Dressing up Nanoparticles: A Membrane Wrap to Induce Formation of the Virological Synapse

Abstract

Next generation nanoparticle-based drug delivery systems require the ability to target specific organelles or subcellular regions in selected target cells. Human immunodeficiency virus type I (HIV-1) particles are evolutionarily optimized nanocarriers that have evolved to avoid intracellular degradation and achieve enrichment at the synapse between mature dendritic cells (mDCs) and T cells by subverting cellular trafficking mechanisms. This study demonstrates that integration of the glycosphingolipid, GM3, in a membrane around a solid nanoparticle (NP) core is sufficient to recapitulate key aspects of the virus particle trafficking in mDCs. GM3 presenting artificial virus NPs (GM3-AVNs) accumulate in CD169⁺, CD81⁺, non-lysosomal compartments in an actin-dependent process that mimics the sequestration of HIV-1. Live-cell optical tracking studies reveal a preferential recruitment and arrest of surface scanning CD4⁺ T cells in direct vicinity to the AVN-enriched compartments. The formed mDC-T cell conjugates exhibit strong morphological similarities between the GM3-AVN-containing mDC-T cell synapse and the HIV-1 virological synapse, indicating that GM3-CD169 interactions alone are sufficient for establishing the mDC-T cell virological synapse. These results emphasize the potential of the GM3-AVN approach for providing therapeutic access to a key step of the host immune response – formation of the synaptic junction between an antigen-presenting cell (mDC) and T cells – for modulating and controlling immune responses.

The immunological synapse¹ defines an organized tight contact between an antigen presenting cell, such as a dendritic cell (DC), and a lymphocyte that plays an important role in ensuring an efficient communication between key components of the immune system. A series of diseases ranging from cancer² to autoimmune disorders³ would benefit from a selective targeting of the immunological synapse with therapeutic NPs and delivery systems. Interestingly, human immunodeficiency virus type 1 (HIV-1) has evolved to subvert cellular trafficking and signaling mechanisms within DCs to achieve accumulation at the DC-T cell contact site.^4–7 These unique cell-to-cell contacts, termed virological synapses⁸ facilitate an efficient transmission of HIV-1 from DCs to CD4⁺ T-cells, which are the preferred host cells of the virus.⁹ In the case of HIV-1, engagement of the sialoadhesin CD169, or Siglec1, expressed on mDCs by the terminal α2–3 sialylated ganglioside GM3 contained in the viral membrane is sufficient to trigger a localization of virus particles to the cellular periphery and, upon initiation of a mDC-T cell contact, induce their trafficking to the mDC-T cell virological synapse.^10–13 The GM3-CD169 mediated cellular uptake and trafficking mechanisms provide, therefore, unique opportunities for site-selective delivery with appropriately designed NP platforms. Site-selective delivery is crucial for surgical delivery approaches that achieve maximum efficacy and minimize collateral damage.¹⁴

Due to their complementarity in size with virus particles, colloidal NPs with rationally tunable surface properties have attracted significant interest as virus model systems that mimic specific functionalities of the evolutionarily optimized virus particles for improved drug delivery.^15–17 Membrane wrapped NPs^18–23, in particular, are an emerging class of hybrid nanomaterials that combine the biological functionality of a lipid membrane with the advantageous photophysical²³, multimodal¹⁹, or delivery properties²¹ of an inorganic NP core for various theranostic applications. In this study we demonstrate that membrane wrapped gold NPs successfully trigger GM3-CD169 dependent cellular trafficking mechanisms in mDCs and achieve a similar intracellular fate as HIV-1 particles. Using single particle tracking analysis, we show that integration of GM3 into these artificial virus nanoparticles (AVNs)²³ is sufficient to achieve their preferential localization at the mDC-T cell synaptic junction, which shares strong molecular resemblances with the virological synapse. These findings highlight a number of potential advantages of the GM3-CD169 strategy, including targeting of a specific cell population (CD169 expressing macrophages and dendritic cells)²⁴, uptake via a non-lysosomal trafficking pathway, and the ability to modulate mDC-T cell signaling by selectively delivering therapeutic NPs to the mDC-T cell synaptic junction.

RESULTS AND DISCUSSION

AVNs were generated through a one-pot assembly strategy in which citrate stabilized gold NPs were incubated with 1-octadecanethiol in the presence of liposomes of defined composition.^{22, 23} Due to the high affinity of the thiol residue to the gold surface, 1-octadecanethiol replaces the citrate and assembles into a 1-octadecanethiol monolayer around the gold NPs. The hydrophobic tails of the lipids readily integrate into this layer, completing the AVN assembly. The resulting AVN structure is schematically depicted in Figure 1a. For each AVN batch we validated the successful formation of a membrane around the gold NP core by inspection in the TEM (Figures 1b and S1), by monitoring the hydrodynamic radius of the particles through dynamic light scattering (DLS) (Figure 1c) and by quantifying the colocalization of the fluorescently labeled membrane with the NP scattering signal in correlated fluorescence / darkfield single particle microscopy (Figure 1e). We also monitored the UV-Vis of the NPs before and after membrane assembly (Figure 1d) to ensure that the AVN formation did not result in NP agglomeration. The absence of a red-shift confirms that the membrane assembly does not induce agglomeration. In fact, the AVN spectrum is even slightly sharper than that of the NP cores before membrane assembly, indicating a colloidal stabilization of the NPs due to membrane formation. A successful membrane assembly resulted in an increase of the average hydrodynamic NP diameter from 91±1nm to 102±3nm and yielded colocalization probabilities of >95%.

Figure 1. Characterization of GM3 containing AVNs.

a. Scheme of the AVN structure comprising a lipid monolayer anchored to the gold NP core through 1-octadecanethiol. b. High resolution TEM image of a representative GM3-AVN shows a visible membrane around the gold nanoparticle core, indicative of successful membrane assembly. c. Hydrodynamic diameter (nm) of GM3 AVNs and 80nm gold (Au) NPs (used as AVN core) measured by Dynamic Light Scattering (DLS). d. UV-VIS spectra of GM3-AVNs and 80nm gold NPs show peak wavelengths centered at 544nm. e. Darkfield image (red, top), fluorescent image (green, bottom), and superimposed image (right) confirm a strong colocalization (yellow) of the fluorescently labeled membrane and the gold NP core for immobilized GM3-AVNs. A colocalization level of approximately 96% indicates a highly efficient membrane wrapping of the gold NP core.

The HIV-1 membrane contains cholesterol, virus-encoded glycoproteins, and various host cell-derived phospholipids and glycosphingolipids.^{25, 26} We used a simplified lipid membrane composition for the AVNs that comprised only dipalmitoylphosphatidylcholine (DPPC) and cholesterol as major components (Table S1). Phosphatidylserine (PS, 0.1mol%) was added to maintain a virus-like surface charge (ζ≈ −25mV), and the inclusion of <0.1mol% Topfluor labeled cholesterol in the membrane allowed the detection of membrane formation through fluorescence microscopy. We included GM3 at a nominal concentration of 3mol%. A characterization of the surface charge in DPPC/cholesterol/GM3 wrapped AVNs (Figure S2) indicated that this feeding ratio led to an effective concentration of ≈1mol% GM3 in the AVN membrane, which is in excellent agreement with the expected GM3 concentration in HIV-1. The resulting GM3-AVNs showed specific binding to CD169-expressing lipopolysaccharide (LPS)-activated DCs as determined by inductively coupled plasma mass spectroscopy (ICP-MS, Figure 2a), flow cytometry (Figure 2b), and optical cell binding studies (Figure S3).

Figure 2. GM3-AVNs recapitulate GM3-CD169 mediated HIV-1 sequestration into peripheral non-lysosomal compartments in mDCs.

a. Mature DCs were challenged with GM3-AVNs, GalCer-AVNs or blank controls at three different AVN/mDC ratios (1×10⁷/3×10⁵; 1×10⁸/3×10⁵; 1×10⁹/3×10⁵) for 5min at 37° C before the number of gold (Au) NPs (mean ± S.D.) bound per mDC was determined by ICP-MS. Data were collected in two independent experiments (two donors) for each dosage. The design of the AVNs is included in the inset.b. Mature DCs were challenged with fluorescently labeled GM3-AVNs, GalCer-AVNs or blank controls at three different AVN/mDC ratios (1×10⁷/3×10⁵; 3×10⁷/3×10⁵; 1×10⁸/3×10⁵) for 15 or 30 min at 37°C before the mean fluorescence intensities were determined by flow cytometry. Data for each dosage represent means of duplicate measurements from one donor. c. Staining of lysosomes (lysotracker; top panel), CD81 (middle panel), and CD169 (bottom panel) in GM3-AVNs containing mDCs. The AVN/mDC ratio was 1×10⁸/3×10⁵ and the incubation time was 1h. Scale bars = 1μm. d. SEM of an mDC section shows a peripheral AVN cluster located close to the cell surface. The Energy Dispersive X-ray analysis (EDAX) (inset, left right corner) shows the elementary composition of the enlarged field, which confirms that the imaged NPs are gold NPs.

GM3-CD169-mediated AVN Capture by mDCs

Our protein-free AVN design with a simple lipidome is ideally suited to investigate GM3's function in mediating HIV-1 capture and mDC-T cell virological synapse formation. Importantly, GM3-AVN binding to mDCs triggers unique uptake and trafficking mechanisms that results in the segregation of AVNs in peripheral nonlysosomal (Figure 2c, top row), CD81⁺ (Figure 2c, middle row) compartments.²³ Immunofluorescence staining of CD169 (Figure 2c, bottom row) confirms now that these unique compartments are also enriched in CD169, which provides further evidence of a GM3-CD169 binding mediated segregation mechanism. We emphasize that the enrichment of CD169 at the site of GM3-AVN clustering in mDCs closely resembles the colocalization of CD169 with HIV-1 particles in CD81⁺ peripheral compartments.^{10, 13, 23} Interestingly, these GM3-AVN containing compartments resemble the recently described HIV-1 containing CD169⁺ plasma membrane in mDCs.²⁷ Since the widefield optical images in Figure 2c do not provide specific information about the intracellular location of the AVN enriched compartments, we prepared mDC sections for imaging in the scanning electron microscope (SEM). Figure 2d shows a representative SEM image of a cross-section of a GM3-AVN treated mDC after an incubation period of 1h. Numerous dendrites characteristic of mDCs are prominently seen, and a gold (Figure 2d, inset) AVN cluster was located just below the cell surface.

Together, the performed optical and electron microscopic studies verify that GM3-AVNs accumulate in CD81⁺, CD169⁺, non-lysosomal compartments close to the cell surface. The segregation of GM3-AVNs in these unique compartments indicates a successful mimicry of the GM3-CD169-dependent trafficking of HIV-1 particles in mDCs.^{28, 29}

Characterization of GM3-AVN Motility and Clustering in mDCs

The optical properties of the 80nm gold nanoparticle core of AVNs are determined by coherent electron oscillations (plasmons) whose resonances lie in the visible range of the electromagnetic spectrum.³⁰ Due to resonant light scattering, AVNs provide large scattering cross-sections and can be imaged with high contrast and without blinking or bleaching on or in DCs under darkfield illumination.^{31, 32} We utilized the bright nature of the AVNs to experimentally address the kinetics of CD169-mediated clustering of GM3-AVNs in peripheral membrane compartments in mDCs. Mature DCs were incubated with GM3-AVNs (1×10⁸/3×10⁵ AVN/cell ratio) in serum-free growth medium at 37°C for increasing time intervals (5min, 15min, 30min, 45min, and 60min) before the AVNs were removed and the spatial distribution of the AVNs on the single cell level was determined through darkfield microscopy.

We subdivided the AVN-treated cells into three groups (No AVN, Cluster and Random) as defined in Figure 3a. The fraction of AVN containing cells (sum of Cluster and Random) increases initially as function of time (Figure 3b, green curve) and plateaus after approximately 20min when ~50% of all mDCs contain AVNs. We emphasize that this incomplete binding is not particular to AVNs. Similar binding levels are achieved with HIV-1 Gag virus like particles (VLPs),¹² and we rationalize this observation with variations in the Poisson-distributed CD169 expression level within primary cells (mDC). The ratio of the Cluster and Random phenotypes (blue and red curves in Figure 3b) changes as a function of time. Within the first 10min the relative contribution from the Cluster phenotype increases sharply from 30% to 57% at the expense of the Random phenotype. Subsequently, the rate of conversion of the Random into the Cluster phenotype becomes more gradual with increasing time (after 60 min the Cluster/Random ratio is 68% / 32%).

Figure 3. Characterization of the temporo-spatial GM3-AVN distribution in mDCs.

a. Illustration of the GM3-AVN distribution pattern observed: No AVN (left), Cluster (middle), and Random (right). Scale bar = 1μm. b. Fraction (±S.D.) of mDCs containing GM3-AVNs (green curve) and relative occurrence (±S.D.) of Cluster (red curve) versus Random phenotype (blue curve) in mDCs as a function of time. For each time point data were collected from >200 mDCs from three different donors in three independent experiments. c. Relative occurrence of No AVN, Cluster, and Random phenotypes after treatment with Colchicine, Cytochalasin D, or Jasplakinolide, as well as for the control without treatment. Presented data are averages (±S.D.) of two repeats from one donor.

We found that actin-targeting pharmaceuticals, such as Jasplakinolide, a macrocyclic peptide that inhibits actin filament depolymerization and induces monomer polymerization, as well as Cytochalasin D, an actin polymerization inhibitor, suppress AVN clustering (Figure 3c). In contrast, treatment with the microtubule polymerization inhibitor Colchicine had no effect. These observations indicate a key role for a dynamic cortical actin network in AVN clustering (vide infra).

To validate that the Cluster phenotype emanates from the Random phenotype, we monitored the spatio-temporal distribution of GM3-AVNs within individual cells. To that end, mDCs were initially incubated with GM3-AVNs for 10 min at 4°C. The samples were then transferred into a darkfield microscope equipped with a cage incubator set at 37°C, and all of the AVNs bound to the mDCs in the field of view were tracked simultaneously. Figure 4a shows two images of a GM3-AVN containing mDC; one at the beginning of the image acquisition and another after 8min24sec. The initially randomly distributed AVNs show an obvious coalescence, which is entirely consistent with the conversion of the Random into the Clustered phenotype. While we cannot exclude some degree of NP agglomeration in solution, especially at longer incubation times, the observation of AVN clustering in individual cells in real-time together with the actin-dependence of the clustering process (Figure 3c) indicate a GM3 triggered cellular response as origin of the AVN segregation shown in Figure 3b.

Figure 4. Real-time tracking of GM3-AVN in live mDCs.

a. Representative images of a GM3-AVNs containing mDC at t = 0 min and t = 8min24s. Scale bars = 1μm. b. Trajectories (~20s) of six GM3-AVNs in mDCs (yellow arrows in a). c. Moment scaling spectrum (MSS) slopes (S_MSS) plotted as function of diffusion coefficient D₂ for 52 trajectories. d. Histogram of average speeds for the recorded trajectories. e. Histogram of maximum speeds for the recorded trajectories.

Optical single virus tracking has become a reliable tool to obtain mechanistic information about virus uptake, trafficking, and transfer.^33–37 We reasoned that AVN tracking is equally well suited to provide additional mechanistic insight into the virus clustering mechanism in mDCs. To visualize this trafficking mechanism in real time, we recorded movies of individual mDCs at different time points after AVN exposure in a darkfield microscope. We constructed trajectories for all of the AVNs detectable on the imaged mDC top surfaces and found a heterogeneous AVN motility characterized by different motion patterns at different time points. This effect can even be observed within individual trajectories. Figure 4b shows representative trajectories (approximately 20s) of the AVNs marked as 1–6 in Figure 4a at selected time points. The starting points of the individual trajectories are marked by arrows. AVNs 1, 2, and 5 show obvious differences in their motility within the duration of the movie and contain segments of directed as well as of more random motion. To analyze the motion pattern of AVNs during the clustering process more systematically, we calculated the moment scaling spectra (MSS) introduced by Ferrari et al.³⁸ for a total of 52 trajectories with durations of approx. 20s from ten AVNs recorded on three different cells. The MSS slope (S_MSS) represents a quantitative metric for characterizing the motion of diffusing NPs, and S_MSS values of 0, 0.5, and 1 correspond to immobilized, free, and directed diffusion, respectively.^{39, 40} Figure 4c summarizes the obtained S_MSS values in a plot against the linear diffusion coefficient D_2. The average diffusion coefficient $\overline{{\mathrm{D}}_2}=5.5\times {10}^{-3}\mu {\mathrm{m}}^2∕\mathrm{s}$ is significantly lower than for free NP diffusion on a membrane.⁴¹ The S_MSS values show a broad distribution in the range between 0.02 and 0.82, confirming variations in the lateral motility between individual AVNs and for individual AVNs at different points in time. In particular, the motion patterns of individual CD169 tethered GM3-AVNs indicate that periods of confined motion are interrupted by relatively short periods of directed motion with S_MSS>0.5. The latter are responsible for achieving a segregation of initially randomly distributed AVNs into larger clusters.

The average and maximum speeds for the recorded trajectories are graphed in Figure 4d and Figure 4e, respectively. The average speed for all trajectories is 0.16 ± 0.11μm/s and more than 70% of the AVNs achieve maximum speeds <0.5μm/s. Although a few AVNs (e.g. AVN1 in Figure 4b) show a faster directed motion, indicative of a microtubule-dependent molecular motor driven process, the motion pattern of the AVN majority is compatible with processive, actin-based myosin motors, which achieve speeds between 0.1 – 0.5μm/s in vivo.⁴² The measured AVN speed distribution and the observation that a perturbation of the cortical actin network (de-)polymerization inhibits AVN clustering (Figure 3c) are – in principle – consistent with a role of the actin-binding myosin in the observed AVN translocation. CD169 is, however, a transmembrane protein with a very short cytoplasmic tail,²⁴ and it is unclear whether it can directly bind to myosin motor proteins associated with the cortical actin network. Another possible explanation of the observed motion pattern, which does not require direct interactions between CD169 and motor proteins, is based on a concerted restructuring of the cortical actin network to guide the diffusion of CD169. According to this model, CD169 transmembrane proteins are confined to cortical actin defined membrane compartments (corrals)⁴³. Binding of GM3-displaying AVNs or virus particles induces either a directed restructuring of the cortical actin network in the vicinity of the receptor or makes the network permissible in one specific direction. The CD169 receptors present in a cortical actin confined membrane microdomain explore the accessible space through diffusion and, thus, follow the trajectory defined by the dynamic cortical actin network.

Independent of the exact mechanism of AVN translocation, our studies show clearly that GM3-CD169 binding is sufficient to initiate a coordinated transport of GM3-AVN bound CD169 receptors over relatively large cellular distances. Considering that CD169 is a transmembrane protein without known transduction motifs,⁴⁴ it remains unclear how nanoconjugated CD169 can induce this cell-wide response. We postulate that the receptor initiates the observed cellular response by recruiting one or more yet to be identified co-factor(s). A future identification of these co-factors would not only result in an improved understanding of the HIV-1 trans infection mechanism, but might also identify a potential pathway for a general intracellular NP delivery into non-lysosomal, non-degradative compartments.

AVN-Polarization in mDC-T Cell Contacts

To determine if GM3-AVNs successfully access trafficking mechanisms that have previously been shown to lead to an accumulation of HIV-1 particles at the mDC-T cell synapse,^{10, 11} we probed the fate of GM3-AVN clusters in mDCs upon initiation of mDC-CD4⁺ T cell conjugates. Mature DCs were pre-treated with GM3-AVNs for 1.5h to ensure completion of AVN cluster formation before CD4⁺ T cells were added (1/1 mDC/T cell ratio). The cells were then incubated for 5 - 240min before they were inspected in the darkfield microscope to classify the formed mDC-T cell conjugates. We differentiated five classes of mDC-T cell conjugates as shown in Figure 5a. Our categorization was based on the orientation of the AVNs with regard to the T cell and followed previous strategies⁴⁵ for quantifying the localization of HIV-1 particles at the mDC-T cell synaptic junction. Depending on the relative location of AVN cluster and T cell, we refer to the mDC-T cell conjugates as Oriented (AVN cluster is located in the same hemisphere of the mDC as the T cell) and Non-Oriented (AVN cluster is localized in the opposite hemisphere of the mDC as the T cell). Other classifications included No AVN, Random (bound AVNs are not clustered), or Transferred to T cell, but these phenotypes were less frequent (Figure S4) and play only a secondary role. To avoid any ambiguity in the analysis we restricted our analysis to mDC-T cell conjugates with at most two T cells per mDC. For those conjugates with two T cells, the mDC was segmented such that one hemisphere contained both T cells (Figure S5).

Figure 5. Classification of mDC-T cell conjugates with GM3-AVN clusters.

a. Examples of Oriented, Non-Oriented, Random, No AVN spatial distribution patterns. The classification is determined by the relative orientation of AVN cluster and T cell (see text). In case of the Oriented pattern the AVN cluster is located in the same cell hemisphere as the T cell, whereas for the Non-Oriented pattern, T cell and AVN cluster are located in two different hemispheres of the mDC as indicated by the dashed lines. Scale bars = 1μm. b. Total occurrence (±S.D.) of mDC-T conjugates containing GM3-AVN clusters as function of time. c. Relative fraction (±S.D.) of Oriented and Non-Oriented patterns as function of time. Data were collected from two donors in two independent experiments. More than 200 conjugates were counted for each time point. Error bars are standard deviations.

The fraction of mDCs that exhibit AVN segregation and that have at least one T cell bound increases quickly to approximately 71% of the entire ensemble within the first 15min after addition of T cells but remains almost constant for longer incubation times (Figure 5b). In the next step, we sub-divided these mDC-T cell conjugates into the Oriented and Non-Oriented subgroups. The results of this classification for approximately 200 mDCs for each time point derived from two different donors are summarized in Figure 5c. The Oriented pattern shows a continuous increase within the first 60min, while the Non-Oriented pattern decreases (Figure 5c). The Oriented/Non-Oriented ratio changes from 1/1 to 3/1 within the first 60min and remains approximately constant for longer incubation times.

To further characterize the mDC-T cell interface in the case of the Oriented phenotype, we performed immunofluorescence studies and evaluated the colocalization between AVN clusters and known markers of virological synapses at the mDC-T cell contact after 2h of co-incubation. CD169 has been demonstrated to induce HIV-1 trafficking to the infectious synapse upon T cell contact and to show a strong polarized distribution towards the mDC-T cell interface.^{10, 13} Likewise, the tetraspanin CD81 is also enriched at the mDC-T cell virological synapse.^{46, 47} Finally, the restructuring of the cell membranes at an infectious synapse requires actin, and high local F-actin concentrations have been shown to colocalize with the site of HIV-1 particle accumulation at the virological synapse.⁴⁸ Figure 6 shows correlated darkfield / fluorescence images of mDC-T cell conjugates with an Oriented GM3-AVN distribution stained for CD169 (a), CD81 (b), and F-actin (c). All of these markers colocalize with the GM3-AVN cluster at the mDC-T cell synapse, confirming an mDC-T cell interface that shares significant morphological and functional similarities with HIV-1 induced virological synapses.

Figure 6. Characterization of Oriented mDC-T cell conjugates through correlated darkfield /fluorescence imaging.

The cells were stained for CD169 (a), CD81 (b) andF-actin (c). The left rowshows the darkfield color image, the middle row the monochromatic fluorescence image, and the rightrow shows a false-color overlay. Colocalization is indicated in yellow (AVN =green, fluorescencemarker = red).T cells were stained with CellTracker Orange andare plotted in cyan pseudo color.Nuclei were stained with DAPI fluorescence stain and are marked blue. Scale bars = 1μm.

Mechanisms of GM3-AVN Enrichment at the Virological Synapse

The anti-correlated changes in the occurrence of Oriented and Non-Oriented mDC-T cell conjugates within the first 60 min of observation (Figure 5c) imply a conversion of the Non-Oriented into the Oriented pattern as consequence of some relative motion of AVN clusters and T cells towards each other. To test this hypothesis, we continuously imaged mDC-T cell conjugates as function of time and monitored the relative alignment of AVN cluster and T cell. In these experiments mDCs pulsed with GM3-AVNs were immobilized on poly-lysine treated glass bottom dishes. AVN cluster containing mDCs were located in the darkfield microscope equipped with a cage incubator and centered in the field of view. CD4⁺ T cells pre-stained with CellTracker Orange were subsequently locally delivered with a micropipette mounted on a micromanipulator. The T cells bound efficiently to the mDCs upon contact formation, and a binding event defined the starting point for the optical tracking. Our optical set-up allowed the acquisition of both fluorescence and elastic scattering images, ensuring a precise localization of both fluorescently marked T cell and AVN cluster.

The live cell imaging studies revealed two different relative motion patterns between AVN clusters and T cells. The most commonly observed motion is illustrated in Figure 7a. The series of images of a mDC-T cell conjugate with an AVN cluster recorded at different time points shows a rapid motion (`rolling') of the T cell across the mDC surface within the first ten minutes that leads the T cell into the vicinity of the AVN cluster (green arrow). Interestingly, the T cell motion stops almost entirely once the cell has reached the vicinity of the AVN cluster. We continued to image the mDC-T cell conjugate for another 20min and did not observe any further movement of the T cell. Overall, the movement of the T cell along the mDC surface converts the AVN pattern from the Non-Oriented into the Oriented form in the course of approximately 10min. We observed similar T cell rolling behaviors that eventually resulted in an immobilization of the T cell in the vicinity of an AVN cluster for six different mDC-T cell conjugates in three independent experiments. The observed T cell behavior is consistent with the known rapid scanning of T cells over DC surfaces to identify cognate peptide-MHC complexes prior to prolonged arrest and forming stable mDC-T cell synapses.^49–51 Our T cell tracking data suggest that peripheral compartments containing GM3-AVNs in mDCs might be segregated with unique surface features that trigger arrest of scanning T cells and initiate, thus, the formation of a stable synaptic contact.

Figure 7. Real-time monitoring of the Oriented mDC-T cell pattern formation.

a. Snapshots of selected time points showing a T cell rolling across the mDC surface towards the GM3-AVN cluster (bright white dot marked by green arrow). b. Snapshots of selected time points showing AVN cluster migrating towards T cell. Scale bars = 1μm. The images show dark field images with superimposed fluorescence image of the labeled T cells.Fluorescence images only served the purpose of identifying the T cells and were not recorded for all frames.

Figure 7b illustrates a second pattern of observed mDC-T cell motility. Here, the T cells in contact with the mDC do not systematically scan over the cell surface but, instead, remain located close to their original binding sites. Interestingly, in this case of immobilized T cells the AVN cluster (green arrow) migrates towards one of the bound T cells and reaches the mDC-T cell contact area within 30min. The observed average speed for the AVN cluster displacement is ~100 nm/min, which is distinctly slower than what was observed for the individual GM3-AVNs in Figure 4d. The direct motion of the AVN cluster was much less frequent than the T cell scanning of Figure 7a, and we cannot exclude that the immobilization of the T cells was a consequence of their binding to the glass bottom of the dish. Future tracking studies in free-floating mDC-T cell conjugates will be able to determine the relevance of this second motion pattern.

SUMMARY AND CONCLUSION

The systematic AVN imaging studies performed in this work suggest the mechanistic model summarized in Figure 8 for the observed AVN concentration at the virological synapse. AVNs bind to mDCs and induce the formation of peripheral AVN clusters. These AVN enriched compartments are then detected by surface scanning T cells, which bind at or close to this site and initiate the formation of a virological synapse at the site of the AVN clustering. According to this model GM3-presenting AVNs recapitulate key steps of the HIV-1 trans-infection mechanism in mDCs. In particular, we have shown that the inclusion of GM3 in a self-assembled membrane around a gold NP suffices to achieve an efficient and selective binding to mDCs and to subsequently induce a segregation of the GM3-NPs in tetraspanin CD81⁺, CD169⁺ nonlysosomal compartments within 10 – 60min post NP challenge. The observation that GM3-AVN segregation requires an intact dynamic cortical F-actin network but is not affected by microtubule disruption through colchicine indicates an F-actin dependent AVN segregation mechanism. The optical tracking of T cells on GM3-AVN containing mDCs has shown that the formation of GM3-AVN enriched compartments impacts mDC-T cell behavior. Most importantly, addition of CD4⁺ T cells to mDCs with pre-formed GM3-AVN clusters was observed to frequently result in the arrest of mDC surface scanning T cells in the vicinity of a GM3-AVN cluster. In contrast, trafficking of GM3-AVN clusters towards immobilized T cells occurred with much lower frequency. The mDC-T cell synaptic junctions containing GM3-AVN clusters are enriched in CD169, CD81, and F-actin and, thus, show a remarkable resemblance to previously described HIV-1 containing mDC-T cell virological synapses.^{10, 11, 27, 46, 52} While the enrichment of AVNs at the mDC-T cell interface was observed frequently, an actual transfer of AVNs from mDC to T cell was rare (less than 5% of the total AVN containing mDC-T conjugates, see Figure S4) under our experimental conditions, even after prolonged co-culturing of mDCs and T cells. This finding indicates that while GM3 is sufficient to induce the formation of a mDC-T cell virological synapse, additional factors besides GM3, such as engagement of CD4 on the T cell surface by the virus glycoprotein, gp120, might be required to achieve an efficient transfer of AVNs to T cells. To this end, insertion of T cell-derived filopodia into mDC invaginations harboring HIV-1 particles has been suggested to initiate directed release of mDC-bound virus particles towards the T cell surface.⁴⁸

Figure 8. Model of AVN concentration at mDC-T cell virological synapse consistent with the experimental observations.

GM3 presenting AVNs are segregated in peripheral membrane compartments that attract surface scanning T cells. The T cell binding induces the formation of a virological synapse.

Despite these limitations with regard to AVN transfer from mDC to T cell, the demonstrated ability to achieve an efficient localization at the mDC-T cell synaptic junction, and to trigger the enrichment of virological synapse characteristic molecular functionalities proves the capacity of GM3-AVNs to target key events in the adaptive immune response. GM3-AVNs are, therefore, promising functional NPs with tangible translational potential for enhanced immunization and therapeutic strategies. This work also underlines the value of self-assembled AVNs with defined composition as biophysical nanotools to study lipid-based inter-cell signaling and self-recognition mechanisms and their parasitization through enveloped viruses.

METHODS and MATERIALS

AVN preparation

Membrane wrapped nanoparticles were generated by co-incubating citrate stabilized gold nanoparticles with octadecane thiol in the presence of a liposome with defined lipid composition that act as lipid reservoir. A total of 1μmol lipid mixture (Table S1) was dissolved in 100μL chloroform, rotary evaporated dried to form a thin layer on 25mL round bottom flask, and further dehydrated under vacuum overnight. 1mL 20mM HEPES buffer (pH 7.2) was then added under argon to the lipid dry layer, and vigorously agitated to form a cloudy solution. The solution was subsequently sonicated with a probe sonicator (Fisher Scientific Model 120 Sonic Dismembrator, 50/60Hz)) for 2.5min or until the solution became clear. The formed liposome solution could be stored at 4°C under argon until further use. 1mL of 1.1 × 10¹⁰ particle mL⁻¹ 80nm gold NPs were pelleted by centrifugation at 600g. The pellet was then added into 0.5mL of the prepared liposome solution, and filled to 1mL with 20mM HEPES buffer. 200ul of 0.01mg mL⁻¹ 1-octadecanethiol (in ethanol) was added to the solution and agitated overnight to ensure thorough mixing and reaction. AVNs were washed by repeated centrifugation and resuspension (3 times, 600g) in ddi water, and finally resuspended into 100μL 20mM HEPES buffer.

AVN characterization

The hydrodynamic radius and zeta potential were obtained from 1×10⁷ particle mL⁻¹ containing AVN dispersions in 10mM NaCl (pH 7.0) using a Zetasizer Nano ZS90 (Malvern, Worcestershire, UK). The absorption spectra were acquired on an Agilent Cary 5000 UV/VIS Spectrometer. Base line correction was performed with a 10mM NaCl solution. AVN concentrations were calculated from the recorded spectra according to Lamber-Beer's Law (ε= 9.124 × 10¹⁰M⁻¹cm⁻¹). AVNs containing Topfluor labeled cholesterol (Avanti Lipids) were then drop-cast onto glass coverslips to validate successful membrane assembly by colocalizing membrane fluorescence and NP scattering signals. To prepare samples for inspection in the TEM, the AVN solution was diluted to 1×10⁷ particles mL⁻¹ with ddi water and drop-cast onto carbon-coated TEM grids. Excess solution was removed by a clean filter paper after 1min incubation, and the samples were dried and stored in vacuum before imaging using a JEOL JEM 2010 HRTEM with 200kV acceleration voltage.

ICP-MS quantification of AVN-mDC binding

1×10⁷, 1×10⁸, 1×10⁹ AVNs containing no GM3 or GalCer, 3% GalCer, or 3% GM3, were added to 3×10⁵ mDCs and incubated in 100 μL serum free growth medium (Roswell Park Memorial Institute Medium, RPMI-1640) at 37°C 5% CO₂ for 5min. Free unbound AVNs were then removed by centrifugation at 270g for 5min. The cells were subsequently fixed with 4% paraformaldehyde and dialyzed with 1x PBS buffer (pH = 7.0) overnight. After collection through centrifugation, the cells were counted. The cells were dissolved in Aqua Regia, and the concentration of gold in the samples was quantified with a Thermo (VG) PlasmaQuad (PQ) ExCell Inductively Coupled Plasma Mass Spectrometer (ICPMS).

Flow cytometry quantification of AVN-mDC binding

All AVNs contained 0.5% TopFluor PC (Avanti). 1×10⁷, 3×10⁷, 1×10⁸ AVNs with 3% GM3 or GalCer or without any glycosphingolipids (blank) were added to 3×10⁵ mDCs and incubated in 100 μL serum free growth medium (RPMI-1640) at 37°C 5% CO₂ for 15 or 30 min. Free unbound AVNs were then removed by centrifugation at 270g for 5min. The cells were subsequently fixed with 4% paraformaldehyde for 15min at room temperature. The fluorescence intensity distribution for each sample was tested with a BD LSRII instrument and analyzed through FlowJo software.

Fluorescence staining of AVN loaded mDCs and mDC-T cell conjugates

In a typical experiment 3×10⁵ mDCs were incubated in 100 μL serum free RPMI-1640 with 1×10⁸ GM3-AVNs for defined times. AVNs were then removed by centrifugation (270 g) and resuspended in serum free RPMI. For some experiments, LysoTracker DeepRed (Invitrogen) (final concentration of 100nM) and Hoechst 33342 were added and cells were incubated for 30min at 37°C to stain the lysosome and nucleus, respectively. The cells were then washed by centrifugation and fixed with 4% paraformaldehyde (10min, room temperature). To stain CD81 or CD169, mDCs were first fixed with 4% parafomaldehyde and then permeabilized with 0.5% Triton X100, blocked with 20% normal human serum (NHS). For CD81 staining, α-CD81 mAb (BD; 1:50 dilution) and Alexa488 conjugated secondary antibody (Invitrogen; 1:50 dilution) were added successively. For CD169 staining, Alexa647-conjugated α-CD169 mAb (Biolegend; 1:50 dilution) was used. To stain for F-actin, paraformaldehyde fixed mDCs were treated with fluorescein phalloidin (Life Technologies, 1:40 dilution) following the manufacture's protocol. To stain mDC-T cell conjugates, the staining procedures were identical with the exception that AVN-treated mDCs were incubated with CD4⁺ T cells in 1/1 ratio at 37°C for 2h in serum free RPMI-1640 prior to any staining procedure.

Statistical analysis of AVN clustering and mDC-T cell conjugate formation

For AVN cluster formation, 1×10⁸ GM3-containing AVNs were added to 3×10⁵ DCs, and incubated in 100 μL serum free RPMI-1640 at 37°C for 5min, 10min, 20min, 30min, 45min and 60min. Free unbound AVNs were then removed by gentle washing through centrifugation and resuspension (2×) and fixed in 4% paraformaldehyde. The cells were then cytospun onto glass coverslips (24mm × 60mm) and inspected in the dark-field microscope. For mDC-T cell conjugates, 1×10⁸ GM3-containing AVNs were added to 3×10⁵ DCs and incubated at 37°C for 1.5h in 100 μL serum free RPMI-1640, before the DCs were collected through centrifugation at 270g to purify from free unbound AVNs. The DCs were then mixed with CD4⁺ T cells in a 1:1 ratio and incubated in serum free RPMI-1640 at 37°C for 5min, 15min, 30min, 45min, 60min, 120min, and 240min respectively. The cells were then fixed with 4% paraformaldehyde, and cytospun onto glass coverslips for further imaging and counting. Counts of DCs and conjugates were collected from recorded movies across the entire imaging area, to ensure the randomness and fairness for the sample set. Data were collected from 3 different donors in 3 independent experiments (n=3) and more than 200 mDCs were counted for each time point

Image acquisition and processing

Optical imaging was performed with a modified Olympus IX71 inverted microscope (Figure S6) equipped with a cage incubator system. Darkfield scattering images were illuminated with a 100W tungsten lamp through a high NA oil darkfield condenser (NA=1.2–1.4), and were collected with a 60x oil objective with adjustable NA (NA=0.65–1.25). Color darkfield images and movies were recorded with a Nikon D3100 SLR through an eyepiece adapter. Monochromatic dark-field images and movies were recorded with an Andor Ixon electron multiplying charge coupled device detector (EMCCD). Fluorescence imaging was performed under epi-illumination using appropriate filter sets. Images were aligned and superimposed using ImageJ. Particle trajectories were recorded as short movies and MSS Slopes and particle speeds were analyzed with home written MATLAB codes.⁵³

Imaging AVN clustering

1–10⁸ GM3-containing AVNs were mixed with 3–10⁵ DCs in 100 μL serum free growth medium at 4°C for 10min; unbound AVNs were removed by washing through centrifugation and resuspension. The DCs were then placed into poly-lysine treated round glass bottom dishes. CO₂-independent medium (Life Technologies) was added after the cells sedimented to the bottom, to ensure adequate environmental conditions for live cell imaging studies. The cells were then transferred into the microscope at 37°C and data acquisition was initiated.

Tracking the relative motion of AVN clusters and T cells

3×10⁵ DCs were pretreated with 1×10⁸ GM3 containing AVNs for 1.5h at 37°C in 100 μL serum free growth medium, and washed by centrifugation (270g) and resuspension. T cells were stained with CellTracker Orange for 30min at 37°C, and washed to remove the free dye. In the microscope incubator, the DCs were dropped onto poly-lysine pretreated round glass bottom dishes and incubated at 37°C for 5min. Candidate mDCs containing a peripheral AVN cluster were then located under darkfield illumination. A stream of T cells were locally administered using a micromanipulator equipped with a micropipette. The interactions of T cells with mDCs were continuously tracked under darkfield illumination and fluorescence images were recorded at selected time points.

SEM sample preparation and imaging

5.6 ×10⁶ mDCs were incubated with 5.6×10⁹ GM3-containing AVN2 for 1hour at 37°C. Cells were washed twice with PBS and fixed with 4% PFA and 1% glutaraldehyde in 0.1 M PHEM buffer (60 mM PIPES, 25 mM HEPES, 2 mM MgCl₂ and 10 mM EGTA). Cells were further fixed with 2% osmium tetroxide, dehydrated in ethanol and embedded in epoxy resin.⁵⁴ Semi-thin sections (0.8 μm) of embedded cells were stained with 1% toluidine blue and inspected by light microscopy to verify presence of sufficient cells. Cell sections of approximately 1μm thickness were dried on a glass cover slip which was attached to a SEM sample holder (stub) with double-sided copper tape. The sample were then coated with a layer of carbon (50–80 nm) in a DV-502 high vacuum evaporator (Denton Vacuum, Cherry Hill, NJ). A strip of PELCO Colloidal Graphite (Ted Pella, Redding, CA) connected glass cover slip to metal stub for grounding. The SEM sample was then imaged with a Zeiss Supra 55VP Field Emission Scanning Electron Microscope at an acceleration voltage of 2–5kV, and elementarily analyzed by the Energy Dispersive X-ray Spectroscopy (EDAX) module contained in the SEM.

Virus Like Particles

HIV Gag-eGFP VLPs were generated as previously described¹², by transient transfection (calcium phosphate mediated) of HEK293T cells with a pGag–eGFP expression plasmid. VLPs were collected 2 days post transfection, purified from cell debris by filtering through 450nm pore size filters, and stored at −80 °C prior to use.

Dendritic cells and T cells

Primary monocyte-derived DCs were differentiated from CD14⁺ peripheral blood monocytes, and matured with LPS(100ng mL⁻¹) for 2 days prior to use.¹⁰ Primary human CD4⁺ T cells were positively isolated from CD14-depleted peripheral blood mononuclear cells (PBMCs) using CD4-conjugated magnetic beads and LS MACS cell separation columns (Miltenyi Biotech). Positively isolated CD4⁺ T cells were activated with 2% PHA (Invitrogen) for 2 days, washed and cultured in IL-2 (50 U/mL) containing RPMI supplemented with 10% FBS.