Distinct Uptake Mechanisms but Similar Intracellular Processing of Two Different Toll-like Receptor Ligand-Peptide Conjugates in Dendritic Cells*

Covalent conjugation of Toll-like receptor ligands (TLR-L) to synthetic antigenic peptides strongly improves antigen presentation in vitro and T lymphocyte priming in vivo. These molecularly well defined TLR-L-peptide conjugates, constitute an attractive vaccination modality, sharing the peptide antigen and a defined adjuvant in one single molecule. We have analyzed the intracellular trafficking and processing of two TLR-L conjugates in dendritic cells (DCs). Long synthetic peptides containing an ovalbumin cytotoxic T-cell epitope were chemically conjugated to two different TLR-Ls the TLR2 ligand, Pam3CysSK4 (Pam) or the TLR9 ligand CpG. Rapid and enhanced uptake of both types of TLR-L-conjugated peptide occurred in DCs. Moreover, TLR-L conjugation greatly enhanced antigen presentation, a process that was dependent on endosomal acidification, proteasomal cleavage, and TAP translocation. The uptake of the CpG∼conjugate was independent of endosomally-expressed TLR9 as reported previously. Unexpectedly, we found that Pam∼conjugated peptides were likewise internalized independently of the expression of cell surface-expressed TLR2. Further characterization of the uptake mechanisms revealed that TLR2-L employed a different uptake route than TLR9-L. Inhibition of clathrin- or caveolin-dependent endocytosis greatly reduced uptake and antigen presentation of the Pam-conjugate. In contrast, internalization and antigen presentation of CpG∼conjugates was independent of clathrin-coated pits but partly dependent on caveolae formation. Importantly, in contrast to the TLR-independent uptake of the conjugates, TLR expression and downstream TLR signaling was required for dendritic cell maturation and for priming of naïve CD8+ T-cells. Together, our data show that targeting to two distinct TLRs requires distinct uptake mechanism but follows similar trafficking and intracellular processing pathways leading to optimal antigen presentation and T-cell priming.


Covalent conjugation of Toll-like receptor ligands (TLR-L) to synthetic antigenic peptides strongly improves antigen presentation in vitro and T lymphocyte priming in vivo.
These molecularly well defined TLR-L-peptide conjugates, constitute an attractive vaccination modality, sharing the peptide antigen and a defined adjuvant in one single molecule. We have analyzed the intracellular trafficking and processing of two TLR-L conjugates in dendritic cells (DCs). Long synthetic peptides containing an ovalbumin cytotoxic T-cell epitope were chemically conjugated to two different TLR-Ls the TLR2 ligand, Pam 3 CysSK 4 (Pam) or the TLR9 ligand CpG. Rapid and enhanced uptake of both types of TLR-L-conjugated peptide occurred in DCs. Moreover, TLR-L conjugation greatly enhanced antigen presentation, a process that was dependent on endosomal acidification, proteasomal cleavage, and TAP translocation. The uptake of the CpGϳ conjugate was independent of endosomally-expressed TLR9 as reported previously. Unexpectedly, we found that Pamϳconjugated peptides were likewise internalized independently of the expression of cell surface-expressed TLR2. Further characterization of the uptake mechanisms revealed that TLR2-L employed a different uptake route than TLR9-L. Inhibition of clathrin-or caveolindependent endocytosis greatly reduced uptake and antigen presentation of the Pam-conjugate. In contrast, internalization and antigen presentation of CpGϳconjugates was independent of clathrin-coated pits but partly dependent on caveolae formation. Importantly, in contrast to the TLR-independent uptake of the conjugates, TLR expression and downstream TLR signaling was required for dendritic cell maturation and for priming of naïve CD8 ؉ T-cells. Together, our data show that targeting to two distinct TLRs requires distinct uptake mechanism but follows similar trafficking and intracellular processing pathways leading to optimal antigen presentation and T-cell priming.
Toll-like receptors (TLR) 3 are germ line-encoded receptors expressed mainly on cells of the innate immune system, such as granulocytes, macrophages, and dendritic cells (DCs). These receptors are important in sensing infectious agents through recognition of pathogen-associated molecules and act as a communicator between innate and adaptive immune responses. The receptors are expressed either on the cell surface or in the endosomal organelles. This compartmentalization of the TLR correlates with the type of ligands with which they interact. The TLRs expressed on the cell surface bind to extracellular components of the microorganisms (such as bacterial LPS to TLR4, bacterial lipopeptide to TLR2). In contrast, the TLRs found in the endosomes bind to ligands derived from intracellular molecules of the pathogen, such as unmethylated CpG DNA sequences to TLR9 and single-stranded RNA to TLR7 (1). Studies have shown that ligands interacting with the latter type of TLRs are internalized independently of the TLRs (2). Upon engagement of the ligand to its receptor, a cascade of intracellular signaling events is initiated that involves docking of different adaptor molecules such as MyD88 and TRAM to the TLR receptors and recruitment of proteins belonging to the IRAKfamily, which ultimately culminate in the activation of the NF␤ transcription factor and gene transcription leading to production of proinflammatory cytokines (3).
DCs are both initiators and regulators of T cell responses (4). Dendritic cells constantly screen the environment for potential foreign antigens by a variety of mechanisms such as phagocytosis, macropinocytosis, caveolin-mediated, or clathrindependent endocytosis. The manner of uptake is dependent on the size and nature of material to be internalized (5,6). As specialized antigen-presenting cells, DCs have the capacity to efficiently process exogenous proteins and present the peptides in major histocompatibility complex (MHC) class I molecules, a process known as cross-presentation. In this scenario exogenously derived antigens are internalized and translocate from the endosomal route into the cytosol, where the proteasome complex processes the antigen. The generated peptides are transported from the cytosol into the endoplasmic reticulum via the peptide transporter TAP (7), after which the peptides undergo further trimming and are finally loaded onto MHC class I molecules, which translocate to the cell surface, where the peptide is presented to CD8 ϩ T-cells.
The ability of DCs to cross-present peptides on MHC class I to CD8 ϩ T-cells together with the capacity of TLR ligands to deliver maturation signals has inspired efforts to explore the use of DCs as a vaccine vehicle in the fight against infectious diseases and cancer (8,9). Covalent linkage of immunogenic peptides to the TLR9 ligand, CpG DNA, or TLR2 ligands, like Pam 3 CysSS and Pam 3 CysSK 4 , induces a more prominent T-cell response than administration of free TLR2-L or TLR9-L mixed with protein (2, 10 -16).
To explore the mode of action of TLR-L antigen conjugates, we have designed well defined synthetic vaccines composed of peptides containing the model antigen ovalbumin CD8 ϩ cytotoxic T-lymphocyte (CTL) epitope (SIINFEKL) chemically linked to either the TLR2-ligand, Pam 3 CysSK 4 , or the TLR9 ligand, CpG. These conjugates were used to study the uptake, intracellular routing, and processing. We show that not only TLR9-L conjugates but also the TLR2-L conjugates are taken up independently of TLR expression albeit through two distinct internalization mechanisms. Down-stream processing route for MHC class I antigen presentation, however, were similar and requires endosomal acidification, TAP translocation, and proteasomal processing. Importantly, whereas the uptake of both types of TLR-L conjugates was independent of TLR expression, priming of specific CD8 ϩ T-cell response required TLR signaling in dendritic cells.

Cell Lines
Freshly isolated DCs were cultured from mouse bone marrow cells as described elsewhere (17). D1 cell line, a long term growth factor-dependent immature splenic DC line derived from B6 (H-2 b ) mice, was cultured as described (18). B3Z is a T-cell hybridoma specific for the H-2K b CTL epitope SIIN-FEKL that expresses a ␤-galactosidase construct under the regulation of the NF-AT element from the IL2 promoter (19). EG7 (EL4-OVA) (20) was cultured in complete medium with 400 g of G418 (Invitrogen). Table 1 shows the conjugates and peptides used in this study.

Chemicals
HCTU was purchased from IRIS Biotech GmbH (Germany), and Pam 3 Cys-OH was from Bachem. PyBOP (benzotriazole-1yl-oxytrispyrrolidinophosphonium hexafluorophosphate) was purchased from MultiSynTech GmbH. Reactive fluorescent dyes BODIPY-FL N-(2-aminoethyl) maleimide, Alexa Fluor 488 C 5 maleimide, and Alexa Fluor 488 carboxylic acid succinimidyl ester were purchased from Invitrogen. Fmoc amino acids were from SENN Chemicals or from MultiSynTech GmbH. Tentagel-based resins were purchased at Rapp Polymere GmbH. All chemicals and solvents used in the solid phase peptide synthesis were from Biosolve. Chemicals, resins, and solvents used in the solid phase DNA synthesis except Beaucage reagent and Control Pore Glass support used to introduce 3Ј-thiol modification were from Proligo and were used as received. 3Ј-Thiol modifier C3 S-S Control Pore Glass and Beaucage reagent were purchased at Glen Research. All chemicals were used as received.

General Methods
Mass spectra were recorded on a PE SCIEX API 165 (PerkinElmer Life Sciences) mass spectrometer. Analytical liquid chromatography/mass spectroscopy was conducted on a Jasco system using an Alltima C 18 analytical column (4.6 ϫ 150 mm, 5-m particle size, flow 1.0 ml/min) with detection at 214 and 254 nm. The solvent system was 100% water (A), 100% acetonitrile (B), and 1% trifluoroacetic acid (C). Gradients of A to B were applied over 15 min, keeping C isocratic at 10%. Purifications of the synthetic peptides were conducted on a BioCAD Vision automated HPLC system (PerSeptive Biosystems, Inc.) supplied with a Alltima C 18 column (10 ϫ 250 mm, 5-m particle size, running at 4 ml/min). A Varian DMS 200 UV-visible spectrophotometer was used to measure UV absorption MALDI-time-of-flight spectra were recorded on a Voyager-DE PRO mass spectrometer (Perseptive Biosystems, Inc.).

Oligonucleotide Synthesis
DMT-based solid-phase phosphorothioate oligonucleotide (ODN) synthesis was performed on an Expedite automated instrument (Perseptive Biosystems) starting from Control Pore Glass support with the 3Ј-thiol modifier. The syntheses were performed on a 10-mol scale according to established methods (22). Elongation was performed using 3Ј-phosphoramidite derivatives of DMT-protected nucleosides 5Ј-DMT-A(TAC)-OH, 5Ј-DMT-C(TAC)-OH, 5Ј-DMT-G(TAC)-OH, and 5Ј-DMT-T-OH under the agency of dicyanoimidazole. After each coupling, remaining free 5Ј-hydroxyls were blocked using a capping solution (t-butylphenoxyacetic anhydride (Tac 2 O)/ 1-methylimidazole in tetrahydrofuran/pyridine) followed by sulfurization of the phosphite linkage to the phosphorothioate linkage using the Beaucage reagent. Next, the 5Ј-DMT protecting group was removed by trichloroacetic acid, after which elongation was continued. After final DMT removal the DNA oligomer was cleaved from the resin by 25% ammonium hydroxide solution to give a 3Ј-disulfide modified ODN (ODN-SSpropyl-OH). ODNs were purified on a Q-Sepharose column pre-equilibrated with 50 mM NaOAc applying a gradient of 2 M NaCl in 50 mM NaOAc. Fractions containing the pure product were combined and dialyzed three times with Millipore water using dialysis tubing with 1-kDa cut-off (Spectrum). Quantification was performed by UV absorbance at 260 nm. Sequences of the ODN prepared in this study were CpG, 5Ј-TCCAT-GACGTTCCTGACGTT-3Ј; GpC, 5Ј-TCCATGAGCTTCC-TGATG-3Ј.

ODN-Peptide Conjugates
For CpG-OVA 247-264 , 3Ј-disulfide modified CpG-SS-propyl-OH (266 nmol) was converted to 3Ј-SH-modified CpG-SH overnight with dithiothreitol (DTT)-containing buffer (35 mg of DTT, 26 mg of NaOAc⅐3H 2 O, 1 ml of water). Dithiothreitol was removed from the mixture using a PD-10 desalting column (Amersham Biosciences) that was pre-equilibrated with 25 ml of a 50 mM phosphate buffer (25 mM Na 2 HPO 4, 25 mM NaH 2 PO 4 , 1 mM EDTA in water, continuously degassed with helium. Filtrate (3.25 ml) was directly transferred to a tube containing 5 mg of maleimidopropionyl peptide Mal-OVA 247-264 . The resulting solution was sonicated and placed under blanket of argon. The tube was sealed and shaken for 2 days at room temperature. The mixture was purified over a Superdex 75 gel filtration column using isocratic elution with 0.15 M triethylammonium acetate. Fractions containing the product were collected and lyophilized. Excess triethylammonium acetate was removed by lyophilization from water (3 times). Quantification was performed by UV absorbance (260 nm). CpG-OVA 247-264A5K and GpC-OVA 247-264 were prepared as described for CpG-OVA 247-264 starting from the corresponding ODNs and maleimidopropionyl peptides.
CpG IL-12p40 Enzyme-linked Immunosorbent Assay-DCs (4 ϫ 10 4 ) were plated into a 96-well round bottom plate and incubated for 24 -48 h with the compounds indicated in the legend to Fig. 1. Supernatants were harvested and tested for IL-12 p40/ p70 content using a standard sandwich enzyme-linked immunosorbent assay as previous described (23).

Confocal Microscopy
DCs were plated out into glass-bottom Petri dishes (MatTek) 2 days before the experiment. Cells were incubated either with the fluorescence-labeled TLR-L peptide conjugate or the fluorescence-labeled peptide at 37°C for different time periods at the concentrations indicated in the figure legends. In some experiments, as indicated, cells were coincubated with 1 M Lysotracker for 5 min to stain endosomal/lysosomal compartments. In experiments with inhibitors, cells were preincubated for 30 min either with 50 M MDC or 10 g/ml filipin followed by extensively washing before incubation with the TLR-L conjugate either alone (pretreatment) or in the presence of inhibitors (coincubation). Cells were then washed and imaged using an inverted Leica SP2 confocal microscope. Dual color images were acquired by sequential scanning, with only one laser line per scan to avoid cross-excitation. The images were processed using the software program ImageJ.

Flow Cytometry
For analyzing the effect of the different compounds on dendritic cell phenotypic profile, DCs were incubated with the different compounds at a final concentration of 1 M for 48 h. Subsequently, cells were harvested, resuspended in fluorescence-activated cell buffer (phosphate-buffered saline, 0.1% bovine serum albumin), and incubated for 20 min with the following panel of monoclonal antibodies: fluorescein isothiocyanate-anti-CD86 (clone GL-1), phycoerythrin-anti-I-A b (clone M5/114 15.2), phycoerythrin-anti-CD40 (clone 3/23), allophycocyanin-anti-K b (24). Cells were washed twice and fixed with 0.5% paraformaldehyde before being subjected to flow cytometry measurements.

MHC Class I-restricted Antigen Presentation Assay
DCs were incubated for 2 h (unless stated otherwise in the legends to Figs. 6 and 8) with 1) the parent peptide (DEVS-GLEQLESIINFEKLAAAAAK, OVA 247-264A5K , or DEVS-GLEQLESIINFEKL, OVA 247-264 ), 2) the peptide conjugate, or 3) the mixture of the parent peptide and the Pam 3 CySK 4 or CpG at the indicated concentrations. Cells were washed five times with medium before the T-cell hybridoma B3Z cells were added and incubated for 16h at 37°C.
Antigen presentation of the ovalbumin cytotoxic T-cell epitope SIINFEKL (OVA 257-264 ) in H-2K b was detected by activation of B3Z cells measured by a colorimetric assay using chlorophenol red-␤-D-galactopyranoside as substrate to detect lacZ activity in B3Z lysates, as described (23).
In some experiments cells were preincubated for 30 min with titrated amounts of epoxomicin ranging from 0.01 to 10 M epoxomicin or with 3 mM NH 4 Cl or preincubated for 60 min with titrated amounts of monodansylcadaverine (25-50 M) or with filipin (10 -20 g/ml) before adding the peptide compound still in the presence of the inhibitors. Cell viability was confirmed by tryphan blue exclusion at the indicated concentration range of inhibitors.

Intracellular Cytokine Staining
An aliquot of spleen cells after re-stimulation and Ficoll purification (see above) were subjected to stimulation in vitro with or without 1 g/ml OVA 257-264 peptide (H2-K b restricted SIINFEKL) overnight in the presence of GolgiPlug (BD Pharmingen). Cells were then washed twice with fluorescence-activated cell sorter buffer and stained with phycoerythrin-conjugated monoclonal rat anti-mouse CD8b2 antibody. Cells were subjected to intracellular cytokine staining using the Cytofix/ Cytoperm kit according to the manufacturer's instructions (BD Pharmingen). Intracellular interferon-␥ was stained with allophycocyanin-conjugated rat anti-mouse interferon-␥. All antibodies were purchased from BD Pharmingen. Flow cytometry analysis was performed using FACSCalibur with CELLQuest software (BD Biosciences). Splenocytes without peptide stimulation were used as a negative control.

In Vivo Uptake Studies
To monitor the uptake of the TLR-L-conjugated peptides and the free peptide in vivo, mice were injected s.c. either with 5 nmol of Alexa488 Fluor-labeled CpGϳconjugated peptide, 5 nmol of BODIPY-FL Pamϳconjugated peptide, CpG mixed with Alexa488 Fluor-labeled peptide, or Pam mixed with BODIPY-FL-labeled peptide. Three days later mice were sacrificed, and a single cell suspension of the draining lymph nodes was stained for CD11c (clone HL3) before being subjected to flow cytometry analysis.

RESULTS
To study the uptake, trafficking, and processing of two distinct TLR-ligand peptide conjugates in dendritic cells for MHC class I presentation to CD8 ϩ cytotoxic T-lymphocytes, we selected the TLR9 ligand CpG and the TLR2 ligand Pam 3 CysSK 4 (Pam) based on the fact that these two ligands interact with two distinct receptors located either in the endosomal compartment (TLR9) or on the plasma membrane (TLR2). As a model antigen we made use of peptides containing the CTL epitope (SIINFEKL, designated OVA 257-264 ) derived from the ovalbumin protein. Two different peptide length variants were synthesized, an extended peptide that required proteasome-dependent processing on both the N and C terminus to release the CTL epitope (DEVS-GLEQLESIINFEKLAAAAAK) designated OVA 247-264A5K , and a shorter peptide, of which C-terminal processing by the proteasome is not required (DEVSGLEQLESIINFEKL), designated OVA 247-264 . For example, peptide OVA 247-264A5K conjugated to Pam 3 CysSK 4 is designated as PamϳOVA 247-264A5K .
Crucially, all of our compounds (listed in Table 1) are produced synthetically and are, therefore, chemically well defined, of high purity, and of constant quality, avoiding potential contamination with other TLR ligands such as LPS, which commonly occur in purified TLR ligand preparations.
Robust Induction of Naïve CD8-specific T-cells Mediated by the Conjugates-To establish the quality of our generated TLR-L-peptide conjugates, we first investigated the induction of an endogenous T-cell response following s.c. injection of either TLR-L-conjugated peptide or free peptide into naïve C57BL/6 mice. After 10 days, the induction of OVA 247-264 -specific CD8 ϩ T-cells was analyzed. As is shown in Fig. 1A, the magnitude of the OVA 257-264 -specific T-cell response induction by either CpGϳOVA 247-264A5K or PamϳOVA 247-264A5K was significantly higher than that in mice injected with non-conjugated OVA 247-264A5K peptide mixed together with either free CpG or free Pam. To address whether the induction of specific T-cells depended on activation of the DCs, we injected GpCϳOVA 247-264 conjugate, which is a non-stimulatory oligonucleotide, as shown by its lack of capacity to induce IL-12 production by DCs (Fig. 1C). Injection of the GpCϳpeptide conjugate into naïve mice led to a significantly lower induction of specific CD8 ϩ T-cells than of CpGϳconjugated peptide but was still as high as the response obtained after mixing of peptide with the CpG (Fig. 1A). Importantly, only when a stimulatory TLR-L conjugate was given, were the majority of CD8 ϩ T-cells able to produce interferon-␥ (Fig. 1B), indicating that signaling via the TLR is essential for the generation of large numbers of functional T-cells in vivo. These results suggest that the enhanced induction of specific T-cells is primarily the result of efficient delivery of the TLR-L-conjugated peptide into the antigen-presenting cell.
TLR Conjugates Activate Dendritic Cells-Next we analyzed the ability of the different conjugates to induce maturation of DCs. As evident from Table 2, increased surface marker expression of CD40, CD86, MHC I, and MHC II was observed in dendritic cells after treatment with the different TLR-L conjugates to a similar extent as the free TLR-Ls. To confirm the involvement of the TLRs in DC activation, we isolated bone marrow-derived dendritic cells (BMDCs) from wild type (WT) mice, TLR2-deficient mice, and TLR9-deficient mice and stimulated these cells with the different conjugates followed by phenotypic characterization by staining for different surface markers associated with DC maturation (25). No up-regulation of the cell surface markers CD86 and MHC class II was detected when BMDCs from TLR2-deficient mice or TLR9-deficient mice were stimulated with Pamϳconjugated peptide or CpGϳconjugated peptide, respectively (Fig. 2). This impaired up-regulation was not due to a general defect in the maturation signaling pathway, because stimulation with LPS could induce up-regulation of CD86 and MHC II in BMDCs derived from both TLR2-and TLR9-deficient mice to a similar extent as observed for BMDCs derived from wild type mice. Taken together, these results demonstrate that the conjugates are as effective as free TLR ligand in activating the DCs and show that the expression of the cognate TLR is required for activation of DCs by the TLR-L-peptide conjugates.
Efficient Uptake of CpG-and Pam-conjugated Antigen Peptides by Dendritic Cells-Having demonstrated that TLR signaling was important for DC activation and priming of T-cells, we decided to compare the downstream cellular mechanism used by the two types of TLR-L with respect to uptake, routing, and cellular processing. Therefore, the efficiency of antigen uptake of conjugated versus non-conjugated peptide by DCs was determined. DCs were incubated with either Pamϳconjugated peptide or free peptide. Both compounds were labeled with a fluorophore (attached on the peptide backbone), which allowed us to monitor the internalization of these compounds by confocal microscopy analysis. Introduction of the fluorophore (either Alexa488 or BODIPY-FL) into the conjugates did not alter the ability of the conjugates to activate DCs as comparable levels of the IL-12 cytokine were produced by the fluorescent conjugates and the dark conjugates. 5 As indicated by the increased intensity of fluorescence inside the cells, the Pamϳconjugated peptides were taken up far more efficiently than the non-conjugated peptide (Fig. 3A). Interestingly, the Pamϳconjugated peptide was found to accumulate in hot spots (indicated by arrows in Fig. 3A), whereas a more diffuse pattern was observed in DCs incubated with the peptide alone. Quantification of the mean fluorescence intensity (MFI) revealed a more than 4-fold higher fluorescence in DCs incubated with the Pamϳconjugate compared with DCs incubated with the peptide (Fig. 3B). Similarly, CpGϳconjugated peptide was internalized more efficiently than the unconjugated peptide by DCs (Fig. 3C). In addition, the non-stimulatory GpCϳconjugate (MFI 63 Ϯ 7.3) was inter- nalized to a similar extent as the stimulatory CpGϳconjugate (MFI 72 Ϯ 9.1) (Fig. 3D).
These comparisons indicate that the fluorescent TLR-L conjugates are taken up much more efficiently by DCs than unconjugated peptides in vitro. To examine the uptake efficiency in vivo, mice were injected with either Alexa488 Fluor-labeled CpGϳconjugate or peptide labeled with Alexa488 Fluor mixed with dark CpG. Three days later draining lymph node cells were stained for the DC surface marker CD11c. In line with the in vitro results, a significantly higher proportion of CD11c ϩ cells had taken up the CpGϳconjugated peptide (2.5%) when compared with the population of DCs that ingested unconjugated peptide (0.3%) or non-injected mice (0.1%; Fig. 3E). A similar tendency was observed upon comparison of BOPIPY-FL-labeled Pamϳconjugate with peptide labeled with BODIPY-FL mixed with dark Pam (Fig. 3F). Collectively, these results indicate that it is the covalent linking of the peptide to the TLR-L that is responsible for the enhanced uptake by the DCs.
Conjugates Translocate to the Endosomal/Lysosomal Compartment Independently of TLR Expression-Our antigen uptake studies revealed that also the TLR9-L conjugate was taken up more efficient despite that TLR9 is not surface-expressed. Therefore, to evaluate the relevance of TLR expression for the enhanced uptake, BMDCs purified from WT, TLR2 Ϫ/Ϫ , and TLR9 Ϫ/Ϫ mice were incubated with Alexa488 Fluor-labeled TLR-L conjugates. As shown in Fig. 4, A and C, BMDCs from TLR9 Ϫ/Ϫ mice internalized CpGϳconjugates to a similar extent as BMDCs from wild type mice, as reported previously (2).
Surprisingly, similar experiment carried out with Pamϳconjugate and BMDCs from WT mice and TLR2 Ϫ/Ϫ mice showed that also the Pam conjugate, despite the fact that the receptor for Pam (TLR2) is located on the cell surface (1,26), were internalized equally well by both types of DCs (Fig. 4,  B and D).
To monitor the intracellular localization of the conjugates, we performed a co-localization study between the conjugates (green) and the endosomes (red). As seen in Fig. 4E, both the CpGϳ and Pamϳconjugates are co-localized partially with an endosomal tracker (Lysotracker) in a pattern characteristic for the endosomal vehicles. Moreover, no overall difference in the uptake kinetic or in the trafficking of the compounds could be detected when comparing BMDCs from wild type mice to mice deficient for either TLR9 or TLR2 expression. 5 These results indicate that TLR expression is not required for uptake and that the conjugate re-locates to the endosomal compartment.
Conjugation of Peptide Leads to Pronounced Enhancement in Antigen Presentation in Vitro-Having established that the conjugates were taken up much more efficiently than the free peptide harboring the OVA CTL epitope SIINFEKL, we next addressed the effect of conjugation of peptide to TLR-L on antigen presentation. DCs were loaded with either the CpGϳconjugated peptide, Pamϳconjugated peptide, the CpG    or Pam mixed with the peptide, or the peptide alone (Fig. 5, A and B) before incubation with the peptide-specific T-cell hybridoma B3Z cells that recognize the H-2K b , SIINFEKL CTL epitope (27). As the concentration of the compounds decreased, antigen recognition was rapidly lost when DCs were incubated either with the peptide alone or with the peptide mixed with CpG but not when the CpGϳconjugated peptide was used. This indicates that the conjugation of peptides to TLR-L enhances antigen presentation (Fig.  5A). Likewise, an increased antigen presentation was observed for the Pamϳconjugated peptide (Fig. 5B).
In this case the difference in antigen presentation between conjugate and non-conjugate was even more prominent. Incubation with a mixture of free TLR-L (Pam or CpG) and the peptides resulted in a decreased antigen presentation by DCs when compared with loading with peptide alone or with conjugated peptide. This might be related to decreased uptake, since it has previously been reported that the endocytotic capacity of DCs declines upon encountering maturation signals (28,29).
To gain insight in the kinetics of antigen presentation, DCs were incubated for various time periods with either TLR-L-conjugated peptide, peptide mixed with TLR-L, or peptide alone. As shown in Fig. 5, C and D, it required 24 -48 h of continuous presence of peptide mixed with CpG or Pam or of peptide alone to reach the level of antigen presentation acquired already after 2 h of incubation with the conjugated peptide, as measured by an equal ability to stimulate the peptidespecific B3Z hybridoma T-cells. Thus, conjugation greatly improves the swiftness of presentation of antigen by DCs for stimulation of T-cells.
The confocal microscopy results indicated that uptake of the conjugates occurred independently from the expression of the respective TLRs. Therefore, we next evaluated the impact of TLR expression upon antigen presentation. To this end BMDCs from WT, TLR2 Ϫ/Ϫ , and TLR9 Ϫ/Ϫ mice loaded with the conjugates were incubated in vitro together and subsequently incubated with the peptide-specific T-cell hybridoma B3Z cell line. In line with the confocal uptake studies, BMDCs derived from WT or the TLR2-or TLR9-deficient mice strains were recognized to the same extent (Fig. 6, A and B).
Disruption of Clathrin Formation and Caveolae Clustering Blocks Antigen Presentation of TLR-L Conjugates-DCs can take up exogenous antigens via different mechanisms such as clathrin-mediated endocytosis, fluid phase endocytosis, and macropinocytosis (30). To define the pathways by which the TLR-L-peptide conjugates were internalized, DCs were pretreated with different inhibitors before loading with the TLR-L conjugates. Macropinocytosis inhibitor 5-(N,N-dimethyl)amiloride (29) had no effect on antigen presentation. 5 On the other hand pretreatment with filipin, a sterol binding agent that disrupt caveolae structures (31) and thereby lipid-raft formation, reduced antigen presentation of both Pamϳconjugates as well as CpGϳconjugates in a dose-dependent manner, whereas antigen presentation of the CTL epitope OVA 257-264 was only marginal affected (Fig. 6C). Because lipid raft formation is involved both in clathrin-dependent endocytosis as well as caveolae-dependent internalization (32), we analyzed the impact of MDC, a specific inhibitor of clathrin formation (33,34), upon antigen presentation. Interestingly, whereas antigen presentation of CpGϳconjugated peptides was not affected by MDC, antigen presentation of Pamϳconjugated peptides was abrogated in a dose-dependent fashion (Fig. 6D). To further explore the distinct uptake mechanisms used by the two types of conjugates, confocal microscopy was performed on DCs treated with the two inhibitors. As evident from Fig. 7, both MDC and filipin abolished the internalization of the Pam conjugate in terms of mean fluorescence per cell, whereas inhibition of clathrin formation had a less pronounced effect on internalization of the CpGϳconjugate (Fig. 7C). On the other hand, inhibition of caveolin formation by filipin reduced the mean fluorescence of cells incubated with the CpGϳconjugate. The selective effect of the inhibitors was not due to a direct effect on one conjugate, as preincubation of cells with the inhibitor followed by extensive washing before incubation with the conjugates (indicated as preincubation in Fig. 7) in the absence of inhibitors led to similar results. Thus, these results indicate that the two TLR-L conjugates are internalized by distinct uptake receptors.
Antigen Presentation Depends upon Endosomal Acidification, Proteasome Activity, and TAP Translocation-Next, we examined the impact of different proteases upon antigen pres-entation. To address this issue, we made use of the C-terminal extended peptides (OVA 247-264A5K ), which require both N-and C-terminal processing to release the SIINFEKL CTL peptide epitope. DCs were pretreated either with epoxomicin, which inhibits the proteasome, or NH 4 Cl, which increases the pH in the acidic endosome/lysosome environment and thereby inhibits the proteases that depend on acidification (35)(36)(37)(38) before incubation with either of the conjugates. As evident from Fig. 8, A and B, when DCs were pretreated with the lysosomotropic agent, NH 4 Cl, a decrease in antigen presentation was seen ranging from 45% inhibition for the Pamϳconjugates to 70% for the CpGϳconjugated peptide. Similarly, inhibition of the proteasome activity resulted in an overall decrease in antigen presentation of both the CpGϳconjugated and Pamϳconjugated peptide (up to 50% inhibition) in a dose-dependent manner (Fig. 8C). To ascertain that the inhibitory effect observed was not due to an overall decrease in the surface expression of MHC class I, DCs that had been pretreated with either epoxomicin or NH 4 Cl were incubated with the minimal CTL epitope OVA 257-264 . As expected, the inhibitors did not cause a major affect upon antigen presentation of exogenous loaded OVA 257-264 peptide (Fig. 8, B and C).
After proteasomal processing, CTL peptide epitopes need to be translocated into the luminal side of the endoplasmic reticulum via the transporter complex TAP to be loaded onto MHC class I molecules. To address the involvement of TAP for cross presentation of the TLR-L conjugates, BMDCs from either wild type mice or TAP-deficient mice were loaded with the TLR-L conjugates or the minimal CTL epitope OVA 257-264 . As evident in Fig. 8D, antigen presentation of both the TLR2L and TLR9L conjugates depended upon TAP activity, as the presentation was abrogated in TAP-deficient mice. Importantly, antigen presentation of the minimal CTL epitope OVA 257-264 was not affected in the TAP-deficient DCs, showing that the observed TAP dependence of the conjugates was not due to an overall reduced surface expression of MHC class I in the TAP-deficient DCs. In contrast, when using BMDCs from mice deficient in both TAP and ␤2-microglobulin expression (TAP Ϫ/Ϫ ␤2m Ϫ/Ϫ ) completely lacking MHC class I surface expression (39), antigen presentation was completely lost for all peptides (Fig. 8D). Collectively, these results indicate that endosomal acidification, proteasomal activity, and TAP translocation are required for antigen presentation of the TLR-L peptide conjugates and suggest that the peptide (conjugate) translocates via the endosomal compartment into the cytosol where the peptide undergoes proteasomal processing before being loaded onto MHC class I molecules in the endoplasmic reticulum.

DISCUSSION
In this study we analyzed the cellular uptake and trafficking of two distinct TLR ligand-antigen conjugates that ultimately lead to the induction of an efficient CTL response. Strikingly, one single s.c. immunization with conjugate in saline induced an impressive systemic expansion of antigen-specific CD8 T-cells (Fig. 1). Thus, conjugation resulted in a stronger systemic response than what was observed for the mixed vaccines.
Our immunofluorescence analysis revealed that conjugates of both types of TLR ligands were taken up very efficiently compared with unconjugated peptides (Fig. 3). Internalization was a very rapid process since uptake studies showed that already within 15-30 min, the major part of the conjugates could be found in endosome/lysosome compartments (Fig. 4). 5 For this we have used fluorescent conjugates; these may have slightly different properties from those of the unmodified conjugates, which could influence the uptake and function. However, the fluorescent conjugates induced DC maturation to a similar extent as the unmodified conjugates. 5 In line with our findings, CpG linked to fluorescein isothiocyanate-labeled ovalbumin protein was recently shown to translocate to LAMP-1-positive endosomal-lysosomal compartments (10). Further support of enhanced uptake mediated by the conjugates was provided from our in vivo uptake analysis, which revealed a 6 -8-fold increase in uptake of the CpG-conjugated peptide by CD11c ϩ cells and a 2-fold increase in uptake of the Pam-conjugated peptide by CD11c ϩ cells compared with uptake of non-conjugated peptide (Fig. 3, E and F). We found that CpGϳconjugated peptides were internalized independently from the expression of TLR9 and could also support antigen presentation in vitro independently of TLR9 expression (Fig. 4). Accordingly, Wagner and co-workers (2) showed that cross-presentation of OVA-linked CpG occurred independently from TLR9 expression but that TLR9 expression nevertheless was essential for activation of the DCs. At first sight these findings might seem paradoxical; however, TLR9 are mainly expressed in the endoplasmic reticulum followed by recruitment to the endosomes upon dendritic cell maturation (40).
Unexpectedly, considering the cell surface expression of TLR2, we found that internalization of the Pamϳconjugate was taken up independently from the expression of TLR2. BMDCs isolated from TLR2-deficient mice internalized the Pamϳconjugate to a comparable level as BMDCs from wild type mice, and antigen recognition was unaffected in BMDCs from TLR2-deficient mice. This could not be attributed to a side effect mediated by the peptide part, since TLR2-independent internalization was also observed when incubating the cells with free Pam. 5 Importantly, inhibition of clathrindependent endocytosis or caveolae formation abrogated both uptake and antigen presentation of Pam conjugates (Figs. 6 and 7), whereas internalization of CpG conjugates was independent of clathrin-coated pits but dependent on caveolae formation. These results indicate that other (distinct) receptors than the TLRs are involved in the uptake of the TLR conjugates, although the exact nature of these receptors remains to be established. Other TLR2 ligands have been reported to be internalized independently from the expression of TLR2. Outer membrane protein A, a conserved major component of the outer membrane of Enterobacteriaceae family that triggers cytokine production in macrophages and DCs (41), was recently shown to be internalized by the scavenger receptor LOX-1 independently from the expression of TLR2 (42). Moreover, lipoteichoic acid has also been reported to be internalized independently from TLR2 expression (43), although the receptor involved in the uptake of lipoteichoic acid still remains to be identified. Therefore, the contribution of TLR2 and other receptors expressed on the cell surface to the uptake of Pam and other TLR ligands remains to be established.
Optimal presentation of the peptide antigen cargo in the conjugates required endosomal acidification (Fig. 8), although it cannot be ruled out that the fusion of early endosomal vesicles with late endosomal vesicles is hampered by the lysosomotropic agent NH 4 Cl (37, 44 -46).These results imply that endosomal proteases, such as cathepins (45,47), might be involved in the processing of the TLR-L-peptide conjugate. Furthermore, proteasomal cleavage was required because proteasome inhibition greatly decreased antigen presentation of the TLR-L conjugates (Fig. 8C). Because proteasomes are mainly located in the cytosol (48,49), our results indicate that the peptide/conjugate, after being released in the endosomal compartment, translocate to the cytosol to undergo proteasomal processing (45). In this regard it was recently reported that the translocon complex SEC61 could be involved in facilitating the translocation of peptide from the endosomes to the cytosol (50). Moreover, abrogation of translocation of . Cells were incubated with the peptides for 3 h in the presence of inhibitors before being processed as outlined under "Experimental Procedures." Ϫ, untreated; ϩ, treated with NH 4 Cl. Results are the means of triplicates ϮS.E. D, BMDCs from WT mice, TAP-deficient mice (TAP Ϫ/Ϫ ), or mice deficient in TAP and ␤2-microglobulin (TAP Ϫ/Ϫ ␤2m Ϫ/Ϫ ) were preloaded either with CpG-conjugates, Pam-conjugate, or OVA 257-264 before being processed as outlined under "Experimental Procedures." peptides from the cytosol to the endoplasmic reticulum completely abrogated antigen presentation of both types of TLR-Lpeptide conjugates (Fig. 8D). Aside from being internalized efficiently, we found that all of the TLR-L conjugates retained their capacity to activate DCs to a comparable level as the free TLR-L both in terms of production of the Th1-favoring cytokine, IL-12, 5 and up-regulation of DC maturation surface markers (Table 2). Importantly, TLR expression was required for DC activation, since BMDCs lacking TLR2 or TLR9 were not able to up-regulate co-stimulatory molecules upon stimulation with either Pamϳconjugate or CpGϳconjugate, respectively (Fig.  2). In addition, conjugation of the non-stimulatory GpC oligonucleotide to the peptide antigen resulted in inefficient CTL priming (Fig. 1), showing that the DC activation of TLR-L peptide conjugates is essential. Therefore, intracellular signaling of TLR is crucially important for effective CTL priming by the conjugates.
In summary, we demonstrate that well defined synthetic TLR-L-peptide conjugates induce a robust and systemic response of specific T-cells due to the combined action of enhanced antigen uptake, improved MHC class I antigen presentation, and dendritic cell maturation. Our data show that targeting to two different TLRs requires a distinct uptake mechanism independent of TLR expression but follows similar trafficking and intracellular processing pathways leading to optimal antigen presentation and T-cell priming. The chemical properties of these conjugates, which ensure that the same DCs that take up the antigen receives simultaneously a proper maturation signal, is likely to be the mechanism behind the superior activity of the peptide conjugates. Accordingly, Medzhitov and Blander (51) recently reported that synchronous entrance of TLR-L and antigen enhanced MHC class II presentation, although in their system antigen and TLR-L was delivered on microspheres in a non-covalent manner. The collective features of our TLR-L peptide conjugates together with their convenient manufacture and handling makes synthetic peptide-TLR-ligand conjugates an attractive novel vaccine modality.