Abstract
Dendritic cells (DCs) play critical roles in developing immune defenses. One important aspect is interaction with pathogen-associated molecular patterns (PAMPs)/danger-associated molecular patterns, including di- and triacylated lipopeptides. Isolated or synthetic lipopeptides are potent vaccine adjuvants, interacting with cell surface TLR2 heterodimers. In contrast, deep embedment within bacteria cell walls would impair lipopeptide interaction with cell surface TLR2, requiring degradation for PAMP recognition. Accordingly, DC processing in the absence of surface TLR2 ligation was defined using synthetic virus-like particles (SVLPs) carrying hydrophobic TLR2 PAMPs within di- and triacylated lipopeptide cores (P2Cys-SVLPs and P3Cys-SVLPs) compared with SVLPs lacking immunomodulatory lipopeptides. DCs rapidly and efficiently internalized SVLPs, which was dominated by slow endocytic processing via macropinocytosis, although some caveolar endocytosis was implicated. This delivered SVLPs primarily into macropinosomes often interacting with EEA-1+ early endosomes. Although endoplasmic reticulum association was occasionally noted, association with recycling/sorting structures was not observed. Involvement of LysoTracker+ structures slowly increased with time, with SVLPs present in such structures ultimately dominating. Only SVLPs carrying di- and triacylated lipopeptide cores induced DC activation and maturation independently of surface TLR2 ligation. Intracellular recognition of SVLP TLR2 ligands was confirmed by observing SVLPs’ association with internal TLR2, which had similar kinetics to SVLP association with LysoTracker. This related to inflammatory cytokine induction by SVLP+ DCs, with adaptive immune response activation ex vivo/in vivo. Importantly, particular DCs, not monocytes, recognized intracellular exposure of the TLR2 PAMPs carried by di- and triacylated SVLP cores, which indicates subset-distinct recognition of functional internal TLR2 ligands. Thus, vaccines carrying hydrophobic TLR2 ligands would interact with particular DCs for efficient induction of specific immunity in the absence of additional adjuvant.
Introduction
Detection of pathogen-associated molecular patterns (PAMPs) or danger-associated molecular patterns (DAMPs) (1) is an important process during dendritic cell (DC) promotion and regulation of innate and adaptive immune defenses (2, 3). Among PAMP/DAMP recognition receptors is the TLR family, including cell surface TLR2 heterodimers with TLR1 and TLR6 (4). These cell surface TLR2s are known to be ligated by structural components close to pathogen surfaces, such as teichoic acids (TAs) and peptidoglycans (PGNs) in bacterial cell walls (5, 6), leading to accumulation with CD36 in lipid rafts and cointernalization (7–9).
Isolated and synthetic bacterial wall components, including Staphylococcus aureus lipoprotein SitC (7), lipoteichoic acid (LTA; S. aureus), and lipopeptides (such as FSL-1 from Mycoplasma salivarium and synthetic Pam3CysSK4 [P3C]), have been widely studied as TLR2 ligands (8). Using keratinocytes and transfected HEK cells expressing TLR1/2/6, CD14, and CD36, ligand-induced internalization of pre-existing cell surface TLR2/1 and TLR2/6 heterodimers was noted (8). LTA ligation of cell surface TLR2 promoted cointernalization into the Golgi (8, 10), although this favored TLR2/TLR6 over TLR2/TLR1.
Unfortunately, these studies did not use DCs. With murine macrophages, mannose-binding lectin (binds LTA and PGN) increased ligand-promoted TLR2 internalization into peripheral vesicular structures (11). The mannose-binding lectin reached LysoTracker+ structures, although there was no assessment of TLR2 colocalization. Synthetic bacterial wall components (MALP-2, P3C, and Pam2CysSK4 [P2C]) can induce TLR2 internalization into dextran+ and LysoTracker+ structures in RAW 264.7 macrophages (12). Polarization of TLR2 was also noted around zymosan, leading to cointernalization within the same phagosomal structures of RAW-TT10 macrophages (13).
The nature of the observations in the above reports suggests that the important signaling was interaction with cell surface TLR2 prior to processing of the endocytosed material. With regard to intact bacteria, the bacterial wall PGN structure and thickness may facilitate binding of wall TA to surface TLR2 (14), but the deep embedment of hydrophobic lipoproteins within bacterial walls (15) would require degradation for exposure as TLR2 ligands. Under such conditions, it is unclear how surface TLR2 heterodimers would prove effective receptors for the PAMPs of the bacterial hydrophobic lipoproteins. Intracellular TLR2 has been described in unstimulated RAW 264.7 murine macrophages transfected to express human TLR2-GFP (12). The perinuclear localization typical of endoplasmic reticulum (ER)/Golgi staining patterns suggests that this reflected TLR2 synthesis and the subsequent pathway delivering TLR2 to the cell surface for expression.
There is no evidence that internal TLR2 molecules are biologically active; the aforementioned reports used ligation of cell surface TLR2, leading to its internalization. Moreover, there are no reports on how DCs behave, which is particularly important considering their critical roles in developing immune responses and the roles played by TLR ligation therein. Accordingly, we used synthetic virus-like particles (SVLPs) that interact efficiently with DCs and monocytes (16) to assess intracellular TLR2 compartments. SVLPs are formed by self-assembly of diacylated lipopeptide monomers (P2Cys-SVLPs [P2C-SVLPs]) and triacylated lipopeptide monomers (P3Cys-SVLPs [P3C-SVLPs]) (17), such that the lipopeptide TLR2 ligands reside within the core of the particles. They do not interact with cell surface TLRs. Consequently, exposure of TLR2 ligands within the hydrophobic di- and triacylated lipopeptide cores requires intracellular degradation. This allowed identification of how DCs detect PAMPs within di- and triacylated lipopeptides independently of surface TLR2 involvement. Primary porcine DCs and monocyte-derived DCs (MoDCs) were used to ensure measurements relevant to DCs rather than cell lines, as well as to facilitate preparation of cells in adequate amounts at regular intervals from the same donor animals. The results demonstrated that DCs react with TLR2 ligands within the hydrophobic cores of P2C-SVLPs and P3C-SVLPs (P2C/P3C-SVLPs), relating to interaction of the SVLPs with internal TLR2+ compartments. The recognition followed particular routes of endocytic uptake/compartmentalization, leading to cytokine induction and costimulatory molecule upregulation. The latter related to DC maturation and the induction of adaptive immune responses.
Materials and Methods
Ethics statement
Blood donor pigs provided the blood cells, which were collected from the vena jugularis and purified for the generation of MoDCs or sorting of blood DCs. This was performed under permission from the Canton of Bern (Switzerland) through animal license number 26/11. Immunization of pigs with SVLPs was conducted under license number 112-09 issued by the Canton of Bern. All experiments were performed according to the laws on the care and use of laboratory animals in the Canton of Bern and in Switzerland.
SVLP components
Assembly of SVLPs is shown in Fig. 1A. Synthesis of the lipopeptides Pam2Cys-IT*, Pam3Cys-IT*, and 1,3-dipalmitoyl-sn-glycero-phosphatidylethanolamine-IT* (PE-IT*), the structures of which are shown in Supplemental Fig. 1, followed the previously described process (16, 17). For conjugation of Alexa Fluor 488 fluorescent dye, a solution of excess Alexa Fluor 488 C5-maleimide (Invitrogen) in 50% aqueous MeCN (1.2 equivalents) was prepared and added to a solution of the lipopeptide in 50% aqueous MeCN, carefully adjusting the pH to 6.5 and stirring at room temperature for 2–3 h. Conjugates were purified by reverse-phase HPLC on an Interchrom UP5WC4.25M column (250 × 10.0 mm) using a linear gradient of 10–100% MeCN in H2O (+0.1% trifluoroacetic acid [TFA]) in 25 min. Free cysteines were blocked at the thiol with N-ethylmaleimide. The lipopeptides were analyzed by analytical HPLC using an Interchrom UP5WC4-25QS column (21 × 4.6 mm) and a linear gradient of MeCN in H2O (+ 0.1% TFA), and the molecular mass was confirmed by electrospray ionization–mass spectrometry (MS) and/or MALDI-TOF MS. HPLC retention times and the calculated and found molecular masses were determined (summarized in Fig. 1B). All peptides and lipopeptides were >96% pure by HPLC.
SVLP generation and analytical data. (A) Structure of di- and triacylated lipopeptide monomers for assembly of SVLPs, showing lipid positions forming the SVLP core. (B) Analytical data for the peptides and lipopeptides described in (A). HPLC analysis was performed using an Interchrom UP5WC4.25QS column (4.6 × 25 mm) and a gradient of 50–100% MeCN in H2O (+ 0.1% TFA) in 16 min for Alexa Fluor–labeled lipopeptides and 30–100% MeCN in H2O (+ 0.1% TFA) in 25 min for others. The calculated molecular mass (Da) shown is the average calculated molecular mass. The found molecular mass (Da) was determined by electrospray ionization–MS or MALDI-TOF with an accuracy of 0.1%.
SVLP generation and analytical data. (A) Structure of di- and triacylated lipopeptide monomers for assembly of SVLPs, showing lipid positions forming the SVLP core. (B) Analytical data for the peptides and lipopeptides described in (A). HPLC analysis was performed using an Interchrom UP5WC4.25QS column (4.6 × 25 mm) and a gradient of 50–100% MeCN in H2O (+ 0.1% TFA) in 16 min for Alexa Fluor–labeled lipopeptides and 30–100% MeCN in H2O (+ 0.1% TFA) in 25 min for others. The calculated molecular mass (Da) shown is the average calculated molecular mass. The found molecular mass (Da) was determined by electrospray ionization–MS or MALDI-TOF with an accuracy of 0.1%.
Blood DCs and MoDCs
These were prepared from porcine PBMCs (18) using anti-CD172a (signal regulatory protein α) magnetic sorting (16, 19) to yield CD172alo (dominated by the classical DC [cDC]1 subset) and CD172ahi (monocytes plus the cDC2 subset) fractions (20) (see Supplemental Fig. 3A for the gating strategy for flow cytometry).
Porcine monocytes were differentiated into MoDCs over 3–5 d at 39°C (body temperature of pigs used for all porcine cell cultures) with porcine GM-CSF (150 ng/ml) and IL-4 (100 U/ml) in DMEM/GlutaMAX (Life Technologies)/10% v/v porcine serum (Sigma) (19). Bone marrow–derived DCs (BMDCs) were derived from porcine bone marrow hematopoietic cells in a similar manner to the MoDCs but using only GM-CSF for the derivation (21). DCs and monocytes were cultured at 1 × 106 cells per milliliter together with SVLPs or stimulants, as described in 15Results.
Abs, markers, and reagents
Abs were against CD172a (74-22-15A; Washington State University Monoclonal Antibody Center), CD86 (HA5-287; Beckman Coulter), MHC class II (MHCII), EEA-1, clathrin H chain (BD Transduction Laboratories), and α-tubulin (Santa Cruz Biotechnology). ER was labeled with ER-Tracker (Molecular Probes; Life Technologies), lipid rafts were labeled with Alexa Fluor 594– or Alexa Fluor 647–cholera toxin subunit B (CTB), and macropinosomes were labeled with Alexa Fluor 594– or Alexa Fluor 647–dextran 10,000 m.w. (Molecular Probes; Life Technologies). Alexa Fluor 546–transferrin was used to assessed clathrin-mediated endocytosis, and DQ-OVA was used to evaluate acidifying compartments (Molecular Probes).
Intracellular structures to be labeled with specific Abs were detected following p-formaldehyde fixation and saponin permeabilization, as described previously (see below) (16). Taking advantage of the fact that this procedure destroyed the detectability of surface TLR2, intracellular TLR2+ structures were identified with anti-TLR2 mAbs (MAB2616, R&D Systems, mouse IgG2b; CD282, Affymetrix, IgG2a and dependent on the isotypes of other Abs being used concomitantly). Detection of TNF-α or IL-6 production within SVLP+ cells used a similar method, using the Ab Alexa Fluor 647–coupled anti-human TNF-α (AF647, IgG1; BioLegend) or anti-porcine IL-6 (MAB686, IgG1; R&D Systems).
Isotype-specific Alexa Fluor 488/Alexa Fluor 546/Alexa Fluor 594/Alexa Fluor 633 conjugates (Molecular Probes) were used for confocal microscopy; isotype-specific PE- or biotin-conjugated IgG F(ab)′2 (Southern Biotechnology Associates) and streptavidin-SPRD (Dako) were used for flow cytometry.
In certain experiments, DCs were treated with 20 mM methyl-β cyclodextran (MBCD; Sigma) for 20 min at 39°C, before shifting onto ice for an additional 30 min, to disrupt lipid rafts (22), as previously described (16). Fresh medium with 2 mM MBCD was added to prevent cholesterol restoration, followed by addition of the SVLPs and Alexa Fluor 594–CTB for 20 min on ice to assess colocalization in the absence of intact lipid rafts.
TLR ligands used were 10 μg/ml P3C (EMC Microcollections, Germany), 10 μg/ml polyinosinic-polycytidylic acid, 10 μg/ml LPS (Sigma), and 10 μg/ml CpG-ODN-D32 (BioSource). The TLR2 agonists P3C, P3C VacciGrade, Pam2CysSK4, and FSL-1 (InvivoGen) were used at 10 μg/ml. Annexin-V–FITC (Bender MedSystems) was used for cell death assessment.
Confocal microscopy
SVLP-treated DCs (1 × 106 per milliliter) in fibronectin-coated Lab-Tek Chambered Coverglasses (Nunc) were fixed (4% w/v p-formaldehyde) and permeabilized (0.3% w/v saponin) for Ab labeling for 30 min on ice. Detection Abs were applied in 0.3% w/v saponin for 20 min on ice. Confocal microscopy used a Leica TCS SL microscope and LCS software (Leica Microsystems), with acquisition at optimum voxel size (16, 23). All microscopy images (taken with an ×63 objective) are representatives of at least five randomly sampled fields, and experiments were repeated at least three times. This allowed selection of ≥100 cells for the colocalization studies, which used unzoomed images, and to calculate the Pearson coefficient of colocalization.
Analysis used Imaris 7.9 or 8.0 software (Bitplane), applying threshold subtractions and γ-corrections set from negative controls. The Surface Module of the Imaris program was used to enhance visualization of the cell surface, and the Filament Tracer Module was used to enhance visualization of microfilaments and microtubules, as described previously (16, 23). The Imaris program colocalization module was used to assess for colocalized voxels of the SVLP signal with that from the labeling of different intracellular compartments, structures, or organelles. This applied algorithmic analyses of the pixels in the channels under investigation, for which regions of interest (ROIs) were selected around the SVLP+ inclusions and the intracellular structure under study. By such means, the gates and thresholds were set for determining the presence of colocalized voxels using the algorithmic module, as described previously (16, 23).
Flow cytometry
Ab labeling as for microscopy was detected using PE or biotin conjugates. Data were acquired using a FACSCalibur with CellQuest Pro software (Becton Dickinson) (16), and FlowJo v10 software (TreeStar) was used to create graphs and zebra plots. Quadrant and gate statistics used Prism 6 (GraphPad). All images and graphs are from single sets of data that are representative of at least three separate experiments.
NF-κB and AP-1 assays
Mouse RAW-Blue cells, derived from RAW267.4 macrophages (InvivoGen), were used as a reporter cell line. They stably express the secreted embryonic alkaline phosphatase (SEAP) gene, which can be triggered by NF-κB and AP-1 transcription factors. RAW-Blue cells were grown in DMEM/GlutaMAX/10% FBS (heat inactivated), following the manufacturer’s instructions. Cells were passaged when they reached 70–80% confluence. QUANTI-Blue (InvivoGen) was used, according to the manufacturer’s instructions, to detect SEAP in the supernatant of RAW-Blue cells. Supernatants (40 μl) from stimulated cells were added to 160 μl of QUANTI-Blue in a 96-well plate, followed by incubation at 37°C for 15–30 min (until the color developed). Absorbance values were obtained by reading at 650 nm.
Innate immune response induction
After overnight stimulation with SVLPs or TLR ligands, IFN-α, IL-1β, and IL-6 production was assessed by ELISA. mAbs K9 and F17 (kindly provided by Dr. B. Charley, INRA) were used for IFN-α, whereas commercial ELISA kits (R&D Systems) were used for IL-1β and IL-6 (19). Cytokine concentration was calculated from standard curves on each ELISA plate by importing the data into Excel, and graphs were prepared using Prism 6 (GraphPad).
IFN-γ ELISPOT assay
PBMCs from naive porcine blood donors were plated in 96-well multiscreen plates (MAIPS4510; Millipore) in DMEM/GlutaMAX/10% FBS/20 μM 2-ME (Invitrogen), followed by stimulation for 48 h at 39°C with SVLP (0.1–10 μg/ml) or mitogen (10 μg/ml ConA or pokeweed mitogen). Data were obtained by IFN-γ ELISPOT, as previously described (24), with data imported into Prism 6 (GraphPad) for preparation of graphs.
Identification of cytokine-producing cells
DC cultures were stimulated overnight with SVLPs or TLR ligands, as described above, with the exception that the cells were analyzed. Cells were fixed and permeabilized as described above for confocal microscopy, 4 h after applying brefeldin to force accumulation of the cytokine to facilitate detection (24). The cytokine under study (TNF-α, IL-6, IL-1β, or IFN-α) was identified by the appropriate Ab specific for that cytokine, together with an Alexa Fluor 547– or Alexa Fluor 633–conjugated secondary Ab for detection.
SVLP-induced immune responses
Pigs were immunized three times intradermally with 100 μg/ml SVLPs in Ca2+/Mg2+-free PBS (without adjuvant), at five injection spots (200 μl per spot) behind the ear. All animals were given two booster immunizations, each after an additional 4 wk. At weeks 2 and 14, blood was collected to assess serum Ab levels by ELISA (25). Briefly, MaxiSorp plates (Nunc) were coated overnight at 4°C with 5 μg/ml SVLPs in carbonate-bicarbonate buffer (pH 9.6). After washing and blocking for 1 h at 37°C with Ca2+/Mg2+-free PBS containing 0.5% w/v BSA (Fluka) and 0.05% v/v Tween 20, serum samples diluted in Ca2+/Mg2+-free PBS/0.05% Tween 20 (Merck)/0.5% BSA (Fluka) were added, bound anti-SVLP Abs were detected with rabbit anti-swine Ig HRP (P0217; Dako), the reaction was developed using OPD (catalog number P-8787; Sigma) peroxidase substrate, and absorbance was read at 450 nm.
Specific lymphocyte responses were assessed by isolating PBMCs at 14 wk and plating as above for the IFN-γ ELISPOT assay. Following the incubation with SVLPs or mitogen controls, data were obtained as previously described (24).
All graphs were prepared using Prism 6 (GraphPad).
Statistical analyses
Analysis of likely colocalization used the colocalization algorithm of Imaris programs. For some calculations, the confocal microscopy images shown in the same figure were used prior to zooming to increase the number of cells used for the algorithmic analysis. In addition, five randomly selected fields were used to ensure that the total number of cells in the analysis was ≥100. This colocalization algorithm was used to calculate the Pearson coefficient for correlation of pixel colocalization within the gate defined by the ROIs of the two fluorescent channels (SVLP and cell marker) under investigation. An initial threshold value for the SVLP (green) channel (channel A) was set based on the values obtained with unlabeled SVLPs interacting with the DCs and DCs lacking the labeling for the intracellular structures under investigation. An ROI was selected for the SVLP channel, using this threshold to set the Mask Dataset for the masking channel. Automatic thresholding was then selected using the Imaris program algorithm developed by Costes and Lockett (26), based on the exclusion of intensity pairs that exhibit no correlation. The automatic threshold was controlled so that it did not include pixels excluded using the above thresholds obtained with unlabeled SVLPs and unlabeled cells. By these means, the thresholds for the mask were calculated, and the colocalization channel was built. From the latter, the statistics for image correlations were obtained. These are displayed as the Pearson coefficient of correlation in colocalized volume. The value shown in the figures is a median value obtained from the image displayed, with error bars calculated using this value together with that from three additional images.
The colocalization channel was also used to identify the percentage of SVLP+ cells showing colocalized voxels. For the graphic presentation of the percentage of SVLP+ cells showing colocalized voxels, as well as the percentage of SVLP+ cells within CD172alo and CD172ahi gates, the percentage of cells showing CD86 upregulation, and the production of the different cytokines assessed, replicate samples from three experiments were analyzed. This permitted statistical determination of p values calculated with an unpaired Student t test using Prism 6 (GraphPad), to generate median ± SE about the mean.
Results
Early events associated with DC endocytosis of SVLPs
Considering the hydrophobic nature of the lipopeptide PAMPs in the di- and triacylated SVLPs (Fig. 1A), exposure of PAMPs for TLR2 binding was first assessed in terms of the endocytic processing used by DCs (using MoDCs and blood DCs). Microscopy analyses initially showed efficient binding on ice during the first minute, with this binding of SVLPs increasing over a 10-min period (Fig. 2A shows MoDCs:SVLPs in green). Flow cytometry demonstrated maximum binding within the first 5-s observation period (5 s on ice plus 5 min washing) (data not shown). This remained stable for ≥60 min on ice, confirmed as surface binding through proteinase K sensitivity (data not shown), similar to that reported previously (16). Confocal microscopy also confirmed that labeled SVLPs were initially on the DC surface and colocalized with CTB, likely indicative of lipid rafts, as previously described for SVLPs with a 1,3-dipalmitoyl-sn-glycero-phosphatidylethanolamine core (PE-SVLPs) (16). This previous work described that MBCD treatment of DCs reduced SVLP interaction without influencing CTB binding. We have confirmed this and determined that, in the absence of lipid rafts caused by MBCD treatment, any colocalization of SVLPs with CTB-labeled areas was closer to random (Pearson coefficient in colocalized volume ≤+0.15, Fig. 2D, CTB+MBCD 0′).
SVLP binding to DCs and early events during endocytosis by DCs. (A) Prechilled MoDCs were incubated with Alexa Fluor 488–SVLPs (1 μg/ml; green; upper panel) or with Alexa Fluor 594–CTB binding cell surface GM1 molecules, for 10 min on ice. In order to facilitate viewing of the labeling, the images were cropped to remove as much of the gray background as possible and to bring the cells close together. This montage is shown by the black lines. (B) MoDCs (upper panels) or blood CD172a-sorted DCs (lower panel) were incubated with Alexa Fluor 488–SVLPs (green) for 60 min at 39°C. Then cells were labeled with Ab against CD14 (red) or CD172a (cyan) without fixation or permeabilization. CD14 and CD172a labeling was enhanced using the Surface Module of the Imaris analysis program (Imaris 8.0.2, Bitplane) to increase appreciation of the cell surface and shape. (C) Prechilled MoDCs were incubated with Alexa Fluor 488–SVLPs (green), Alexa Fluor 594–CTB (red), and Alexa Fluor 647–dextran (10,000 m.w., anionic; cyan) for 30 min on ice and then shifted to 39°C for 10 min. The Surface Module of Imaris software was applied to define cell surfaces from Nomarski interference images (colored white). The circle was zoomed and turned 180° for the middle panel (all three labels) and right panel (SVLP and CTB only). (D) Pearson coefficient of colocalized voxels for the likely colocalization of SVLPs with different markers of early endocytosis pathway components. This used the images, prior to zooming, shown in (C) (CTB and dextran [Dx] at 10′) and (E) (CTB and caveolin [Cav-1] at 30′), as well as SVLPs with CTB on ice before transfer to 39°C in the presence or absence of MBCD (CTB+MBCD 0′, CTB 0′), clathrin (Clath at 10′), and transferrin (Tf at 1′ and 10′); the images were obtained from five randomly selected fields to ensure a total of ≥100 cells from three replicates. The dotted line indicates the value for random localization (0). (E) As in (C), but with Alexa Fluor 488–SVLPs (green), Alexa Fluor 546–conjugated labeling of Cav-1 (red), and Alexa Fluor 647–CTB (cyan) for 30 min at 39°C. Blend image showing colocalized voxels for SVLP with CTB (white, with white arrowheads) or caveolin (yellow, with yellow arrow) (left panel). Representation of the cell in gray shading obtained using the Surface Module of the Imaris program applied to the Trans channel used for displaying a Nomarski interference–like image (right panel). All representative images are as described in 2Materials and Methods.
SVLP binding to DCs and early events during endocytosis by DCs. (A) Prechilled MoDCs were incubated with Alexa Fluor 488–SVLPs (1 μg/ml; green; upper panel) or with Alexa Fluor 594–CTB binding cell surface GM1 molecules, for 10 min on ice. In order to facilitate viewing of the labeling, the images were cropped to remove as much of the gray background as possible and to bring the cells close together. This montage is shown by the black lines. (B) MoDCs (upper panels) or blood CD172a-sorted DCs (lower panel) were incubated with Alexa Fluor 488–SVLPs (green) for 60 min at 39°C. Then cells were labeled with Ab against CD14 (red) or CD172a (cyan) without fixation or permeabilization. CD14 and CD172a labeling was enhanced using the Surface Module of the Imaris analysis program (Imaris 8.0.2, Bitplane) to increase appreciation of the cell surface and shape. (C) Prechilled MoDCs were incubated with Alexa Fluor 488–SVLPs (green), Alexa Fluor 594–CTB (red), and Alexa Fluor 647–dextran (10,000 m.w., anionic; cyan) for 30 min on ice and then shifted to 39°C for 10 min. The Surface Module of Imaris software was applied to define cell surfaces from Nomarski interference images (colored white). The circle was zoomed and turned 180° for the middle panel (all three labels) and right panel (SVLP and CTB only). (D) Pearson coefficient of colocalized voxels for the likely colocalization of SVLPs with different markers of early endocytosis pathway components. This used the images, prior to zooming, shown in (C) (CTB and dextran [Dx] at 10′) and (E) (CTB and caveolin [Cav-1] at 30′), as well as SVLPs with CTB on ice before transfer to 39°C in the presence or absence of MBCD (CTB+MBCD 0′, CTB 0′), clathrin (Clath at 10′), and transferrin (Tf at 1′ and 10′); the images were obtained from five randomly selected fields to ensure a total of ≥100 cells from three replicates. The dotted line indicates the value for random localization (0). (E) As in (C), but with Alexa Fluor 488–SVLPs (green), Alexa Fluor 546–conjugated labeling of Cav-1 (red), and Alexa Fluor 647–CTB (cyan) for 30 min at 39°C. Blend image showing colocalized voxels for SVLP with CTB (white, with white arrowheads) or caveolin (yellow, with yellow arrow) (left panel). Representation of the cell in gray shading obtained using the Surface Module of the Imaris program applied to the Trans channel used for displaying a Nomarski interference–like image (right panel). All representative images are as described in 2Materials and Methods.
Further analysis of the endocytic process used by DCs with SVLPs was performed with MoDCs and blood DCs, prepared as described in 2Materials and Methods. This used an initial 30 min on ice for binding, followed by washing to remove unbound material before shifting to 39°C. Thereby, the kinetics of internalization (increase in proteinase K resistance with time) was similar to that reported previously for PE-SVLPs (16) (data not shown). Following the 39°C shift, SVLPs rapidly appeared on the cytoplasmic side of the plasma membrane (Fig. 2B, 2C, 2E). Internalization was observed with CD172a+CD14+ MoDCs, CD172ahi blood DCs, CD172alo blood DCs, and CD172ahi CD14+ monocytes (Fig. 2B). The SVLPs carrying TLR2 ligands in their hydrophobic cores were not interacting with cell surface TLR2; the concomitant presence of anti-TLR2 Ab did not alter the image shown in Fig. 2B. Prominent colocalization and polarization with CTB was observed at the time of shifting to 39°C (Fig. 2D, CTB 0′) and remained clear during the first 10 min (Fig. 2C). This was observed in the majority of SVLP+ cells (Supplemental Fig. 2A, CTB 10′), and proved significant with a Pearson coefficient of +0.65 (ranging from greater than +0.4 to +0.8 for different images) (Fig. 2D, CTB 10′). Such observations also relate to our previous findings using PE-SVLPs (16).
Selection of DC endocytic pathways for SVLP internalization
Association with clathrin was rare (Supplemental Fig. 2A, Clath 10′). Any indication of SVLP interaction with clathrin+ structures was noted primarily between 5 and 10 min. This was closer to a random association, as indicated by the Pearson coefficient ≤+0.2, and often was closer to 0 (Fig. 2D, Clath 10′). Such observations implicated a minor involvement of clathrin-mediated endocytosis for DC uptake of SVLPs. The slow kinetics of SVLP internalization further signified dominant clathrin-independent endocytosis.
SVLP association with transferrin+ structures was also rare and detectable only between 1 and 10 min (Supplemental Fig. 2A, Tf 1′, Tf 10′). As with clathrin, the Pearson coefficient for colocalization of SVLPs with transferrin was closer to a random association (+0.1 to +0.22) (Fig. 2D, Tf 1′, Tf 10′). These results implied that sorting/recycling endosomal compartments were not major contributors to DC processing of SVLPs.
SVLPs continued to colocalize with CTB close to the DC surface from 10 to 30 min (Fig. 2E, colocalized voxels are white; white arrowheads show SVLP/CTB colocalization). As observed within the first 10 min (see Fig. 2C), the colocalization of SVLPs with CTB continued to be significant, with the Pearson coefficient ranging from greater than +0.2 to +0.6 for different images taken after 30 min (Fig. 2D, CTB 30′). It was interesting that the median value and range obtained from different images for this Pearson coefficient at 30 min were lower than those observed after 10 min (Fig. 2D, CTB 10′), as was the percentage of cells showing potentially colocalized voxels (Supplemental Fig. 2A, CTB 10′ versus CTB 30′).
Caveolin+ structures associating with SVLPs were also noted between 10 and 30 min (Fig. 2E, colocalized voxels are white; yellow arrows show SVLP/caveolin-1 colocalization). Colocalized voxels with caveolin were observed in more cells than with clathrin or transferrin, but in fewer cells than with CTB (Supplemental Fig. 2A, Cav-1 30′). In contrast with the clearly significant colocalization of SVLPs with CTB, the Pearson coefficient for the association of SVLPs with caveolin-1 was low, ranging from +0.2 to less than +0.3 (Fig. 2D, Cav-1 30′).
Non-di/non-triacylated SVLP internalization by DCs was reported to use a prominent macropinocytosis (16). Yet, during the initial 10-min period when SVLP/CTB colocalization was high, neither SVLPs nor CTB associated with dextran-containing structures (Fig. 2C). The Pearson coefficient for SVLPs with dextran was ≤0.2 at 10 min, with a similar range in values as for SVLPs with clathrin (Fig. 2D, Dx 10′), and the percentage of cells showing apparently colocalized voxels was low (Supplemental Fig. 2A, Dx 10′). This demonstrated that dextran binding was absent from the lipid rafts associated with SVLPs and CTB.
Assessment of DC macropinocytosis of SVLPs
The results shown in Fig. 2C and 2D demonstrated a lack of SVLP colocalization with dextran during the first 10 min of endocytosis, and SVLPs did not appear to associate with the macropinosomes known to accumulate dextran molecules. Accordingly, an additional labeling was performed using dextran in the red channel and MHCII in the blue channel. During the first 10–20 min of endocytosis, the internalized SVLPs remained more polarized than dextran, with the latter becoming more disperses within the cell (Fig. 3A, 10, 20 min). This related to the results with CTB/dextran, showing that dextran was not associated with the CTB+ lipid raft structures used for initiating endocytosis of SVLPs (Fig. 2C, 2D). Moreover, the Pearson coefficient for SVLP association with dextran at 10 min (with the SVLP/dextran/MHCII concomitant analysis) was ∼0 (Fig. 3B, Dx 10′); again, this confirmed that the SVLP/dextran results obtained with the SVLP/CTB/dextran labeling (Fig. 2C, 2D) showed a more random association of dextran with SVLPs. By 20 min, the percentage of cells with apparently colocalized SVLP/dextran voxels had increased slightly over that observed at 10 min (Supplemental Fig. 2A, Dx 10′, Dx 20′), while the Pearson coefficient had also increased; this remained less than or equal to +0.2 because some images gave a coefficient close to 0 (Fig. 3B, Dx 20′).
SVLP uptake kinetics by DCs in terms of association with different vesicular structures of endocytic pathways. Prechilled MoDCs were incubated with 1 μg/ml Alexa Fluor 488–SVLPs together with live cell markers (dextran, ER-Tracker) for 30 min on ice and then shifted to and incubated at 39°C for the times shown below. Thereafter, the cells were labeled for internal MHCII or for microtubules or with DiI (after fixation and permeabilization, as described in 2Materials and Methods, to label preferentially internal structures). (A) Alexa Fluor 488–SVLPs (green) with Alexa Fluor 546–dextran (10,000 m.w., anionic, red) and Alexa Fluor 633–conjugate labeling of MHCII (cyan) after 10–30 min at 39°C. Images are three-dimensional plots with the Surface Module of Imaris software applied to define cell surfaces from Nomarski interference images, colored different shades of gray; all are blend images with the exception of the lower left panel, which contains a MIP image. (B) Pearson coefficient of colocalized voxels for the likely colocalization of SVLPs with dextran (Dx) and MHCII, using images from the experiments shown in (A), as well as additional images to provide the kinetics of association from 10 min to 4 h and images obtained from a total of three replicates. The dotted line indicates the value for random localization (0). (C) Alexa Fluor 488–SVLPs (green) together with Alexa Fluor 546–conjugated labeling of α-tubulin (red) and Alexa Fluor 633–conjugated labeling of MHCII (cyan) at 30 min after the temperature shift; the blend three-dimensional plot was prepared with the Filament Module of the Imaris analysis software applied to the microtubule (red) staining. (D) Pearson coefficient of colocalized voxels for the likely colocalization of SVLPs with ER-Tracker (ERtr) after 10, 20, and 60 min at 39°C (images not shown), with CTB after 60 min at 39°C (Fig. 4A), and with lipophilic dye DiI after 60 min at 39°C (Fig. 4A), using the images mentioned along with three additional images. The dotted line indicates the value for random localization (0). All representative images in (A) and (C) are as described in 2Materials and Methods. The calculation of the Pearson coefficients in (B) and (D) was obtained with images prior to zooming and images obtained from five randomly selected fields to ensure a total number of ≥100 cells from three replicates.
SVLP uptake kinetics by DCs in terms of association with different vesicular structures of endocytic pathways. Prechilled MoDCs were incubated with 1 μg/ml Alexa Fluor 488–SVLPs together with live cell markers (dextran, ER-Tracker) for 30 min on ice and then shifted to and incubated at 39°C for the times shown below. Thereafter, the cells were labeled for internal MHCII or for microtubules or with DiI (after fixation and permeabilization, as described in 2Materials and Methods, to label preferentially internal structures). (A) Alexa Fluor 488–SVLPs (green) with Alexa Fluor 546–dextran (10,000 m.w., anionic, red) and Alexa Fluor 633–conjugate labeling of MHCII (cyan) after 10–30 min at 39°C. Images are three-dimensional plots with the Surface Module of Imaris software applied to define cell surfaces from Nomarski interference images, colored different shades of gray; all are blend images with the exception of the lower left panel, which contains a MIP image. (B) Pearson coefficient of colocalized voxels for the likely colocalization of SVLPs with dextran (Dx) and MHCII, using images from the experiments shown in (A), as well as additional images to provide the kinetics of association from 10 min to 4 h and images obtained from a total of three replicates. The dotted line indicates the value for random localization (0). (C) Alexa Fluor 488–SVLPs (green) together with Alexa Fluor 546–conjugated labeling of α-tubulin (red) and Alexa Fluor 633–conjugated labeling of MHCII (cyan) at 30 min after the temperature shift; the blend three-dimensional plot was prepared with the Filament Module of the Imaris analysis software applied to the microtubule (red) staining. (D) Pearson coefficient of colocalized voxels for the likely colocalization of SVLPs with ER-Tracker (ERtr) after 10, 20, and 60 min at 39°C (images not shown), with CTB after 60 min at 39°C (Fig. 4A), and with lipophilic dye DiI after 60 min at 39°C (Fig. 4A), using the images mentioned along with three additional images. The dotted line indicates the value for random localization (0). All representative images in (A) and (C) are as described in 2Materials and Methods. The calculation of the Pearson coefficients in (B) and (D) was obtained with images prior to zooming and images obtained from five randomly selected fields to ensure a total number of ≥100 cells from three replicates.
During this initial 20 min of endocytosis, SVLP+ vesicles also remained more peripheral than internal MHCII+ structures (Fig. 3A, 10, 20 min). The percentage of cells with apparently colocalized voxels was also particularly low (Supplemental Fig. 2A, MHC 10′, MHC 20′). Although SVLPs and MHCII+ structures associated with microtubules, SVLP+ vesicles were associated more with microtubule extensions compared with MHCII+ structures, which were closer to the microtubule organizing center (Fig. 3C). Although some of the MHCII+ structures did appear more peripheral, they were still not associating with SVLPs (Pearson coefficient ≤0, Fig. 3B).
Although internalization of SVLPs did not follow the route taken by endocytosed dextran, the important role of macropinocytosis did become evident. From 20 min to 4 h, there was a progressive increase in the percentage of cells with apparently colocalized voxels (Supplemental Fig. 2A, Dx 20′, Dx 30′, Dx 60′, Dx 4 h). Importantly, by 30 min, SVLP+ vesicles were clearly colocalized with dextran (Fig. 3A, 30 min, cell surface in white, colocalization in yellow). The Pearson coefficient for SVLP/dextran colocalization had increased to approximately +0.4 by this 30-min time point, although there still was some variation depending on the experiment (Fig. 3B, Dx 30′). Continued observation in different experiments confirmed this colocalization, with further increases in the Pearson coefficient to between +0.5 and +0.7 by 60 min and to between +0.7 and +0.8 by 4 h (Fig. 3B, Dx 60′, Dx 4 h). Moreover, there was a very narrow variation in the values obtained from different images and experiments at this latter time point. During this time period, the Pearson coefficient for SVLP with MHCII+ structures varied from −0.2 to +0.15, with no clear indication of an increase over time (Fig. 3B, MHCII 60′, MHCII 4 h), as reflected in the low percentage of cells positive for colocalized voxels (Supplemental Fig. 2A, MHC 60′, MHC 4 h).
ER retrograde transport and vesicular trafficking during endocytosis of SVLPs
The results in Fig. 3 show the involvement of macropinosomes during the endocytosis of SVLPs. Yet, the initial colocalization with CTB rather than dextran (Fig. 2C) may also reflect retrograde transport into the ER together with CTB, despite the low levels of SVLP colocalization with transferrin (closer to random; Fig. 2D), suggesting a lack of association with recycling endosomes. Certainly, not all SVLP+ structures remained associated with CTB, or with dextran at later time points (Supplemental Fig. 2A), suggesting that more than one endocytic route is used by the cells. There was a clear reduction in the Pearson coefficient for SVLP/CTB colocalization from 10 to 30 min (Fig. 2D), although this may depend on the experiment; analysis with SVLP/CTB/DiI labeling gave a high Pearson coefficient for SVLP/CTB colocalization as late as 60 min (Fig. 3D, CTB 60′). Accordingly, SVLP colocalization with ER-Tracker+ structures was assessed to determine whether an early association with the ER could be defined.
An apparent colocalization of SVLPs with ER-Tracker+ structures was observable between 20 and 60 min (data not shown). However, the majority of SVLP+ structures were not associated with ER-Tracker, and the percentage of cells showing apparently colocalized voxels remained <40% (Supplemental Fig. 2A, ERtr 10′, ERtr 20′, ERtr 60′). Indeed, at 10 and 20 min, the Pearson coefficient for colocalization of SVLPs with ER-Tracker was approximately +0.2 (Fig. 3D, ERtr 10′, ERtr 20′). Only at 60 min was there an increase to between +0.3 and less than +0.4 (Fig. 3D, ERtr 60′); the percentage of cells positive for apparently colocalized voxels also increased at 60 min (Supplemental Fig. 2A, ERtr 60′). Although these results would reflect a relative infrequency of SVLP delivery into ER-Tracker+ structures, such events could not be discounted.
Applying the lipophilic dye DiI to p-formaldehyde–fixed cells favored DiI labeling of intracellular structures rather than the cell surface, which revealed a time-dependent association of SVLPs with DiI+ structures. At the aforementioned 60-min time point showing SVLP/CTB colocalization, SVLPs also associated with vesicular bodies, as reflected by colocalization with DiI (Fig. 4A) with a Pearson coefficient of +0.7 (Fig. 3D, DiI 60′). SVLPs were noted in small and more complex membrane-bound structures within the cell cytoplasm (data not shown). This colocalization with intracellular DiI+ structures appeared to be initiated as early as 5 min, and ∼40% of cells displayed apparently colocalized voxels (Supplemental Fig. 2B, DiI 5′). Nonetheless, the Pearson coefficient showed these events to be random, remaining close to 0 (Fig. 4B, DiI 5′). SVLP+ structures did colocalize with DiI+ bodies, particular larger, irregular structures, in a time-dependent manner (Fig. 4A, upper panel, cell surface in grayish white, colocalized voxels overlay in white). This time-dependent association was confirmed by an increase in the Pearson coefficient for SVLP/DiI colocalization to ≥0.6 by 90 min (Fig. 4B), as reflected by an increase in the percentage of cells with colocalized voxels (Supplemental Fig. 2B).
Endocytosis of SVLPs by DCs relative to potential retrograde transport or association with the endolysosomal pathway. Prechilled MoDCs were incubated with 1 μg/ml Alexa Fluor 488–SVLPs and live cell markers (CTB, DQ-OVA) for 30 min on ice and then shifted to and incubated at 39°C for the times shown below, followed by labeling with Ab (EEA-1) or with DiI (after fixation and permeabilization, as described in 2Materials and Methods). (A) Alexa Fluor 488–SVLPs (green) with Alexa Fluor 647–CTB (cyan) at 60 min after temperature shift followed by p-formaldehyde fixation to allow DiI (red) labeling of intracellular membranes; the image is selected to show DiI labeling of large membranous structures. Images are three-dimensional plots with the Surface Module of the Imaris analysis software applied to the cell surface defined by the Nomarski interference image of cells (colored grayish white). Colocalization of SVLPs with DiI or CTB are shown separately, as colocalized voxels (white; right panels). (B) Pearson coefficient of colocalized voxels for the likely colocalization of SVLPs with DiI using images from three replicates at each time point, providing the kinetics of association from 5 to 90 min. The dotted line indicates the value for random localization (0). (C) Alexa Fluor 488–SVLPs (green) together with self-quenched DQ-BODIPY-TR-OVA (DQ-OVA, red) at 30 min after the temperature shift. MIP three-dimensional (3D) plot showing colocalization (yellow inclusions) of SVLP (green) with the dequenched albumin (red) (upper panel). MIP 3D plot showing only colocalized voxels (white) from the upper panel (lower panel). (D) Self-quenched DQ–Alexa Fluor 488–SVLPs (DQ-SVLP, green) together with Alexa Fluor 546–conjugated labeling of EEA-1 (red) at 2 h after the temperature shift. Blend 3D image of dequenched Alexa Fluor 488–SVLPs (green) and labeled EEA-1 (red), with the Surface Module of Imaris analysis software applied to the cell surface defined by the Nomarski interference image of cells (shades of gray) (top panel). MIP 3D plot of the top panel showing colocalization (yellow) of dequenched SVLP (green) with EEA-1 (red) (middle panel). Same MIP 3D plot as in the middle panel showing only the colocalized voxels (white) (bottom panel). All representative images in (A)–(D) are as described in 2Materials and Methods. (E) Pearson coefficient of colocalized voxels for the likely colocalization of SVLPs with dequenched DQ-OVA and of dequenched SVLPs with EEA-1, using the images shown in (C) and (D), together with three replicates. The dotted line indicates the value for random localization (0). (F) Pearson coefficient of colocalized voxels for the likely colocalization of SVLPs with LysoTracker after 30 min or after 1, 2, 4, 6, or 8 h at 39°C (images not shown), using three replicate images for each time point. The dotted line indicates the value for random localization (0). The calculation of the Pearson coefficients in (B), (E), and (F) was obtained with images prior to zooming and with images obtained from five randomly selected fields to ensure a total number of ≥100 cells from three replicates.
Endocytosis of SVLPs by DCs relative to potential retrograde transport or association with the endolysosomal pathway. Prechilled MoDCs were incubated with 1 μg/ml Alexa Fluor 488–SVLPs and live cell markers (CTB, DQ-OVA) for 30 min on ice and then shifted to and incubated at 39°C for the times shown below, followed by labeling with Ab (EEA-1) or with DiI (after fixation and permeabilization, as described in 2Materials and Methods). (A) Alexa Fluor 488–SVLPs (green) with Alexa Fluor 647–CTB (cyan) at 60 min after temperature shift followed by p-formaldehyde fixation to allow DiI (red) labeling of intracellular membranes; the image is selected to show DiI labeling of large membranous structures. Images are three-dimensional plots with the Surface Module of the Imaris analysis software applied to the cell surface defined by the Nomarski interference image of cells (colored grayish white). Colocalization of SVLPs with DiI or CTB are shown separately, as colocalized voxels (white; right panels). (B) Pearson coefficient of colocalized voxels for the likely colocalization of SVLPs with DiI using images from three replicates at each time point, providing the kinetics of association from 5 to 90 min. The dotted line indicates the value for random localization (0). (C) Alexa Fluor 488–SVLPs (green) together with self-quenched DQ-BODIPY-TR-OVA (DQ-OVA, red) at 30 min after the temperature shift. MIP three-dimensional (3D) plot showing colocalization (yellow inclusions) of SVLP (green) with the dequenched albumin (red) (upper panel). MIP 3D plot showing only colocalized voxels (white) from the upper panel (lower panel). (D) Self-quenched DQ–Alexa Fluor 488–SVLPs (DQ-SVLP, green) together with Alexa Fluor 546–conjugated labeling of EEA-1 (red) at 2 h after the temperature shift. Blend 3D image of dequenched Alexa Fluor 488–SVLPs (green) and labeled EEA-1 (red), with the Surface Module of Imaris analysis software applied to the cell surface defined by the Nomarski interference image of cells (shades of gray) (top panel). MIP 3D plot of the top panel showing colocalization (yellow) of dequenched SVLP (green) with EEA-1 (red) (middle panel). Same MIP 3D plot as in the middle panel showing only the colocalized voxels (white) (bottom panel). All representative images in (A)–(D) are as described in 2Materials and Methods. (E) Pearson coefficient of colocalized voxels for the likely colocalization of SVLPs with dequenched DQ-OVA and of dequenched SVLPs with EEA-1, using the images shown in (C) and (D), together with three replicates. The dotted line indicates the value for random localization (0). (F) Pearson coefficient of colocalized voxels for the likely colocalization of SVLPs with LysoTracker after 30 min or after 1, 2, 4, 6, or 8 h at 39°C (images not shown), using three replicate images for each time point. The dotted line indicates the value for random localization (0). The calculation of the Pearson coefficients in (B), (E), and (F) was obtained with images prior to zooming and with images obtained from five randomly selected fields to ensure a total number of ≥100 cells from three replicates.
Endosomal and lysosomal associations with SVLPs
DQ-OVA was used to identify early endosomal activity because the DQ signal only becomes detectable upon endosomal dequenching of the fluorochrome label. The observed involvement of macropinosomes during SVLP endocytosis (Fig. 3A, 3B) did lead to colocalization of SVLPs with DQ-OVA (Fig. 4C, upper panel, colocalization in yellow; lower panel, colocalized voxels in white), with a Pearson coefficient of colocalization between +0.55 and +0.75 (Fig. 4E, DQ-OVA 30′). DQ dequenching indicated the initiation of processing/degradation that likely involves early endosomes interacting with macropinosomes. Indeed, SVLP+ structures were observed in association with early endosomal (EEA-1+) structures (Fig. 4D, top panel, cell surface in grayish white). Not all SVLP+ vesicles were EEA-1+ at any given moment, but the presence of detectable DQ-SVLPs (quenched in a similar manner to DQ-OVA and requiring degradation for visualization) indicates processing. Certainly, SVLP/EEA-1 colocalization was prominent by 2 h (Fig. 4D, middle panel, colocalization in yellow; bottom panel, colocalized voxels in white). The Pearson coefficient for SVLP/EEA-1 colocalization was lower than for SVLP/DQ-OVA colocalization, with a median value of +0.45 (Fig. 4E, EEA1 2 h); a similar relationship was observed for the percentage of cells showing apparently colocalized voxels (Supplemental Fig. 2A, DQ-OVA 30′, EEA-1 2 h). Although many dequenched SVLP+ structures were not associated with EEA-1+ vesicles, this may reflect the manner by which EEA-1+ vesicles interact with SVLP+ macropinosomes, as reported previously (16).
Involvement of early endosomes should result in endosomal maturation toward LysoTracker+ late endosomal- and lysosomal-like structures. During the first 30 min, only a minority of SVLP+ structures significantly colabeled with LysoTracker (data not shown), despite the fact that 30–45% of cells display apparently colocalized voxels (Supplemental Fig. 2A, LysoTracker, 30′); the Pearson coefficient was closer to random (+0 to +0.15, Fig. 4F, LysoTracker 30′). The Pearson coefficient gradually increased with time: +0.15 to +0.3 at 1 h, at least +0.65 at 2 h, and at least +0.8 by 4 h (Fig. 4F, LysoTracker 1, 2, 4 h). An increase in the percentage of cells with apparently colocalized voxels was also observed (Supplemental Fig. 2A, LysoTracker 1, 2, 4 h). SVLP–LysoTracker colocalization was still detectable at 6 and 8 h (data not shown), although the Pearson coefficient was now lower (median of +0.3 and +0.35, respectively) than that at 2 or 4 h (Fig. 4F, LysoTracker 6, 8 h). Moreover, there was a decrease in the percentage of cells with apparently colocalized voxels between 6 and 8 h (Supplemental Fig. 2A, LysoTracker 6, 8 h), which may have reflected increased degradation of SVLPs in LysoTracker+ structures with time.
Association of endocytosed SVLPs with intracellular TLR2
One possible consequence of the observed interaction between SVLP+ and LysoTracker+ compartments is degradation to expose the hydrophobic TLR2 ligands of P2C/P3C-SVLPs. Accordingly, DCs were allowed to process endocytosed Alexa Fluor 488–labeled P2C-SVLPs for 1 and 2 h. The cells were fixed and permeabilized as described in 2Materials and Methods (the fixation/permeabilization procedure destroyed the detectability of cell surface TLR2) to assess internal TLR2-containing compartments of the cells. At both time points, internal TLR2 was clearly discernible (Fig. 5A, SVLPs green, TLR2 red, cell surface light gray). Although certain SVLP+ vesicles were certainly distinguishable from TLR2+ structures (Fig. 5A, cell surface in light gray), a number of structures displayed colocalized SVLPs and TLR2 (Fig. 5A, colocalized voxels in yellow; colocalized voxels in white in bottom panel). The colocalization of SVLPs with internal TLR2 was time dependent and was observed with CD172alo and CD172ahi cells (data not shown). The Pearson coefficient for SVLP/TLR2 colocalization increased from a median of 0.4 at 1 h to a median of +0.65 (range +0.5 to +0.74) at 2 h (Fig. 5B). This latter value was maintained out to 6 h (Fig. 5B, 6 h). The percentage of cells with apparently colocalized SVLP/TLR2 voxels also increased from 1 to 6 h (Supplemental Fig. 2A, TLR2 1, 2, 6 h).
Association of SVLPs with internal TLR2 and intracellular production of cytokines. Blood CD172a-sorted DCs were incubated with 1 μg/ml Alexa Fluor 488–SVLPs for 1 to 6 h at 39°C and then fixed and permeabilized for labeling with Abs against TLR2, TNF-α, or IL-6, as described in 2Materials and Methods (the fixation and permeabilization facilitated preferential labeling of internal structures). (A) Alexa Fluor 488–SVLPs (green) relationship to internal TLR2+ structures (red) at 1 and 2 h at 39°C. Overlay of SVLP and TLR2 labels (left panels), with colocalization shown in yellow or white. Colocalized voxels alone (right panels) shown in yellow or white. Using either blend images (first and third rows) or MIP images (second and fourth rows), the cell is highlighted as shades of gray by applying the Surface Module of Imaris analysis software on the cell surface defined by the Nomarski interference image of cells; this allows appreciation of the intracellular nature of the staining. (B) Pearson coefficient of colocalized voxels for the likely colocalization of SVLPs with internal TLR2 over time, using images from the experiments shown in (C), as well as additional images, including a total of three replicates, to provide the kinetics of association from 1 to 6 h. The dotted line indicates the value for random localization (0). Calculation of Pearson coefficients was obtained from images prior to zooming and from images obtained from three randomly selected fields to ensure a total number of ≥100 cells from three replicates. (C) Microscopic analysis of CD172a-sorted blood cells following endocytosis of P2C-SVLP488 for 4 h at 39°C to determine the presence of TNF-α (cyan) in SVLP+ (green) cells. The cell is highlighted as shades of gray by applying the Surface Module of Imaris analysis software on the cell surface defined by the Nomarski interference image of cells; this allows appreciation of the intracellular nature of the staining. (D) As in (C) with P2C-SVLP488 (green), but showing staining for IL-6 (cyan). All representative images are as described in 2Materials and Methods.
Association of SVLPs with internal TLR2 and intracellular production of cytokines. Blood CD172a-sorted DCs were incubated with 1 μg/ml Alexa Fluor 488–SVLPs for 1 to 6 h at 39°C and then fixed and permeabilized for labeling with Abs against TLR2, TNF-α, or IL-6, as described in 2Materials and Methods (the fixation and permeabilization facilitated preferential labeling of internal structures). (A) Alexa Fluor 488–SVLPs (green) relationship to internal TLR2+ structures (red) at 1 and 2 h at 39°C. Overlay of SVLP and TLR2 labels (left panels), with colocalization shown in yellow or white. Colocalized voxels alone (right panels) shown in yellow or white. Using either blend images (first and third rows) or MIP images (second and fourth rows), the cell is highlighted as shades of gray by applying the Surface Module of Imaris analysis software on the cell surface defined by the Nomarski interference image of cells; this allows appreciation of the intracellular nature of the staining. (B) Pearson coefficient of colocalized voxels for the likely colocalization of SVLPs with internal TLR2 over time, using images from the experiments shown in (C), as well as additional images, including a total of three replicates, to provide the kinetics of association from 1 to 6 h. The dotted line indicates the value for random localization (0). Calculation of Pearson coefficients was obtained from images prior to zooming and from images obtained from three randomly selected fields to ensure a total number of ≥100 cells from three replicates. (C) Microscopic analysis of CD172a-sorted blood cells following endocytosis of P2C-SVLP488 for 4 h at 39°C to determine the presence of TNF-α (cyan) in SVLP+ (green) cells. The cell is highlighted as shades of gray by applying the Surface Module of Imaris analysis software on the cell surface defined by the Nomarski interference image of cells; this allows appreciation of the intracellular nature of the staining. (D) As in (C) with P2C-SVLP488 (green), but showing staining for IL-6 (cyan). All representative images are as described in 2Materials and Methods.
Intracellular cytokine induction
The observed interaction between P2C-SVLPs and internal TLR2 suggested pattern recognition receptor detection of TLR2 ligands in the hydrophobic SVLP core, likely exposed through DC processing. To determine whether these events (Fig. 5A) were of a functional consequence, induction of intracellular TNF-α and IL-6 was assessed with respect to the concomitant presence of detectable SVLPs. Microscopic analysis of blood CD172a+ cells showed that certain SVLP+ cells were also TNF+ (Fig. 5C); there was no distinction between CD172alo and CD172ahi cells. It was also interesting to note that the SVLP-containing structures were distinct from those in which TNF-α was detectable (Fig. 5C), and no colocalization was noted (Pearson coefficient ≤0; data not shown). This would relate to SVLPs interacting with TLR2+ vesicles (and/or being processed via MHC class II rich compartments [MIIC]), whereas the induced TNF-α would be present within the ER/Golgi.
The presence of IL-6 in SVLP+ cells was also assessed microscopically in blood CD172a+ cells following interaction with P2C-SVLPs. As with TNF-α, IL-6 induction was detectable in SVLP+ cells (Fig. 5D). Also similar to TNF-α induction, there was no distinction between CD172alo and CD172ahi cells, and SVLP+ structures were distinct from those carrying IL-6.
Cytokine secretion by SVLP-treated DCs
For the induced intracellular cytokines (Fig. 5) to be of consequence for immune development, they should be secreted. Assessment of SVLP-induced cytokine secretion initially used BMDCs because of their higher sensitivity to TLR ligands (Fig. 6A versus Fig. 6B). Blood CD172a+ cells were also used for evaluation of the cDC1, cDC2, and plasmacytoid DC (pDC) subsets. In addition to TNF-α and IL-6, IL-1β and IFN-α were assessed.
Capacity of SVLPs carrying TLR2 ligands in their hydrophobic di- or triacylated cores to induce particular innate immune cell cytokines relative to SVLP interaction with blood DCs and monocytes. (A) GM-CSF–derived BMDCs were treated with P3C-SVLPs (triacylated) or P2C-SVLPs (diacylated) in comparison with PE-SVLPs (no di- or triacylation), control TLR ligands, or no treatment (cells alone). Following overnight culture at 39°C, supernatants were collected and tested (ELISA) for IL-1β (upper panel) and IL-6 (lower panel). Bars represents means (± SE) of three replicate samples representing three replicate experiments. (B) As in (A), but using CD172a-sorted blood cells tested for IL-1β (upper panel) and IFN-α (lower panel). (C) As in (B), but using CD14-sorted (CD14−) blood cells. (D) Influence of TLR2 agonists on SVLP interaction with CD172a+ cells. The flow cytometry graphs and the bar charts together show the influence of P3C, P2C, or FSL-1 (10 μg/ml) on the interaction of P3C-SVLPs (1, 0.2 μg/ml), P2C-SVLPs (1, 0.2 μg/ml), and PE-SVLPs (1 μg/ml) with the cells. (E) CD172a-sorted blood cell interaction with different concentrations of Alexa Fluor 488–labeled P3C-SVLPs (PE-SVLPs with triacylated lipids; P3C-SVLP488) compared with Alexa Fluor 488–labeled PE-SVLPs (SVLP488). The study also compared the influence of additional free P3C lipopeptide (+P3C). The SVLP signal is shown relative to the level of CD172a expression. (F) Graphical presentation of the percentage cells from (D) in CD172alo gates compared with CD172ahi gates showing the presence of the SVLPs, relative to SVLP concentration and the presence of free P3C lipopeptide. Comparison is also shown for cells alone (cells) and P3C lipopeptide alone (P3C) (see Supplemental Fig. 3B for cells and P3C alone).
Capacity of SVLPs carrying TLR2 ligands in their hydrophobic di- or triacylated cores to induce particular innate immune cell cytokines relative to SVLP interaction with blood DCs and monocytes. (A) GM-CSF–derived BMDCs were treated with P3C-SVLPs (triacylated) or P2C-SVLPs (diacylated) in comparison with PE-SVLPs (no di- or triacylation), control TLR ligands, or no treatment (cells alone). Following overnight culture at 39°C, supernatants were collected and tested (ELISA) for IL-1β (upper panel) and IL-6 (lower panel). Bars represents means (± SE) of three replicate samples representing three replicate experiments. (B) As in (A), but using CD172a-sorted blood cells tested for IL-1β (upper panel) and IFN-α (lower panel). (C) As in (B), but using CD14-sorted (CD14−) blood cells. (D) Influence of TLR2 agonists on SVLP interaction with CD172a+ cells. The flow cytometry graphs and the bar charts together show the influence of P3C, P2C, or FSL-1 (10 μg/ml) on the interaction of P3C-SVLPs (1, 0.2 μg/ml), P2C-SVLPs (1, 0.2 μg/ml), and PE-SVLPs (1 μg/ml) with the cells. (E) CD172a-sorted blood cell interaction with different concentrations of Alexa Fluor 488–labeled P3C-SVLPs (PE-SVLPs with triacylated lipids; P3C-SVLP488) compared with Alexa Fluor 488–labeled PE-SVLPs (SVLP488). The study also compared the influence of additional free P3C lipopeptide (+P3C). The SVLP signal is shown relative to the level of CD172a expression. (F) Graphical presentation of the percentage cells from (D) in CD172alo gates compared with CD172ahi gates showing the presence of the SVLPs, relative to SVLP concentration and the presence of free P3C lipopeptide. Comparison is also shown for cells alone (cells) and P3C lipopeptide alone (P3C) (see Supplemental Fig. 3B for cells and P3C alone).
Di- and triacylated SVLPs induced IL-1β production by BMDCs and CD172a+ cells in a concentration-dependent manner (Fig. 6A, 6B, IL-β, P3C-SVLP, P2C-SVLP). It appeared that PE-SVLPs may also have induced IL-1β levels with blood CD172a+ cells, but this probably reflected the particularly low levels observed with the cells-alone control (Fig. 6B, SVLP, cells). Certainly, this was not observed with BMDCs (Fig. 6A, SVLP, cells). Although there was a slight increase in the IL-1β levels obtained from BMDCs when the PE-SVLP concentration increased from 0.5 to 2 μg/ml, this always remained lower than the levels obtained with the control cells. Moreover, the IL-1β levels obtained with PE-SVLPs and blood CD172a+ cells were always lower than those obtained with CpG (Fig. 6B, SVLP, CpG) and were independent of SVLP concentration. This contrasted with the clear capacity of P2C/P3C-SVLPs to induce high levels of IL-1β, reaching levels higher than those induced by the P3C lipopeptide control, in a concentration-dependent manner (Fig. 6A, 6B, P3C-SVLP, P2C-SVLP). Similar observations were made for TNF-α induction (data not shown).
When assessed further using CD14-sorted cells to separate monocytes from DCs, all treatments, including the P3C lipopeptide control, induced lower levels of IL-1β compared with BMDCs and blood CD172a+ cells (Fig. 6C, IL-1β). Nonetheless, P2C/P3C-SVLPs did induce IL-1β in a concentration-dependent manner, and the levels were similar to the P3C lipopeptide control at the higher concentrations of SVLPs (Fig. 6C). It should be noted that CD14 sorting leads to a loss of DCs within the CD14− population, which may have influenced the lower levels of IL-1β that were induced.
The microscopic detection of induced IL-6 (Fig. 5D) also led to secretion of the cytokine by CD172a+ cells (data not shown) and BMDCs (Fig. 6A, IL-6). Nonetheless, this was clearly influenced by SVLPs per se rather than the di- or triacylated core; thus, only BMDCs are shown (Fig. 6A, IL-6). Such SVLP-dependent activation of cells was also observed with triggering murine RAW-Blue cell SEAP. The transcription factors (NF-κB and AP-1) involved in this process were induced primarily by the SVLP structure rather than by the di- or triacylated core (Supplemental Fig. 3D). This suggested that factors other than TLR2 ligation by the TLR2 ligand PAMPs of the di- and triacylated SVLP cores were involved with IL-6 and SEAP induction. Nonetheless, di- and triacylated SVLPs induced higher levels of SEAP than did PE-SVLPs with murine RAW-Blue cells, although this was independent of concentration (Supplemental Fig. 3D). Moreover, low SVLP concentrations with BMDCs did show a clear benefit of di- and triacylated cores (Fig. 6A, IL-6). Although 1 μg/ml P2C-SVLPs was used to induce intracellular IL-6 detectable by microscopy (Fig. 5E) and cytokine production by BMDCs, this was a less effective concentration for SVLPs lacking TLR2 ligand PAMPs (Fig. 6A, IL-6).
With regard to IFN-α, there was little or no induction by SVLPs in BMDCs or in blood cells (Fig. 6A–C). This was not due to an inability of DCs to respond, as reflected by type I IFN–induced CpG-ODN. These results for IFN-α are dealt with in more detail below in the section titled “25DC and monocyte maturation induced by SVLPs.”
Assessment of surface TLR2 during SVLP interaction with DCs and monocytes
As mentioned above, it was considered that P2C-SVLPs were interacting with internal TLR2 rather than surface TLR2 because of the inability of anti-TLR2 Ab to block interaction of the SVLPs with, and their endocytosis by, DCs. This anti-TLR2 Ab was able to interfere with the binding of rhodamine-labeled P3C (data not shown). To confirm that P2C-SVLPs were not binding at the cell surface via their TLR2 ligands in the hydrophobic core, competition assays were performed between SVLPs, P2C-SVLPs, and P3C-SVLPs (1 μg/ml) and the TLR2 agonists P3C, P3C VacciGrade, Pam2CysSK4, and FSL-1 (10 μg/ml); competition of TLR2 agonists with PE-SVLPs (no TLR2 ligands in their cores) was used as a control. An example of the results obtained using homologous competition between P3C and P3C-SVLPs, as well as between P2C and P2C-SVLPs, compared with competition between P3C and PE-SVLPs is shown in Fig. 6D (similar results were obtained for all TLR2 agonists). A summary of the replicate data is also shown for all competitions. This demonstrated that none of the TLR2 agonists reduced the interaction of P2C-SVLPs or P3C-SVLPs with CD172+ cells. The figure shows the data for all CD172+ cells because similar results were obtained for CD172lo and CD172hi subsets (data not shown).
DC and monocyte maturation induced by SVLPs
Another possible consequence of the observed SVLP/TLR-2 interaction is cytokine induction leading to DC maturation. Although TLR2 ligation is not the primary pathway for IFN-α induction, it was considered important to assess this cytokine for potential maturation effects. Interestingly, low levels of IFN-α were induced in CD14− cells by P2C-SVLPs, but they were far inferior to CpG-ODN and even the free P3C lipopeptide (Fig. 6C, IFN-α). In contrast, neither PE-SVLPs nor P2C/P3C-SVLPs induced IFN-α production by BMDCs (Supplemental Fig. 3E). The main type I IFN inducer CpG-ODN also had no effect, confirming the absence of functionally active pDCs (19). Similar ineffectiveness of SVLPs was obtained with blood CD172a+ cells and CD14− cells, wherein pDCs were present as witnessed by CpG-ODN induction of IFN-α (Fig. 6B, 6C, IFN-α).
Despite the lack of IFN-α induction, the presence of TNF-α, IL-6, and IL-1β could influence DC maturation. Accordingly, the relationship between SVLP uptake by DCs and maturation in terms of CD86 expression was assessed. This again used CD172a (signal regulatory protein α)–sorted blood cells to obtain blood DCs and monocytes, together with flow cytometry gating to facilitate the distinction of CD172alo cells (dominated by the cDC1 subset) from CD172ahi cells (cDC2, dominated by monocytes) (Supplemental Fig. 3A). SVLPs carrying triacylated TLR2 ligand cores (P3C-SVLPs) were compared with PE-SVLPs (lack TLR2 ligands) over a 20-h period; free P3C lipopeptide was used as positive control.
As shown above (Fig. 2B), CD172alo and CD172hi cells endocytosed Alexa Fluor 488–SVLPs in a concentration-dependent manner (Fig. 6E, 6F), irrespective of a triacylated core (Fig. 6E; SVLP488, P3C-SVLP488). The signal was clearly due to the endocytosed Alexa Fluor 488–SVLPs, because unlabeled SVLPs did not provide such signals (Supplemental Fig. 3B). Similar levels of endocytosed Alexa Fluor 488–SVLPs and the percentage of cells positive for CD172alo and CD172hi cells were noted at higher SVLP concentrations (Fig. 6E, 6F), although a more prominent interaction with CD172hi cells was noted at lower SVLP concentrations (Fig. 6E, 6F). Free P3C induced an increase in PE-SVLP and P3C-SVLP association with the cells (Fig. 6E, 6F). This influence of free P3C was not due to an increase in the background signal (Supplemental Fig. 3B) but likely reflected the known capacity of P3C to activate DC/monocytes.
CD172alo and CD172hi cells responded to P3C-SVLPs by upregulating CD86 expression (Fig. 7A, P3C-SVLP488). They were also responsive to the P3C lipopeptide control (Fig. 7A, P3C). In contrast, PE-SVLPs lacking triacylated TLR2 ligand lipid cores did not induce notable CD86 upregulation (Fig. 7A, SVLP488). Admixing P3C with PE-SVLPs yielded a pattern of CD86 upregulation resembling that obtained with P3C alone and was independent of SVLP concentration (Fig. 7A, 7B, PE-SVLP488+P3C compared with P3C). In contrast, the capacity of P3C-SVLPs to induce CD86 upregulation in their own right was concentration dependent (Fig. 7A, 7B, P3C-SVLP488). When admixed with P3C, the latter again dominated the influence on CD86 expression (Fig. 7A, 7B, P3C-SVLP488+P3C).
Consequence of SVLP interaction with blood DCs and monocytes, in terms of DC maturation and induction of adaptive immune responses in vitro and in vivo. (A) Expression of CD86 relative to the level of CD172a expression on CD172a-sorted blood cells after 20 h (at 39°C) of treatment with different concentrations (each column) of Alexa Fluor 488–labeled PE-SVLPs (SVLP) or Alexa Fluor 488–labeled P3C-SVLPs (P3C-SVLP488). Comparison was also made on the influence of free P3C lipopeptide (P3C and +P3C). (B) Graphical presentation of the percentage cells from (A) in the CD172alo gates compared with the CD172ahi gates showing CD86 upregulation relative to SVLP concentration and the presence of free P3C lipopeptide. Comparison is also shown for cells alone (cells) and P3C lipopeptide alone (P3C). (C) Comparison of CD86 expression on CD172a-sorted blood cells with respect to Alexa Fluor 488–labeled SVLP (SVLP488; P3C-SVLP488) association, in the presence or absence of free P3C lipopeptide (P3C and + P3C). All representative images in (A)–(C) were prepared as described in 2Materials and Methods. (D) PBMCs isolated from two naive pigs were restimulated with different concentrations of P2C-SVLPs and incubated for 48 h at 39°C. IFN-γ secreting cells were analyzed by ELISPOT (24). The graph is from a single experiment and is representative of three experiments. (E) Animals were immunized with P2C-SVLPs; blood samples were collected after 2 and 14 wk for the preparation of sera, and 14 wk postimmunization for preparation of PBMCs. Anti-SVLP Ab activity was assessed by ELISA (25) using plates coated with homologous SVLPs as Ag (left panel). Fresh PBMCs restimulated with different concentrations of homologous SVLPs were incubated for 48 h at 39°C; IFN-γ SC were analyzed by the above ELISPOT (right panel). All representative images shown were prepared as described in 2Materials and Methods.
Consequence of SVLP interaction with blood DCs and monocytes, in terms of DC maturation and induction of adaptive immune responses in vitro and in vivo. (A) Expression of CD86 relative to the level of CD172a expression on CD172a-sorted blood cells after 20 h (at 39°C) of treatment with different concentrations (each column) of Alexa Fluor 488–labeled PE-SVLPs (SVLP) or Alexa Fluor 488–labeled P3C-SVLPs (P3C-SVLP488). Comparison was also made on the influence of free P3C lipopeptide (P3C and +P3C). (B) Graphical presentation of the percentage cells from (A) in the CD172alo gates compared with the CD172ahi gates showing CD86 upregulation relative to SVLP concentration and the presence of free P3C lipopeptide. Comparison is also shown for cells alone (cells) and P3C lipopeptide alone (P3C). (C) Comparison of CD86 expression on CD172a-sorted blood cells with respect to Alexa Fluor 488–labeled SVLP (SVLP488; P3C-SVLP488) association, in the presence or absence of free P3C lipopeptide (P3C and + P3C). All representative images in (A)–(C) were prepared as described in 2Materials and Methods. (D) PBMCs isolated from two naive pigs were restimulated with different concentrations of P2C-SVLPs and incubated for 48 h at 39°C. IFN-γ secreting cells were analyzed by ELISPOT (24). The graph is from a single experiment and is representative of three experiments. (E) Animals were immunized with P2C-SVLPs; blood samples were collected after 2 and 14 wk for the preparation of sera, and 14 wk postimmunization for preparation of PBMCs. Anti-SVLP Ab activity was assessed by ELISA (25) using plates coated with homologous SVLPs as Ag (left panel). Fresh PBMCs restimulated with different concentrations of homologous SVLPs were incubated for 48 h at 39°C; IFN-γ SC were analyzed by the above ELISPOT (right panel). All representative images shown were prepared as described in 2Materials and Methods.
Interestingly, CD86 upregulation induced by P3C-SVLPs was more notable on CD172alo cells (Fig. 7A, 7B, P3C-SVLP488); up to 35% of these cells were CD86+, which was dependent on the P3C-SVLP concentration employed (Fig. 7B CD172lo, P3C-SVLP488). CD172ahi cells, which are dominated by monocytes, exhibited no clearly detectable CD86 upregulation in response to P3C-SVLPs (Fig. 7A, 7B, P3C-SVLP488); this was similar to the cell control values (Fig. 7B, CD172hi: cells) irrespective of P3C-SVLP concentration. Admixed free P3C increased CD86 expression levels to those obtained with the P3C-alone control (Fig. 7A, 7B, SVLP488+P3C, P3C-SVLP488+P3C).
Therefore, the relationship between CD86 upregulation and SVLP uptake was assessed. As expected from Fig. 7A, CD86 expression on cells interacting with PE-SVLPs (Fig. 7C, SVLP488) was similar to the cell control (Supplemental Fig. 3C, Cells). Also, P3C-induced CD86 was again independent of SVLP association (Supplemental Fig. 3C, P3C compared with Fig. 7C, SVLP488+P3C). With P3C-SVLPs, lower SVLP concentrations resulted in a small proportion of SVLP− cells expressing CD86 (Fig. 7C, P3C-SVLP488). The majority of CD86+ cells were SVLP+, reflecting an association between SVLP interaction and CD86 upregulation. Nonetheless, SVLP uptake by DCs did not guarantee CD86 expression; most SVLP+ cells remained CD86−, and admixed P3C did not increase this beyond the levels observed with P3C alone (Fig. 7C, P3C-SVLP488+P3C).
Cell viability
One possible explanation for a selective expression of CD86 by only certain SVLP+ cells was a loss of viability following uptake of SVLPs. Accordingly, CD172a-sorted blood cells treated with PE-SVLPs and P3C-SVLPs were assessed after 16 h at 39°C for perturbed plasma membrane, detected in terms of cells becoming annexin V–FITC+. Related to previous observations on MoDCs (16), SVLPs did not induce annexin V labeling (data not shown). Only when treated with SVLPs admixed with P3C were increased numbers of annexin V+ cells noted, primarily with CD172ahi cells, but they were still a minor population (data not shown). Thus, the observed influence of SVLPs on CD86 expression (Fig. 7A, 7B) was not due to induced cell death.
Adaptive immune response induction
The consequence of P2C-SVLP–dependent DC modulation on the adaptive immune compartment was initially assessed using PBMCs from naive donor pigs. P2C-SVLPs efficiently induced IFN-γ–producing cells, as assessed by ELISPOT assay (24), in a concentration-dependent manner (Fig. 7D). In contrast, PE-SVLPs, which lack a di- or triacylated core carrying TLR2-PAMPs, did not induce any detectable immune response in the absence of adjuvant (data not shown). This induction of T lymphocyte activity from naive donor pigs was observed primarily with Th lymphocytes, although a low-level response was detected with CD4−CD8hi CTLs (data not shown).
Accordingly, five animals were vaccinated with P2C-SVLPs, in the absence of adjuvant, to assess effector/memory responses. Blood samples taken at 2 and 14 wk demonstrated induction of a specific Ab response maintained over the 14-wk observation period (Fig. 7E, left panel). At 14 wk, SVLP-specific T lymphocyte activity, restimulation with homologous SVLPs in an ELISPOT assay, was also demonstrable; restimulation of PBMCs to produce IFN-γ was dependent on SVLP concentration (Fig. 7E, right panel). As with the in vitro stimulation of cells from naive donor pigs, this was observed primarily with Th lymphocytes; a low-level response detected with CD4−CD8hi CTLs was no different from that observed with cells from the naive donor animals (data not shown).
Discussion
Maintenance of immune homeostasis and induction of adaptive immune responses are critical roles for DCs, providing effective defense against pathogens and efficacy of vaccine delivery. DCs possess receptors for endocytosis and pattern recognition receptors detecting PAMPs and DAMPs, including the plasma membrane TLRs. Cell surface TLR2 heterodimers recognize structural components close to pathogen surfaces, including TA and PGN (5, 6), as well as zymosan (13, 27). TAs are particularly reactive ligands for DC surface TLR2. Synthetic di- and triacylated lipopeptides, as well as lipopeptides isolated from bacterial cell walls, are also potent ligands for surface TLR2 heterodimers, leading to internalization and cell activation (8, 12, 28), hence their application as adjuvants (29–31). Yet, the situation with intact bacteria does not relate to these latter situations. The bacterial cell wall structure may facilitate interaction of TAs for binding bacteria and fungi to surface TLR2 (14). In contrast, the hydrophobicity of lipoproteins/lipopeptides results in embedment deeper within bacterial cell walls (15). Although this would require degradation to expose the PAMPs, bacteria produce extracellular vesicles potentially with outer membrane–linked lipoproteins on their exterior face (15). These could promote lipopeptide-mediated activation of DC surface TLR2s. Such extracellular vesicles may also acquire outer membrane–linked LTA for interaction with the surface TLR2s (32).
Ligation of surface TLR2s can certainly lead to internalization (7–13, 33), but there is a discrepancy concerning intracellular compartmentalization (8, 10–12). Association of TLR2 was reported within the Golgi and dextran+ and LysoTracker+ structures (8, 12). TLR2 within the Golgi/ER may reflect retrograde transport from early endosomes, as well as synthesis of cell surface material; TLR2 associated with dextran or LysoTracker would reflect macropinosomes or late endosomes/lysosomes. Nonetheless, these studies were initiated by interaction of TLR2 ligands with cell surface receptors. The internal TLR2 was resultant upon ligation leading to receptor internalization, with TLR2 being present with its ligand (7–12). There was no evidence that internalized surface TLR2 was functionally active within the cell rather than just reflecting endocytic processing. Moreover, there were no reports on DCs; as the primary APC type (34), they are quite distinctive from keratinocytes, macrophage cell lines, and transfected HEK cells (35, 36) in their functionality subsequent to endocytic processing.
Accordingly, we sought to determine whether DCs possessed intracellular-recognition mechanisms for lipopeptide/lipoprotein-associated PAMPs. SVLPs [PE-SVLPs (16, 17)] known to interact with DCs (16) were modified to carry immunomodulatory di- or triacylated lipopeptide ligands for TLR2 heterodimers (P2C/P3C-SVLPs) within their hydrophobic cores (Fig. 1A, Supplemental Fig. 1) (17). Placing TLR2 ligands within the hydrophobic core (P2C-SVLPs and P3C-SVLPs) relates to their position within the bacterial cell wall because they cannot interact with DC surface TLR2s. This allowed assessment of hydrophobic TLR2 ligand processing without interference from surface TLR2s, which occurs with isolated ligands and whole bacteria because of accessible TLR2 ligands, such as TA.
We reported previously that these PE-SVLPs lacking TLR2 ligands rapidly interacted with DCs (16). With the present work assessing interaction of DCs with SVLPs carrying TLR2 ligands in their hydrophobic core, it was considered important to define the role of endocytic processing of SVLPs, because it is essential for the generation of peptides for Ag presentation and for exposing TLR2 ligands. Accordingly, the endocytic events were assessed in a relative manner with respect to interaction of SVLPs with internal TLR2 and the ultimate analysis of immunogenicity when adjuvant addition was precluded.
SVLP interaction with DCs led to internalization by various endocytic routes differing in their relative roles and kinetics. Clathrin-dependent endocytosis and any involvement of recycling/sorting endosomes may have appeared early; at best, they were minor processes because of their lack of significance, as defined by the Pearson coefficient of colocalization between the endocytic marker and SVLPs. A more apparent association of endocytosed SVLPs with the ER (ER-Tracker+) was noted, but the Pearson coefficient of colocalization remained relatively low. The main endocytic routes involved lipid raft mobilization in association with CTB binding. The majority of SVLP-internalization events led to a gradual and time-dependent association with dextran+ macropinosomes, although some evidence for caveolar endocytosis was noted. Colocalization with CTB, dextran, and caveolin-1 had greater significance, as defined by the Pearson coefficient of colocalization, particularly the CTB and (from 30 min) dextran events. This relationship between SVLP endocytosis and lipid raft involvement and primary endocytosis via macropinocytosis was also reflected in terms of the percentage of cells exhibiting potentially colocalized voxels.
Interaction with early endosomal structures then became more evident, remaining a gradual process over time, which is typical for macropinocytosis and caveolar endocytosis (35, 36). Involvement of early endosomes also paralleled the increase in enzymatic activity, as reflected by dequenching of DQ-OVA. Ultimately, association with late endosomes/lysosomes (LysoTracker+) was noted, which again was a particularly slow process, but it eventually showed high significance, as defined by the Pearson coefficient of colocalization. The lower frequency of association with Lysotracker+ structures after 6 h may reflect the nature of endolysosomal degradation processing the majority of SVLP Ags beyond detectability.
Certainly, the presence of SVLPs within different stages of the endolysosomal compartment would be necessary for the processing leading to Ag presentation. Importantly, this could also provide the environment for release of the hydrophobic core and, therefore, lipopeptide PAMPs. Although SVLP association with MHCII+ structures was not particularly evident, this was not surprising. Processing to provide antigenic epitopes for association with MHCII in the MIIC is likely to result in loss of SVLP detectability. Certainly, the DCs were capable of presenting antigenic peptides derived from the SVLPs (see below concerning the induction of adaptive immune responses).
In contrast to the results obtained for MHCII+ compartments, SVLP association with internal TLR2 was identified. Indeed, SVLP colocalization with intracellular TLR2+ structures exhibited a high Pearson coefficient of correlation. This may reflect exposure of TLR2 ligands contained within SVLP lipopeptide monomers in LysoTracker+ compartments prior to processing into peptides for the MIIC. At least some of this detectable processed material would be delivered into TLR2+ structures, allowing interaction of PAMPs with intracellular TLR2. This was independent of interaction with cell surface TLR2; SVLPs do not bind with cell surface TLR2, nor does anti-TLR2 Ab prevent DC endocytosis of SVLPs. Moreover, the kinetics of association with the intracellular TLR2 was too rapid (relative to the slow kinetics of SVLP endocytosis) for any exocytosis of the processed material. Considering the kinetics of SVLP association with the EEA-1+ and LysoTracker+ compartments, these were closely related to the kinetics of association with intracellular TLR2 during the first 2 h following initiation of endocytosis.
The functional significance of the colocalization between endocytosed SVLPs and intracellular TLR2 was assessed using blood DCs and BMDCs. Di- and triacylated SVLPs induced TNF-α, IL-1β, and IL-6; this was detected intracellularly in SVLP+ cells and as secreted cytokines, which are essential for paracrine and autocrine immunological activities.
DCs and macrophages can produce TNF (37, 38), but the greater migratory capacity of DCs (39–42) would relate more to advancement of adaptive immune defenses, with more sedentary macrophages relating more to local inflammatory responses, although the latter would influence activities of resident and recruited DCs. It is well documented that the major TNF-producing DC subset contains 6-sulfo LacNAc and TNF/INOS-producing DCs (43, 44). Although these have not been identified in the pig, it is likely that they do exist, considering their identification in humans (initially as SLAN DCs) and mouse (initially as Tip-DCs). The CD16 marker used to identify SLAN DCs in humans is not applicable in the pig, because most DCs, monocytes, and macrophages bear CD16. We have also tried to use the anti-human SLAN DC marker on pig cells, but this does not work. It is also possible that DCs other than the equivalent of 6-sulfo LacNAc and TNF/INOS-producing DCs can produce TNF with porcine cells. Nonetheless, the main result is that DCs will produce TNF in response to P2C-SVLPs, as well as TLR2 agonists, such as P2C and P3C. As such, this is an important cytokine for promoting DC maturation.
The consequence of this internal TLR2 ligation leading to cytokine induction was immunologically relevant. P2C/P3C-SVLPs and PE-SVLPs (lacking TLR2 ligands in the core) efficiently interacted with blood DCs and monocytes within the CD172alo and CD172ahi gates in a manner that was dependent on SVLP concentration. Yet, only P2C/P3C-SVLPs with TLR2 ligands induced upregulation of CD86 expression.
These results demonstrate that the observed interaction of SVLPs with intracellular TLR2 was functionally important, leading to signaling and immunological activity without any prerequisite for interaction with cell surface TLR2 heterodimers. Although DC processing of SVLPs used different endocytic pathways, macropinocytosis and caveolar endocytosis dominated, ultimately involving late endosomes/lysosomes likely exposing the di- and triacylated cores for interaction with intracellular TLR2 sites. In addition to PAMP recognition by intracellular TLR2, SVLP processing supplied antigenic peptides, despite the aforementioned difficulty in identifying an association between processed SVLPs and MHC molecules. SVLPs have only a minority of their lipopeptide monomers labeled, to prevent overconjugation and quenching of the fluorochrome signal. Therefore, the majority of lipopeptides being processed into peptides for association with MHC class I and MHCII would not be labeled, yet they should be detectable through induced immune responses. Ag presentation was identified in vitro in terms of induced IFN-γ–producing cells, demonstrating that DC processing had yielded peptides for association with MHC molecules. This was confirmed by vaccination using SVLPs constructed with lipopeptides carrying TLR2 ligands. These SVLPs were clearly immunogenic in vivo in the absence of any additional adjuvant, whereas SVLPs lacking TLR2 ligands were not immunogenic unless they were formulated with an adjuvant. The in vivo readouts were observed as serum Ab and IFN-γ ELISPOTs, demonstrating that at least MHCII presentation had occurred, and the observed interaction with intracellular TLR2 had an immunological significance. Indeed, processing of SVLPs did favor activation of Th lymphocytes, leading to the development of specific Ab production by B lymphocytes.
The overall outcome is that a synthetic vaccine can be designed to provide antigenic epitopes and a hydrophobic core adjuvant within the same particle. As exemplified by di- and triacylated SVLPs, processing by DCs provides for activation/maturation of the DCs and consequential Ag presentation to naive T and B lymphocytes.
Acknowledgements
We thank Res Michel, Hans-Peter Lüthi, and Daniel Brechbühl for blood donor care/sampling; Brigitte Herrmann for microscopy on SVLP association with TLR2, together with intracellular TNF-α and IL-6 detection in SVLP+ cells and competition studies between SVLP and anti-TLR2 Ab or the different TLR2 agonists; Panagiota Milona for acquisition of flow cytometry results for competition studies between SVLP and anti-TLR2 Ab or the different TLR2 agonists; and Francisco Sobrino (CBMSO, Madrid, Spain) for advice on the sequences of the B cell and T cell epitopes used in the synthesis of lipopeptides for construction of the SVLP vaccines.
Footnotes
This work was supported by Swiss National Science Foundation Projects NANOVACC (310000-1198828) and RNA-Targeting (310030-150008).
The online version of this article contains supplemental material.
Abbreviations used in this article:
- BMDC
bone marrow–derived DC
- cDC
classical DC
- CTB
cholera toxin subunit B
- DAMP
danger-associated molecular pattern
- DC
dendritic cell
- ER
endoplasmic reticulum
- INOS
inducible NO synthase
- LTA
lipoteichoic acid
- MBCD
methyl-β cyclodextran
- MHCII
MHC class II
- MIIC
MHC class II rich compartment
- MoDC
monocyte-derived DC
- MS
mass spectrometry
- PAMP
pathogen-associated molecular pattern
- P2C
Pam2CysSK4
- P3C
Pam3CysSK4
- P2C/P3C-SVLP
P2C-SVLP and P3C-SVLP
- P2C-SVLP
P2Cys-SVLP
- P3C-SVLP
Pam3Cys-SVLP
- pDC
plasmacytoid DC
- PE-SVLP
SVLP with a 1,3-dipalmitoyl-sn-glycero-phosphatidylethanolamine core
- PGN
peptidoglycan
- ROI
region of interest
- SVLP
synthetic virus-like particle
- TA
teichoic acid
- TFA
trifluoroacetic acid.
References
Disclosures
A.G. and J.A.R. are cofounders of Virometix AG, a company formed to commercialize SVLPs for vaccine purposes. The other authors have no financial conflicts of interest.