 cytokines upon in vitro re-stimulation compared to non-djuvanted vaccine [15,17]. Here our results show that ISCOMsreparations were more efficient than saponins formulations vac-ine in eliciting Th1 (IFN-� and IL-2) cytokines responses.

Another important issue evaluated here was the induction ofoth systemic and mucosal antibody responses as desirable fea-ure of the intranasal delivery of vaccines [7]. This route offershe advantages of being easier to administer and thus not requir-ng trained healthcare workers. Besides, it has the major potentialf inducing immunity at the portal of entry of many pathogense.g. mucosal sites) [21,39]. Therefore, immune responses from.n. delivered ISCOMs and saponin formulations interestingly differrom s.c. immunizations. The i.n. delivery of ISCOMs significantlyncreased OVA immunogenicity, as seen through the induction oferum specific IgG, IgG1 and IgA antibodies, but neither systemicgG2a responses or DTH reactions were observed. These results par-ially differed from those reported by Hu et al. [40], who showedhat s.c and i.n. immunizations of respiratory syncytial virus enve-ope protein in ISCOMs formulated with Q. saponaria saponinsnduced similar specific IgG2a levels in serum. As a consequence,t has been suggested that immunization route and antigen mod-late the type of immune response elicited by an adjuvant [6].

ndeed, in several reports it has been demonstrated that subcuta-eous administration of ISCOMs induces prominent Th1 responses,hile intranasal deliveries bias the response towards a Th2 profile

2,41]. Remarkably, i.n. immunization with IQB-90 induced localnasal) as well as distal (vaginal and faeces) mucosal production ofVA-specific IgA.

In summary, this is the first work that reports the formulationf ISCOMs with non-Q. saponaria saponins. QB-90 and IQB-90 areignificantly less toxic than Quil A®, both in vivo and in vitro. Wead shown that Ag uptake by BMDCs is largely more efficient when

t is formulated as IQB-90 than soluble saponins. Subcutaneouslyelivered IQB-90 induces strong humoral (Th1 and Th2-types) and

ellular immune responses. Equally important is the i.n. deliveryf IQB-90 approach presented in this work that provides new per-pectives in the significant specific antibody induction responsesoth at the systemic and mucosal levels. Overall, the properties of

[

4 (2016) 1162–1171

IQB-90 reported here demonstrate their potential as a candidatefor further development of prophylactic and therapeutic vaccines,as an alternative to the classic ISCOMs derived from Quil A®.

Conflict of interest statement

The authors declare that they have no conflict of interest.

Acknowledgements

This work was supported by the Programa de Desarrollo de lasCiencias Básicas (PEDECIBA) and Agencia Nacional de Innovacióne Investigación (ANII) from Uruguay and the Conselho Nacional deDesenvolvimento Científico e Tecnológico (CNPq), Financiadora deEstudos e Projetos (FINEP) and Coordenac ão de Aperfeic oamento dePessoal de Nível Superior (CAPES) from Brazil to a CAPES/UDELARproject.

References

[1] Rappuoli R. Twenty-first century vaccines. Philos Trans R Soc Lond B: Biol Sci2011;366:2756–8.

[2] Morelli AB, Becher D, Koernig S, Silva A, Drane D, Maraskovsky E. ISCOMA-TRIX: a novel adjuvant for use in prophylactic and therapeutic vaccines againstinfectious diseases. J Med Microbiol 2012;61:935–43.

[3] O’Hagan DT, De Gregorio E. The path to a successful vaccine adjuvant—‘the longand winding road’. Drug Discovery Today 2009;14:541–51.

[4] Brito LA, Malyala P, O’Hagan DT. Vaccine adjuvant formulations: a pharmaceu-tical perspective. Semin Immunol 2013;25:130–45.

[5] Harandi AM, Medaglini D, Shattock RJ. Vaccine adjuvants: a priority for vaccineresearch. Vaccine 2010;28:2363–6.

[6] Gupta P. Mucosal vaccine delivery and M cell targeting. In: Devarajan PV, JainS, editors. Targeted drug delivery: concepts and design. Springer InternationalPublishing; 2015. p. 313–37.

[7] Zaman M, Chandrudu S, Toth I. Strategies for intranasal delivery of vaccines.Drug Deliv Transl Res 2013;3:100–9.

[8] Garcon N, Hem SL, Friede M. Evolution of adjuvants across the centuries. Vac-cines 2013;5:58–70 (Sixth Edition).

[9] Rey-Ladino J, Ross AG, Cripps AW, McManus DP, Quinn R. Natural products andthe search for novel vaccine adjuvants. Vaccine 2011;29:6464–71.

10] Rosales-Mendoza S, Salazar-Gonzalez JA. Immunological aspects of usingplant cells as delivery vehicles for oral vaccines. Expert Rev Vaccines2014;13:737–49.

11] Kensil CR, Patel U, Lennick M, Marciani D. Separation and characterizationof saponins with adjuvant activity from Quillaja saponaria Molina cortex. JImmunol 1991;146:431–7.

12] Sun HX, Xie Y, Ye YP. Advances in saponin-based adjuvants. Vaccine2009;27:1787–96.

13] O’Hagan DT, Fox CB. New generation adjuvants—from empiricism to rationaldesign. Vaccine 2015;33(Suppl 2):B14–20.

14] Kauffmann C, Machado AM, Fleck JD, Provensi G, Pires VS, Guillaume D, et al.Constituents from leaves of Quillaja brasiliensis. Nat Prod Res 2004;18:153–7.

15] Silveira F, Cibulski SP, Varela AP, Marques JM, Chabalgoity A, de Costa F,et al. Quillaja brasiliensis saponins are less toxic than Quil A and have similarproperties when used as an adjuvant for a viral antigen preparation. Vaccine2011;29:9177–82.

16] Fleck JD, Kauffmann C, Spilki F, Lencina CL, Roehe PM, Gosmann G. Adjuvantactivity of Quillaja brasiliensis saponins on the immune responses to bovineherpesvirus type 1 in mice. Vaccine 2006;24:7129–34.

17] de Costa F, Yendo AC, Cibulski SP, Fleck JD, Roehe PM, Spilki FR, et al. Alternativeinactivated poliovirus vaccines adjuvanted with Quillaja brasiliensis or Quil-Asaponins are equally effective in inducing specific immune responses. PLoS ONE2014;9:e105374.

18] Villacres-Eriksson M, Behboudi S, Morgan AJ, Trinchieri G, Morein B.Immunomodulation by Quillaja saponaria adjuvant formulations: in vivo stim-ulation of interleukin 12 and its effects on the antibody response. Cytokine1997;9:73–82.

19] Lovgren Bengtsson K, Morein B, Osterhaus AD. ISCOM technology-based matrixM adjuvant: success in future vaccines relies on formulation. Expert Rev Vac-cines 2011;10:401–3.

20] Sun HX, Xie Y, Ye YP. ISCOMs and ISCOMATRIX. Vaccine 2009;27:4388–401.21] Sjolander S, Drane D, Davis R, Beezum L, Pearse M, Cox J. Intranasal immunisa-

tion with influenza-ISCOM induces strong mucosal as well as systemic antibodyand cytotoxic T-lymphocyte responses. Vaccine 2001;19:4072–80.

22] Zhao L, Seth A, Wibowo N, Zhao CX, Mitter N, Yu C, et al. Nanoparticle vaccines.

Vaccine 2014;32:327–37.

23] Quirici L, Verza SG, Mastrogiovanni M, Miraballes I, Casanova G, Soule S, et al.Candidate particulate antigen delivery system based on Quillaja brasiliensissaponins. In: Elsevier, editor. 7th Vaccine & ISV congress. Sitges, Spain: Elsevier;2013.

ccine 3

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

S.P. Cibulski et al. / Va

24] Silveira F, Rossi S, Fernandez C, Gosmann G, Schenkel E, Ferreira F. Alum-typeadjuvant effect of non-haemolytic saponins purified from Ilex and Passifloraspp. Phytother Res 2011;25:1783–8.

25] McKenzie A, Watt M, Gittleson C. ISCOMATRIX: vaccines: safety in humanclinical studies. Hum Vaccin 2010;30:6.

26] Sun Y, Liu J, Yu H, Gong C. Isolation and evaluation of immunological adjuvantactivities of saponins from the roots of Pulsatilla chinensis with less adversereactions. Int Immunopharmacol 2010;10:584–90.

27] Lutz MB, Kukutsch N, Ogilvie AL, Rossner S, Koch F, Romani N, et al. Anadvanced culture method for generating large quantities of highly puredendritic cells from mouse bone marrow. J Immunol Methods 1999;223:77–92.

28] West MA, Wallin RP, Matthews SP, Svensson HG, Zaru R, Ljunggren HG, et al.Enhanced dendritic cell antigen capture via toll-like receptor-induced actinremodeling. Science 2004;305:1153–7.

29] Dalsgaard K. A study of the isolation and characterization of the saponin QuilA.Evaluation of its adjuvant activity, with a special reference to the application inthe vaccination of cattle against foot-and-mouth disease. Acta Vet Scand Suppl1978;69:7–40.

30] Ragupathi G, Gardner JR, Livingston PO, Gin DY. Natural and syntheticsaponin adjuvant QS-21 for vaccines against cancer. Expert Rev Vaccines2011;10:463–70.

31] Mastelic B, Ahmed S, Egan WM, Del Giudice G, Golding H, Gust I, et al. Modeof action of adjuvants: implications for vaccine safety and design. Biologicals2010;38:594–601.

32] Brito LA, O’Hagan DT. Designing and building the next generation of improvedvaccine adjuvants. J Control Release 2014;190:563–79.

[

4 (2016) 1162–1171 1171

33] Gauthier C, Legault J, Girard-Lalancette K, Mshvildadze V, Pichette A.Haemolytic activity, cytotoxicity and membrane cell permeabilization of semi-synthetic and natural lupane- and oleanane-type saponins. Bioorg Med Chem2009;17:2002–8.

34] Myschik J, Lendemans DG, McBurney WT, Demana PH, Hook S, Rades T. Onthe preparation, microscopic investigation and application of ISCOMs. Micron2006;37:724–34.

35] Reimer JM, Karlsson KH, Lovgren-Bengtsson K, Magnusson SE, Fuentes A, Stert-man L. Matrix-M adjuvant induces local recruitment, activation and maturationof central immune cells in absence of antigen. PLoS ONE 2012;7:e41451,http://dx.doi.org/10.1371/journal.pone.0041451.

36] Maraskovsky E, Schnurr M, Wilson NS, Robson NC, Boyle J, Drane D. Develop-ment of prophylactic and therapeutic vaccines using the ISCOMATRIX adjuvant.Immunol Cell Biol 2009;87:371–6.

37] Sjolander A, van’t Land B, Lovgren Bengtsson K. Iscoms containing purified Quil-laja saponins upregulate both Th1-like and Th2-like immune responses. CellImmunol 1997;177:69–76.

38] Coffman RL, Sher A, Seder RA. Vaccine adjuvants: putting innate immunity towork. Immunity 2010;33:492–503.

39] Davitt CJ, Lavelle EC. Delivery strategies to enhance oral vaccination againstenteric infections. Adv Drug Deliv Rev 2015;91:52–69.

40] Hu KF, Elvander M, Merza M, Akerblom L, Brandenburg A, Morein B. The

immunostimulating complex (ISCOM) is an efficient mucosal delivery systemfor respiratory syncytial virus (RSV) envelope antigens inducing high local andsystemic antibody responses. Clin Exp Immunol 1998;113:235–43.

41] Sjolander A, Cox JC, Barr IG. ISCOMs: an adjuvant with multiple functions. JLeukoc Biol 1998;64:713–23.