Adjuvating the adjuvant: Facilitated delivery of an immunomodulatory oligonucleotide to TLR9 by a cationic antimicrobial peptide in dendritic cells

Page 1

Ao

MGa

b

c

a

ARR1AA

KACD

1

cwhttiptrKclgectc

0d

Vaccine 29 (2011) 426–436

Contents lists available at ScienceDirect

Vaccine

journa l homepage: www.e lsev ier .com/ locate /vacc ine

djuvating the adjuvant: Facilitated delivery of an immunomodulatoryligonucleotide to TLR9 by a cationic antimicrobial peptide in dendritic cells

ichael C. Aichingera, Michael Ginzlerb, Julian Weghuberc, Lars Zimmermannc, Karin Riedlb,erhard Schützc, Eszter Nagyb, Alexander von Gabainb, Rudolf Schweyena, Tamás Henicsa,∗

Department of Genetics, Max F. Perutz Laboratories, Dr. Bohrgasse 9, 1030 Vienna, AustriaINTERCELL AG, Campus Vienna Biocenter 3, 1030 Vienna, AustriaBiophysics Institute, Johannes Kepler University, Altenbergstrasse 69, 4040 Linz, Austria

r t i c l e i n f o

rticle history:eceived 11 February 2010eceived in revised form6 September 2010ccepted 2 November 2010

a b s t r a c t

IC31® is a novel bi-component vaccine adjuvant consisting of the peptide KLKL5KLK (KLK) and the TLR9agonist oligonucleotide d(IC)13 (ODN1a). While membrane-interacting properties of KLK and immuno-modulating capabilities of ODN1a have been characterized in detail, little is known of how these twomolecules function together and synergize in interacting with their primary target cells, dendritic cells

vailable online 18 November 2010

eywords:djuvantationic antimicrobial peptideendritic cell

(DCs). We have found that KLK-triggered aggregates entrapped ODN1a and these complexes readily asso-ciated with the DC cell surface. KLK stimulated the uptake and internalization of ODN1a via endocytosis,while the bulk of the peptide remained associated with the cell periphery. ODN1a co-localized with earlyand late endosomes as well as endoplasmic reticular structures. ODN1a co-localized with TLR9 positivecompartments following KLK mediated uptake. These features did not depend on the expression of TLR-9.Our results reveal novel mechanisms that allow KLK to enhance the effects of the TLR-9 ligand ODN1a inimmunomodulation.

. Introduction

Current vaccine design exploits both careful selection pro-edures of protective antigens as well as formulations withell-tailored novel adjuvants. Recent developments in the fieldighlight the necessity to supplement vaccines with adjuvantshat are capable of immune potentiating the effects of evenhe best antigens [1]. The bi-component novel adjuvant, IC31®

s a stoichiometrically fix mixture of the natural anti-microbialeptide-derivative KLKL5KLK (herein referred to as KLK) andhe single-stranded DNA-phosphodiesther oligo-d(IC)13 (hereineferred to as ODN1a) [2,3]. Fritz et al. have demonstrated thatLK itself induces sustained adaptive immune responses againsto-injected antigens in vivo by stimulating the expression of highevels of Th2 type antigen-specific antibodies to the model anti-en ovalbumin (OVA) or a commercial influenza vaccine [4]. KLK is

fficient in enhancing association of antigen to antigen-presentingells (APCs, including dendritic cells, DCs) and in forming charac-eristic vaccine depots at the site of injection [4]. Co-injection ofomplete IC31® with appropriate immunogenic peptides or pro-

∗ Corresponding author. Tel.: +43 1 4277 54614; fax: +43 1 4277 9546.E-mail address: [email protected] (T. Henics).

264-410X/$ – see front matter © 2010 Elsevier Ltd. All rights reserved.oi:10.1016/j.vaccine.2010.11.003

© 2010 Elsevier Ltd. All rights reserved.

teins induces a potent antigen-specific, predominantly Th1 type Tcell response and elicits antigen-specific cytotoxic T cell activity[5]. The protein-based vaccines are capable of inducing antigen-specific mixed type 1 and type 2 humoral responses. It has also beendemonstrated that IC31® efficiently stimulates the maturation andactivation of dendritic cells (DCs) that, in turn, enables the inductionand proliferation of naïve T cells [5]. The TLR-9/MyD88 pathwaywas identified as the efferent arm of IC31® action, as immuniza-tion with IC31®-adjuvanted OVA peptides results in the dramaticincrease of peptide-specific IFN-�-producing T cells derived fromwild type, but not from TLR-9 −/− or MyD88 −/− mice [5]. Furtherexperiments revealed that the originally described immunomod-ulatory effects of KLK are overruled by the combined effects seentogether with ODN1a (IC31® effects) and also demonstrated thatKLK considerably enhances the effects elicited by the oligonu-cleotide alone [2,5].

Recently, we have demonstrated that, while KLK appears asa non-pore forming peptide, it readily interacts with variousmembrane vesicles or cells, profoundly altering membrane ultra-structure and fluidity [6]. Depending on the composition of a

membrane-mimicking environment, KLK is capable of undergoingmultiple conformational transitions [6]. These data together withthe assumption that KLK and ODN1a may associate in a charge andconformation based complex, prompted us to make efforts (i) tocharacterize the fate of KLK and ODN1a upon their interaction with

Page 2

Vaccin

dKdfir

afdaetuowTalci

2

2

1hAhFBRapwof

2

CwAw(1pG(Fi

2

iSm

2h

h

M.C. Aichinger et al. /

endritic cells, (ii) to evaluate how membrane-active properties ofLK affects association and internalization of ODN1a by DCs, (iii) toetermine which sub-cellular compartments attracts ODN1a and,nally, (iv) to demonstrate that ODN1a is targeted to its cellulareceptor, TLR-9.

Here we show that KLK-facilitated aggregates entrap ODN1and the resulting complexes readily associate with the DC cell sur-aces. KLK stimulates the uptake and internalization of ODN1a byendritic cells via endocytosis, while the peptide remains associ-ted with the periphery. ODN1a co-localizes with early and latendosomes as well as endoplasmic reticular structures. Associa-ion of KLK–ODN1a complex with the cell surface, as well as ODN1aptake, internalization and cytoplasmic distribution do not dependn the expression of TLR-9. However, ODN1a partially co-localizesith TLR9 positive compartments following KLK mediated uptake.

hese data reveal novel mechanisms that identify KLK as a potentdjuvant component in enhancing beneficial effects of the TLR-9igand ODN1a in immunomodulation, thus mark IC31® as a vac-ine adjuvant already in human trials whose uptake mechanismnto DCs has been characterized in details.

. Materials and methods

.1. Materials

KLK and KPK (10 mg/ml) were from Intercell AG, Austria;000 nmol/ml ODN1a-Cy5 was from Purimex, Germany; Anti-uman/mouse TLR9-FITC was obtained from Eubio, Alexis;nti-human CD71 (TfR)-FITC was from BD Pharmingen; Anti-uman CD107a (Lamp-1)-PE was from BD Pharmingen; FM1-43,M® Lipophilic Styryl Dyes, ER-TrackerTM Green and glibenclamideODIPY® FL were purchased from Invitrogen; DAPI was from Roche.PMI 1640, l-glutamine, sodium-pyruvate, non-essential aminocids, gentamycin, �-mercaptoethanol, fetal bovine serum andenicillin/streptomycin were from PAA-Laboratories (Austria). IL-4as purchased from Peprotech, Eubio, GM-CSF was from Cell-Genix

r derived from X6310 cell supernatants [7]. DMEM-Glutamax wasrom Gibco, Invitrogen.

.2. Preparation of human immature dendritic cells

Whole human heparinized blood was obtained from the Redross and 500 ml of blood was diluted 1:1 in PBS. Diluted bloodas placed on LSM 1077 lymphocyte separation medium (PAA,ustria) gradient and centrifuged for 20 min at 350 × g. Monocytesere harvested and CD14+ cells were selected by MACS purification

MiltenyiBiotech) and incubated in RPMI 1640 supplemented with0% Fetal bovine serum (PAA, Austria), 1% l-glutamine, 1% sodium-yruvate, 0.1% �-mercaptoethanol, 500 U/ml IL-4 and 0.1 �g/mlM-CSF at 37 ◦C in 5% CO2 for 7 days. At day 6 of culture, CD83

below 10%), CD1a (50–70%) and MHCII (70–90%) were assured byACS analysis. At day 7 non-adherent cells were collected and usedn subsequent experiments.

.3. Preparation of mouse bone marrow derived dendritic cells

C57Bl/6 were housed under pathogen free condition accord-ng to guidelines of the Federation of European Laboratory Animalcience Association. Bone marrow-derived DCs were isolated andaintained according to Reutterer et al. [7].

.4. Phenotypic characterization of KLK or IC31®-stimulateduman myeloid DC

Immature human myeloid DC were generated by culture ofighly purified monocytes in X-VIVO-15 medium (Bio Whit-

e 29 (2011) 426–436 427

taker) supplemented with 1.5% human plasma, 800 U/ml GM-CSF(Leukomax®, Novartis) and 1000 U/ml IL-4 (Strathmann Biotech).Cells were fed at day 2 of culture with medium supplementedwith 1600 U/ml GM-CSF and harvested at day 5 of culture. Imma-ture human myeloid DC were stimulated for 24 h with mediumor KLK (10 nmol/ml) at 1.5 × 106 cells/3 ml in 6-well flat-bottomplates. After incubation, cells were harvested and stained withanti-HLA-DR-APC (L243, BD Biosciences), anti-CD83-FITC (HB15e,BD Biosciences), anti-CD40-FITC (5C3, BD Biosciences), anti-CD80-FITC (L307.4, BD Biosciences) and anti-CD86-FITC (2331 FUN-1,BD Biosciences), using adequate isotype control antibodies. Phe-notypic analyses were performed using a FACSCalibur® flowcytometer equipped with CellQuest software (Becton Dickin-son).

2.5. ODN1a-Cy5 uptake experiments

Dendritic cells were placed in an 8 well Lab-Tek II Cham-ber Slide (LAB-TEK®, Nunc) at a density of 1.5–3 × 105 cells perwell and incubated over night at 37 ◦C in 5% CO2 in appropriatemedium. Fresh medium was added to the wells with ODN1a-Cy5 and KLK/KPK added to a final concentration of 0.4 nmol/ml(low dose) or 1.2 nmol/ml (high dose) and 10 nmol/ml, respec-tively. Where indicated 6 nM FM1-43 or 1 �M ER-tracker wasadded to the mixture. Cells were incubated for various lengthsof time, washed once with PBS and then fixed for 30 min at37 ◦C with 4% PFA (Sigma–Aldrich). Cells were washed twice with0.5% Tween in PBS for 1 min, rinsed once with PBS and thenincubated with 2% BSA (Sigma–Aldrich) or 2% goat serum (Vec-tor Laboratories). In case of subsequent antibody staining cellswere exposed to anti-human CD71-FITC and anti-human CD107a-FITC diluted 1:33 in 2% BSA, or to anti-human/mouse TLR9-FITCdiluted 1:100 in 2% goat serum for 1 h at 37 ◦C in a humid cham-ber. Cells were then washed two times with 2% BSA or 2% goatserum and once with PBS and examined using Zeiss Axioplan 2LSM510 Meta laser scanning fluorescence microscope. Images weresubsequently analyzed using the Zeiss LSM ImageBrowser soft-ware.

2.6. Live cell laser confocal imaging

At day 7 dendritic cells were collected and transferred tonew dishes containing sterile glass cover-slides. Cells were incu-bated at 37 ◦C in 5% CO2 over night. ODN1a-Cy5, KLK/KPK anddyes were mixed as described above. The medium was removedcarefully from the cover-slide and a few micro-liters of themixture were placed on the slide. The glass slide was put onan object slide and sealed before the sample was observedusing a Zeiss Axioplan 2 laser scanning fluorescence micro-scope equipped with a LSM510 Meta camera. Images wereprocessed using the Laser Scanning Microscope Version 3.2 SP pro-gram.

2.7. FRET measurements

FRET experiments were performed using FRET-pair formingfluorochrome-labeled components; KLK–TAMRA and ODN1a-Cy5. FRET efficiency was calculated with the acceptor-bleachingtechnique (acceptor: Cy5; Donor: Tamra) according to theformula FRETeff = (Intensitypost − Intensitypre)/Intensitypost withphoto-bleaching corrections. For fluorescent imaging in FRET

experiments, a modified epi-fluorescence microscope was used(Axiovert 200, Zeiss, Germany), which was equipped with atemperature control system (POCmini, Zeiss, Germany). Sampleswere illuminated through a 100× NA = 1.45 Plan-Apochromatobjective (Zeiss) using the 514 nm line (excitation of TAMRA)

Page 3

428 M.C. Aichinger et al. / Vaccine 29 (2011) 426–436

Fig. 1. KLK enhances the uptake of ODN1a in DCs. (A) Human moDCs were exposed to Cy5-labeled ODN1a in the absence (control) or presence of KLK or its derivative,KPK. Red coloration of the cytoplasm indicates uptake of ODN1a. (B) At early phases Cy5-ODN1a and KLK aggregates in various size complexes (upper left panel) and thesea esicur ntainii to the

otumoaaM2w3

ggregates associate with DCs (upper right panel). ODN1a internalizes into distinct vight panel). (C) Time-lapse laser confocal imaging of a DC that engulfs ODN1a conterpretation of the references to color in this figure legend, the reader is referred

f an Ar+-ion laser (Innova, Coherent, USA) and 647 nm (exci-ation of Cy5) of a Kr+-ion laser at excitation intensities ofp to 10 kW/cm2 by using epi-configuration. An acousto-opticodulator (Isomet, 1205C) was used to achieve exact timing

f the laser illumination. After filtering (custom-made dichroicnd emission filters, Chroma, USA), images were recorded on

back-illuminated liquid nitrogen cooled CCD camera (Microax 1300-PB, Roper Scientific). Each CCD pixel corresponds to

00 nm × 200 nm in the sample plane. All live cell experimentsere performed in Hank’s balanced salt solution (HBSS; PAA) at

7 ◦C.

lar structures (lower left panel) and later on dissipates within the cytoplasm (lowerng particles (white arrows and arrowheads). Time scale is from 0 to 60 min. (Forweb version of the article.)

3. Results

3.1. KLK stimulates the uptake and internalization of ODN1a bydendritic cells

Initially we observed that KLK dramatically increased the uptake

and cytoplasmic accumulation of Cy5-labeled ODN1a in bothmouse and human dendritic cells (Fig. 1A). This was DC specificas no uptake and accumulation was seen with HeLa, C2C12, CACOor HEK cells (data not shown). Interestingly, the middle leucine-to-proline substituted derivative of KLK (KPK) completely failed to

Page 4

M.C. Aichinger et al. / Vaccine 29 (2011) 426–436 429

F KLK-FA 1a. (C)9

saoa1deopctatws

3c

vpfltct

ig. 2. KLK and ODN1a dissociates at the cell periphery. (A and B) ODN1a-Cy5 andrrows indicate ODN1a–KLK complexes and arrowheads point to internalized ODN0 min.

timulate the uptake of ODN1a (Fig. 1A). Furthermore, when KLKnd ODN1a were added to cells, we observed immediate formationf heterogeneous size aggregates around the cells that promptlyssociated with the cell surface (Fig. 1B, upper left and right panels).5 min after the administration of the adjuvant compounds, ODN1aistributed in various size vesicle-like structures within periph-ral regions of the cytoplasm, and these structures were dissipatingver time, but were clearly visible after 60 min (Fig. 1B, lower leftanel). At later time points the internalized and peripheral ODN1aontaining vesicles gradually redistributed homogeneously overhe whole cytoplasm of the cells (Fig. 1B, lower right panel). Noggregate formation was evident when KPK was used in iden-ical experiments, and importantly, in this context no ODN1aas detected in any of the cells at any time points (data not

hown).

.2. KLK-based aggregates entrap ODN1a and associate with theell surface

KLK and ODN1a mixture (as IC31®) readily assembled intoarious size aggregates around dendritic cells (DCs) and the com-

lex showed strong association with the cell surface. Time-lapseuorescence confocal imaging revealed a dynamic interaction ofhe aggregates with the cell body and large dendrites of theells (Fig. 1C). Fluorochrome labeling of the peptide (FITC) andhe oligonucleotide (Cy5) demonstrated co-localization of these

ITC were added to DCs and the complex was monitored by laser confocal imaging.DCs were exposed to KLK-FITC and monitored by laser confocal imaging from 0 to

molecules within the aggregates, regardless of whether they asso-ciated with cells or not (Fig. 2A). Although, charge-mediatedassociation of the oppositely charged molecules appeared a fea-sible mechanism to explain aggregate formation, complete lack ofaggregation with KPK (this peptide has no significant net chargedifference as compared to KLK) indicated, that distinct secondarystructural properties were rather responsible for the aggregativeproperty of KLK. Indeed, we have earlier demonstrated that, whileboth KLK and KPK was random coil in dilute aqueous solutions,only KLK, but not KPK arranged into a �-sheeted conformationas determined by circular dichroism measurements [6]. There-fore, we propose that KLK has the capacity to assemble intointer-molecular �-sheeted aggregates that entraps ODN1a and thecomplex associates with negatively charged moieties of the DC cellsurface.

The association/co-localization of KLK and ODN1a was lost assoon as ODN1a had been internalized (Fig. 2B). This suggested, thatKLK remained associated with the cell membrane while ODN1aentered the cytoplasm, best explained by an endocytotic process.In accordance with this, we observed no obvious internalization ofFITC-labeled KLK in human dendritic cells. The peptide stayed asso-

ciated with cortical areas of the cells over time with gradual loss offluorescent signal at later time points (Fig. 2C).

To further prove contact loss between the adjuvant compo-nents upon internalization of ODN1a, we performed fluorescenceresonance energy transfer (FRET) experiments using fluorochrome-

Page 5

430 M.C. Aichinger et al. / Vaccin

Fig. 3. FRET analysis of human Jurkat T cells (A). Jurkat T cells were exposed toODN1a-Cy5 and KLK–TAMRA and analyzed for FRET as described in Section 2. (1)ODN1a-Cy5 and KLK–TAMRA on glass slide. (2) Cy5 and KLK–TAMRA on glass slide.(aat

lmmrbomtdlttavocNripateacs

oligonucleotide were monitored by fluorescent confocal imaging.

3) Extracellular FRET signal average efficiency. Fluorescent images of a Jurkat T cellnd a mouse bone marrow derived dendritic cell channeled to detect Cy5 (ODN1a)nd TAMRA (KLK) emission signals (B). The contour of the mouse DC is outlined inhe figure for better orientation.

abeled components where the flourochromes meet the require-ents as optimal FRET pairs (KLK–TAMRA and ODN1a-Cy5). Thisicroscopy technique used to measure the proximity of two fluo-

ophores. Since fluorophores can be employed to specifically labeliomolecules and the distance condition for FRET is of the orderf the diameter of most biomolecules, FRET is often used to deter-ine when and where two or more biomolecules interact within

heir physiological surroundings. Since energy transfer occurs overistances of 1–10 nm, a FRET signal corresponding to a particular

ocation within a microscope image provides an additional dis-ance accuracy surpassing the optical resolution (∼0.25 mm) ofhe light microscope. The average FRET value between ODN1a-Cy5nd KLK–TAMRA on a glass slide was 8.8% (Fig. 3A), but the FRETalue was concentration dependent. At a concentration ratio (cr)f ODN1a-Cy5: KLK–TAMRA [=intensity (red)/intensity (green)] ofr >10 the FRET value was 10.8%, at cr <10 6.9% (data not shown).o FRET was observed between KLK–TAMRA and Cy5 only (fluo-

ochrome without the nucleotide), suggesting that the FRET shownn Fig. 3A was not due to the high concentrations of the labeled com-onents (Fig. 3A). Next we measured FRET between ODN1a-Cy5nd KLK–TAMRA in the presence of Jurkat T cells. After incuba-ion of the cells with both substances, FRET was calculated in the

xtracellular space and within the cell. In the extracellular space theverage FRET value of 7% was similar to the measurements withoutells (6.9% for cr < 10) (Fig. 3A). Since no intracellular FRET inten-ity was measured, we attempted to obtain fluorescent images of

e 29 (2011) 426–436

Jurkat T cells exposed to ODN1a-Cy5 and KLK–TAMRA. As shownin Fig. 3B, no intracellular signal for ODN1a was registered, whileKLK was readily detectable outside the cell. This data was consis-tent with our observation that no KLK-stimulated uptake of ODN1awas observed with many non-DC cells (see Section 4). Additionally,this result explained the lack of intracellular FRET signal in thesecells. Despite of multiple attempts, we were not able to registerFRET using dendritic cells, possibly due to the highly detailed cellsurface property of DCs, but fluorescence imaging of these cellsclearly revealed ODN1a localization within the cell and – simi-larly to Jurkat T cells – KLK remained outside (Fig. 3B). Togetherthese data favour the assumption that ODN1a is internalized byDCs, while KLK remains localized at the cell periphery, possibly inassociation with the plasma membrane.

3.3. Association of ODN1a with endosomal and endoplasmicreticular compartments

In order to identify the internalization route and mecha-nism by which KLK facilitates the uptake of ODN1a, we usedfluorochrome-labeled specific markers of various endosomal andER compartments, which are known cytoplasmic locations of TLR9.First, DCs were exposed to Cy5-labeled ODN1a in the presence ofKLK and co-stained with the endosomal marker, FM1-43. Theseexperiments revealed clear co-localization of ODN1a with endo-somes (Fig. 4A). The figure demonstrates the sustained nature ofuptake and internalization, as ODN1a containing cell-associatedaggregates were still persistent, while endosomal localization of theoligonucleotide was already apparent (Fig. 4A). Consistent with ear-lier observations [7] Cy5-labeled CpG DNA was also endosomallylocalized in the presence of KLK (Fig. 4B).

In similar experiments, ODN1a was co-localized with the endo-plasmic reticular compartment of DCs, indicating that besidesendosomal internalization a portion of the oligonucleotide alsolocates in the ER (Fig. 5A). Also, CpG DNA was targeted to the ERin response to KLK treatment (Fig. 5B). Thus, the mode of action ofKLK does not seem to be restricted to a particular ODN type. Time-lapse confocal fluorescence imaging identified ODN1a containingaggregates associated with the cell body and red-to-yellow colortransition that indicates internalization and co-localization withthe ER (Fig. 5C).

3.4. Co-localization of ODN1a with early- and late endosomesand endoplasmic reticulum in wild type and TLR9 −/− cells

Considerable work exist demonstrating that CpG DNA is inter-nalized and associates with early endosomes and this mechanismtriggers the translocation of TLR9 from the ER to the so-calledtubular lysosomal compartment (TLC) of DCs [8,9]. According tothese findings, the additional recruitment of adapter molecules,such as MyD88 and subsequent initiation of TLR9 signaling wouldtake place in the TLC. Thus, we first asked if ODN1a and CpG DNAcould co-localize with early and late endosomes in our system.These experiments revealed that both ODN1a and CpG DNA co-localized with TfR and Lamp1-positive structures, indicating thatboth TLR9 agonists partitioned in early and late endosomes of DCs(Fig. 6A and B). Next we were interested to learn whether TLR9was required for KLK-induced uptake and cellular distribution ofODN1a. DCs obtained from TLR9 −/− mice were exposed to labeledODN1a and KLK and subsequent uptake and localization of the

Our results showed that the internalization and cellular distribu-tion of ODN1a was identical to that of WT cells, thus the presence ofTLR9 did not affect KLK-mediated uptake and cellular distributionof ODN1a (Fig. 7).

Page 6

M.C. Aichinger et al. / Vaccine 29 (2011) 426–436 431

F a-Cy5b DNA

3Tm

ooTaWutvse

atsmmsowepsambo

3O

ue

Recent identification of the unique membrane interactive prop-erties of KLK [6] prompted us to further investigate the potentialeffects of this peptide in facilitating the uptake and subsequent

Table 1KLK-and IC32-mediated stimulation of human myeloid DC.

Medium KLK IC31

HLA-DR/CD83Donor 1 0.84%a 4.71% 14.61%Donor 2 0.77% 3.87% 13.02%

HLA-DR/CD1aDonor 1 1.34% 6.94% 21.30%Donor 2 0.90% 9.70% 25.00%

HLA-DR/CD40Donor 1 2.58% 11.07% 24.43%Donor 2 – – –

HLA-DR/CD80Donor 1 1.42% 4.91% 13.52%

ig. 4. ODN1a co-localizes with endosomes. Human moDCs were exposed to ODN1y laser confocal imaging. (A) Cells exposed to ODN1a-Cy5. (B) Cells exposed to CpG

.5. Internalized ODN1a partially co-localizes withLR9-containing compartments in mouse mDCs and humanoDCs

Next, it was important to examine whether the up regulationf ODN1a internalization by KLK was accompanied with bindingf the oligonucleotide to its pattern recognition receptor, TLR9.o monitor this, we stained IC31®-treated cells with the recentlyvailable antibody against TLR9 (Alexis Biochemicals, Germany).e monitored co-localization of Cy5-ODN1a and anti-TLR9-FITC

sing laser-confocal fluorescence imaging. As shown in Fig. 8, par-ial co-localization of ODN1a with TLR9 was detected in distinctesicle-like compartments (Fig. 8, white arrows). Thus, we demon-trated receptor-ligand binding in these cells that also providesvidence of TLR9 positivity of mouse bone marrow derived DCs.

With the demonstration of KLK mediated delivery of its lig-nd, ODN1a to TLR9 positive compartments, we were interestedo see if this mechanism was associated with functional aspects,uch as cytokine production and/or expression of costimulatoryolecules and activation/maturation markers by DCs. Despiteultiple attempts to detect IL-6, IL-12, TNF� and IFN� in the

upernatants of KLK/KPK or IC31® treated human and mouse DCs,ther than minimal levels of IFN�, none of the other cytokinesere apparent in these samples. Moreover, we did not observe any

ffects of KLK with or without ODN1a on the basal level of IFN�roduction (data not shown). However, we were able to demon-trate KLK-mediated upregulation of costimulatory molecules andctivation markers (CD83, CD1a, CD40, CD80 and CD86) on humanyeloid DCs (Table 1). Upregulation of these cell surface molecules

y KLK was much more prominent in the presence of ODN1a, thether component of IC31 (Table 1).

.6. KLK drives simultaneous uptake of a peptide antigen and

DNa1 in dendritic cells

It remained tempting to explore how KLK mediated ODN1aptake affected the delivery of a specific antigen into DCs. Fritzt al. have shown that KLK was capable of enhancing antigen deliv-

and KLK, stained with the endosomal marker FM1-43 and the cells were monitored-Cy5.

ery to APCs in a dose dependent manner [4]. However, no attemptshave been made to monitor localization of ODN1a and the antigenin the same cell. In order to perform this, we co-administered theadjuvant (KLK–ODN1a-Cy5) and a peptide antigen (OVA-SINFEKLpeptide-FITC) and monitored cellular distribution of ODN1a usingfluorescence confocal imaging. We found no detectable uptake ofantigens in the absence of KLK (Fig. 9, upper left panel) but consid-erable OVA was accumulated in cells in the presence of KLK (Fig. 9,upper right panel). We found that ODN1a and the antigen mostlyco-localized in the same extracellular aggregate (Fig. 9, lower pan-els). However, the oligonucleotide and the model antigen seemedto be internalized into distinct vesicular compartments with onlynegligible co-localization (Fig. 9, lower panels).

4. Discussion

Donor 2 1.12% 5.13% 16.77%HLA-DR/CD86

Donor 1 54.66% 60.24% 65.85%Donor 2 71.91% 80.50% 75.46%

a Percentage of double positive cells.

Page 7

432 M.C. Aichinger et al. / Vaccine 29 (2011) 426–436

F 5 andi e-lap( cell w

ctpl

ig. 5. ODN1a co-localizes with the ER. Human moDCs were exposed to ODN1a-Cymaging. (A) Cells exposed to ODN1a-Cy5. (B) Cells exposed to CpG DNA-Cy5. (C) Timarrowhead) and internalizes ODN1a. Arrow indicates the cytoplasmic region of the

ytoplasmic delivery of ODN1a to its intracellular target recep-or. This seemed to be especially tempting in the light of theotent additive action of the two molecules in immunomodu-

ation [10–12]. Our experiments revealed that KLK and ODN1a

KLK, stained with ER-tracker (ER) and the cells were monitored by laser confocalse fluorescence confocal imaging of a DC that encounters an ODN1a–KLK aggregatehere ODN1a co-localizes with the ER compartment.

co-aggregate in complexes that readily associate with the dendriticcell surface. This was attributable to KLK, as no such aggregateformation was observed in the absence of the peptide. Further-more, no aggregate formation was seen when KLK was replaced

Page 8

M.C. Aichinger et al. / Vaccine 29 (2011) 426–436 433

F posede ed byD

wsmiaostsTDai

ig. 6. ODN1a co-localizes with early and late endosomes. Human moDCs were exarly endosome, EE) or Lamp1 (for late endosome, LE) and the cells were monitorNA-Cy5.

ith KPK, a derivative peptide with a middle lysine-to-prolineubstitution. Since no change in the net charge profile was imple-ented by the single amino acid substitution, we assumed that

t was the conformational properties of KLK that mediated itsggregative potency. In agreement with our earlier demonstrationf multiple conformational transitions of KLK [6], we hypothe-ize that the peptide accumulates in a colloid-like complex wherehe solvent-proximal random coil molecules are driven into �-

heeted aggregates towards the hydrophobic centre of the complex.hese KLK-based aggregates entrap ODN1a and antigen presentingCs then engulf the complexes. However, the peptide-nucleic acidggregates dissolve readily upon internalization, as only ODN1as detected in early and late endosomal compartments, following

to ODN1a-Cy5 and KLK, stained with antibodies against transferring receptor (forlaser confocal imaging. (A) Cells exposed to ODN1a-Cy5. (B) Cells exposed to CpG

uptake. Although, a characteristic depot formation is observed aftersubcutaneous injection of ODN1a in the presence of KLK, but notKPK, [2] we have no evidence to indicate that the above mentionedaggregation and uptake mechanisms also apparent in vivo in theinterstitial environment. We made attempts to monitor the fate ofKLK during this process and showed that – at least – the vast major-ity of the peptide remains associated with the cell periphery. Takentogether with our earlier findings indicating that, triggered by the

lipid composition of the membrane-mimicking environment, KLKis capable of acquiring an �-helical conformation [6], we assumethat this conformation is induced by distinct loci of the cell mem-brane. Subsequently, this contact leads to the intercalation of KLKwith the negatively charged inner leaflet, with the simultaneous

Page 9

434 M.C. Aichinger et al. / Vaccine 29 (2011) 426–436

F se bod th thec ates a

dcscwwF

oa(bd(tomo

eh

FcO

ig. 7. TLR9 is not required for ODN1a uptake by DCs. Wild type and TLR9 −/− mouescribed in Section 2. After 60 min of exposure, cells were fixed and co-stained wionfocal imaging. Arrows indicate cell surface-associated ODN1a containing aggreg

isruption of the aggregates and internalization of the oligonu-leotide into distinct vesicular structures. Unique cell surface loci,uch as characteristic lipid rafts at the DC cell surface might explainell specificity. Indeed, none of these phenomena were observedith HeLa, C2C12, CACO, Jurkat T cells or HEK cells regardless ofhether ODN1a was added in the presence or absence of KLK (see

ig. 3 and data not shown).We have also demonstrated that, in accordance with earlier

bservations [4], KLK mediated increased uptake of a model peptidentigen into DCs did not interfere with increased uptake of ODN1aFig. 9). The aggregation and entrapping mechanism by KLK mighte part of the initiation of Ag uptake and likely not charge depen-ent as no such phenomena were observed in the presence of KPKnot shown). We did not observe competition in the internaliza-ion between ODN1a and the tested OVA peptide but the majorityf ODN1a and Ag seemed to localize in distinct vesicular compart-

ents (Fig. 9). However, interference with the KLK mediated uptake

f ODNa1 may take place with other antigens.We observed that ODN1a co-localized with early and late-

ndosomal compartments of DCs. CpG DNA, another TLR9 agonistas been suggested to internalize via the clathrin-mediated

ig. 8. ODN1a co-localizes with TLR9-positive structures. Mouse bone-marrow derived Dells were visualized by laser confocal fluorescence imaging. On the right panel the contDN1a double positive structures, large arrows point to ODN1a positive vesicles.

ne marrow-derived dendritic cells were exposed to Cy5-labeled ODN1a and KLK asendosomal marker FM1-43 and DAPI. Stained cells were visualized by fluorescentnd arrowheads point to internalized particles.

endocytotic pathway [8]. In our system, we saw a basal butobvious CpG uptake in the absence of KLK, and KLK consider-ably enhanced CpG uptake by DCs. ODN1a, on the other hand,is not visibly taken up by DCs in the absence of KLK, butits uptake is dramatically enhanced by KLK. Although, severalendocytotic pathways have been identified, far the best under-stood uptake system is the clathrin-dependent process. Muchless defined are the various clathrin-independent internaliza-tion routes, yet, these mechanisms can account for as much as50% of the total endocytotic activity of the cell [13]. Besidesmacropinocytosis and caveolae-dependent uptake, the clathrin-and caveolae-independent internalization (most widely definedas GPI-enriched endosomal compartments or GEEC-pathway) hasbeen receiving considerable attention because of a number ofunique features, including its responsibility of the major fraction offluid-phase uptake of the cell [13–15]. Although, numerous details

are yet to be identified, it becomes increasingly clear that the endo-cytotic vesicles derived from various internalization pathways mayeventually converge at the level of early and late endosomal com-partments [13,16]. These endosomal structures then regulated bythe same recycling machinery [13].

Cs were exposed to ODN1a-Cy5 and KLK, stained with antibodies against TLR9 andours of a DC are marked with a white dotted line. Small arrows indicate TLR9 and

Page 10

M.C. Aichinger et al. / Vaccine 29 (2011) 426–436 435

F expoK rrows

pomhGpmtoOpoDOplIt[pursIicobe

ig. 9. Fluorescent confocal imaging of mouse bone marrow derived dendritic cellsLK (upper right panel) or in the presence of KLK and Cy5-ODN1a (lower panels). A

Recent studies revealed that KLK specifically blocks the fluid-hase uptake/GEEC-linked internalization pathway, while nobvious effects of KLK was detected on clathrin or caveolin-ediated endocytotic mechanisms. This effect of KLK leads to a

alt in recycling of GPI-APs and to a consequent accumulation ofEEC vesicles (or clathrin-independent carriers, CLICs) at the celleriphery [Weghuber et al., in preparation]. Although, the exactechanism of ODN1a internalization remains elusive, it is feasible

o assume that the oligonucleotide is internalized via more thanne uptake mechanisms; therefore, KLK-mediated stimulation ofDN1a uptake might be a compensatory shift from the GEEC-linkedathway towards clathrin-mediated or other (caveolin-dependentr independent) pathways [13,15]. Alternatively, similarly to CpGNA, clathrin-mediated uptake might be the primary route ofDN1a internalization, which would be flooded upon the fluid-hase recycling block by KLK, resulting thus in a very efficient

oading of the DC cytoplasm with the immunostimulatory ODN1a.n any case, ODN1a co-localizes with endosomal compartmentshat are surveyed by TLR9, its pattern recognition receptor (PRR)17]. Importantly, we were able to co-localize ODN1a with TLR9ositive vesicles in human DCs. This argues, that KLK mediatedptake and internalization of this TLR9 ligand leads to a detectableeceptor-ligand binding, a prerequisite of subsequent biologicalignaling in this important adaptive immunological mechanism.nterestingly, we also co-localized ODN1a with compartments pos-

tive for the selective ER marker, ER-TrackerTM. This seeminglyontradictory finding gains explanation by recent demonstrationf retrograde recycling mechanisms that allow communicationetween various endosomal compartments and the ER [19]. Forxample, ER-phagosome fusions [19,20], or ER fusion with sites of

sed to the FITC-labeled model antigen OVA alone (upper left panel), together withindicate extracellular aggregates containing OVA and ODN1a.

microbial entry [8] have supported localization of various proteinsin unexpected compartments.

The ability if IC31® to promote T or B cell-mediated adaptiveimmunity is lost in TLR9-deficient mice, suggesting that TLR9 sig-naling and subsequent IFN-I responses might be crucial in theimmunostimulatory function of IC31®. Indeed, it has recently beenshown that IC31® (KLK + ODN1a) was ineffective to induce peptideantigen-specific CTL response in IFN-I receptor and Stat1 defi-cient mice [21]. The authors also conclude that DCs rather thanT cells were the primary IFN-I targets in these experiments sincemice lacking the IFN-I receptor in the T cell lineage sowed nodetectable loss of IC31® effects [21]. These data suggest that IC31®

induces IFN-I production by DCs to improve T cell-mediated adap-tive immunity. Although, we were not able to demonstrate KLKand/or IC31® mediated cytokine production in our experimentalsystem, we detected KLK-mediated upregulation of costimulatorymolecules and activation markers (CD83, CD1a, CD40, CD80 andCD86) on human myeloid DCs. Upregulation of these cell surfacemolecules by KLK was much more prominent in the presence ofODN1a, the other component of IC31® (Table 1).

In conclusion, we propose a mechanism during which KLK-mediated specific block of the GEEC-internalization pathway shiftsuptake of ODN1a towards clathrin-mediated or other alternativepathways and this upregulated entry of ODN1a into DCs promptsrelocation of TLR9 from the ER to late endosomal and tubular

lysosomal compartments, as suggested earlier [8]. Redistributionof GPI-APs from the fluid-phase uptake to the clathrin-mediatedendocytic pathway upon inhibition of the Rho-family GTPase cdc42has been described [14,22]. Thus, channeling of ODN1a internaliza-tion provides a hyper-efficient platform of TLR9 ligand recognition

Page 11

4 Vaccin

ararahtuopbta

A

cf

R

[

[

[

[

[

[

[[

[

[

36 M.C. Aichinger et al. /

nd subsequent initiation if signaling and might explain the supe-ior immunostimulatory capacity of IC31® when compared to otherdjuvants [2]. This efficient loading mechanism may also allow toeducing the amount of the TLR agonist to an acceptable level thatvoids the onset of inflammatory side effects. In conclusion, weave identified a CAP-mediated efficient delivery of a TLR9 agonisto its innate PRR, a phenomenon that relies on the earlier describednique membrane-interactive properties of KLK [6]. These featuresf KLK can directly be linked to a novel and selective effect of theeptide on the GPI-AP/GEEC-uptake and recycling system (Weghu-er et al., submitted for publication). These studies add IC31® tohe list of human vaccine adjuvants with a broadly explored mech-nism of action.

cknowledgements

Valuable comments of Eszter Nagy and Karen Lingnau are appre-iated. This work was supported by the BRIDGE Grant (No.: 815438)rom FFG.

eferences

[1] Wack A, Rappuoli R. Vaccinology at the beginning of the 21st century. Curr OpinImmunol 2005;17:411–8.

[2] Lingnau K, Riedl K, von Gabain A. IC31 and IC30, novel types of vaccine adjuvantbased on peptide delivery systems. Expert Rev Vac 2007;6:741–6.

[3] Yamada K, Natori S. Characterization of the anti-microbial peptide from sapecinB, an anti-bacterial protein of sarcophagi peregrine (flesh fly). Biochem J1994;298:623–8.

[4] Fritz JH, Brunner S, Birnstiel M, Buschle M, von Gabain A, Mattner F,

et al. The artificial antimicrobial peptide KLKLLLLLKLK induces predomi-nantly a TH2-type immune response to co-injected antigens. Vaccine 2004;22:3274–84.

[5] Schellack C, Prinz K, Egyed A, Fritz JH, Wittmann B, Ginzler M, et al. IC31, anovel adjuvant signaling via TLR9, induces potent cellular and humoral immuneresponses. Vaccine 2006;24:5461–72.

[

[

e 29 (2011) 426–436

[6] Aichinger MC, Ortbauer M, Reipert S, Zauner W, Bogner P, Froschauer E, et al.Unique membrane-interacting properties of the immunostimulatory cationicpeptide KLKL5KLK (KLK). Cell Biol Int 2008;32:1449–58.

[7] Reutterer B, Stockinger S, Pilz A, Soulat D, Kastner R, Westermayer S, et al. TypeI IFN are host modulators of strain-specific Listeria monocytogenes virulence.Cell Microbiol 2008;10:1116–29.

[8] Latz E, Schoenemeyer A, Visintin A, Fitzgerald KA, Monks BG, Knetter CF, et al.TLR9 signals after translocating from the ER to CpG DNA in the lysosome. NatImmunol 2004;5:190–8.

[9] Bartholomeu DC, Ropert C, Melo MB, Parroche P, Janqueira CF, Teixeira SMR,et al. Recruitment and endo-lysosomal activation of TLR9 in dendritic cellsinfected with Trypanosoma cruzi. J Immunol 2008;181:1333–44.

10] Kamath AT, Rochat AF, Valenti MP, Agger EM, Lingnau K, Andersen P, et al.Adult-like anti-mycobacterial T cell and in vivo dendritic cell responses fol-lowing neonatal immunization with Ag85B-ESAT-6 in the IC31 adjuvant. PloSONE 1998;3:e3683.

11] Kamath AT, Valenti MP, Rochat AF, Agger EM, Lingnau K, von Gabain A, et al. Pro-tective anti-mycobacterial T cell responses through exquisite in vivo activationof vaccine-targeted dendritic cells. Eur J Immunol 2008;38:1247–56.

12] Riedl K, Riedl R, von Gabain A, Nagy E, Lingnau K. The novel adjuvant IC31strongly improves influenza vaccine-specific cellular and humoral immuneresponses in young adult and aged mice. Vaccine 2008;25:3461–8.

13] Gong Q, Huntsman C, Ma D. Clathrin-independent internalization and recycling.J Cell Mol Med 2008;12:126–44.

14] Sabharanjak S, Sharma P, Parton R, Mayor S. GPI-anchored proteins aredelivered to recycling endosomes via a distinct cdc42-regulated, clathrin-independent oinocytic pathway. Dev Cell 2002;2:411–23.

15] Mayor S, Pagano RE. Pathways of clathrin-independent endocytosis. Nat RevMol Cell Biol 2007;8:603–12.

16] Spang A. On the fate of early endosomes. Biol Chem 2009;390:753–9.17] Latz E, Verma A, Visintin A, Gong M, Sirois CM, Klein DCG, et al. Ligand-

induced conformational changes allosterically activate Toll-like receptor 9. NatImmunol 2007;8:772–9.

19] Garin J, Diez R, Kieffer S, Dermine JF, Duclos S, Gagnon E, et al. The phagosomeproteome: insight into phagosome functions. J Cell Biol 2001;8:165–80.

20] Guermonprez P, Saveanu L, Kleijmeer M, Davoust J, Van Endert P, Amigorena S.ER-phagosome fusion defines an MHC class I cross-presentation compartment

in dendritic cells. Nature 2003;425:397–402.

21] Prchal M, Pilz A, Simma O, Lingnau K, von Gabain A, Strobl B, et al. Vaccine2009;27S:G17–20.

22] Lundmark R, Doherty GJ, Howes MT, Cortese K, Vallis Y, Parton RG, et al. TheGTPase-activating protein GRAF1 regulates the CLIC/GEEC endocytic pathway.Curr Biol 2008;18:1802–8.