The study was approved by the institutional review board for biomedical research of Hospital Germans Trias i Pujol (HUGTiP). The participants provided written informed consent to participate in this study.

Peripheral blood mononuclear cells (PBMCs) from HIV-seronegative donors were obtained using Ficoll-Hypaque gradient centrifugation. The monocyte population (>97% CD14⁺) was isolated using CD14-positive magnetic bead–based selection (cat. no. 130-050-201, Miltenyi Biotec). Cells were maintained at 37°C and 5% CO₂ in RPMI medium (cat. no. 11875168, Gibco) supplemented with 10% FBS (cat. no. A5256701, Life Technologies), penicillin/streptomycin at 100 IU/ml (cat. no. 15070-022, Life Technologies), and of granulocyte-macrophage colony-stimulating factor and interleukin-4 at 1000 IU/ml (cat. no. 215-GM-500 and 204-IL-500, R&D) for seven days with replacement of media and cytokines every two days. Activated MDDCs were differentiated by culturing immature MDDCs at day five for a further two days in the presence of 100 ng/ml lipopolysaccharide (LPS, cat.no. L4391-1MG, Sigma-Aldrich) to obtain matured MDDCs.

HEK-293T/17 cells (cat. no. CRL-11268, ATCC repository) were maintained in DMEM (cat.no. 11995073, Gibco) supplemented with 10% FBS and penicillin/streptomycin at 100 IU/ml at 37°C with 8% CO₂. VLP_HIV stocks were generated by transfecting 1E7 HEK-293T/17 cells with 15 µg of pHIV-Gag-eGFP plasmid (cat. no. 11468, NIH AIDS Reagent Program) using 15 µl of LipoD293 (Ver. II) reagent (cat. no. SL100668, SignaGen). After 2 days, supernatants containing VLP_HIV were centrifuged for 5 min at 400 xg, filtered (Millex HV, 0.45 μm; Millipore) and frozen at −80°C until use.

MDDCs were centrifuged, washed with PBS and stained with CellTracker™ Red CMTPX Dye (cat. no. C34552, ThermoFisher). 5E5 cells diluted in 1 ml RPMI were transferred to 35-mm wells for culture imaging (cat. no. 81156, Ibidi) pretreated with poly-L-lysine 20 µg/ml (cat. no. P4832, Sigma-Aldrich). Cells were pulsed with saturating levels of VLP_HIV on ice for 15 min, before being washed with PBS and placed into a microscope stage incubator at 37°C and 5% CO₂. Images were acquired using an Andor Dragonfly 505 spinning disk confocal microscope equipped with an apochromatic 100X/1.49 oil objective and a Sona 4.2 B11 sCMOS camera controlled with Fusion Software. Z-scan mode was set at a 0.2-µm XY pixel size to image the whole cell volume. Laser power was set at a minimum of 0.5% to avoid phototoxicity. Images were acquired every 2 min for 90 min. Eight fields were selected in multiposition mode to maximize the number of imaged cells.

To measure cholesterol coalescence, 5E5 MDDCs were stained for 5 min at room temperature with two lipid probes, either 75 nM of cholesterol-PEG(1k)-Abberior STAR RED dye (35, 36) or 28 nM of DPPE-Abberior STAR RED dye (cat. no. STRED-0200-1MG, Abberior GmbH). Cholesterol-PEG (1 k)-Abberior STAR RED was a gift from Prof. Christian Eggeling (Friedrich Schiller University Jena & Leibniz Institute of Photonic Technology, Jena, Germany). The probe was synthesized by fluorescent labelling of cholesterol-PEG(1000)-NH2 (PG2-AMCS, Nanocs Inc., NY, USA) with the amine reactive dye Abberior STAR RED-NHS (STRED-0002-1MG, Abberior GmbH). Cells were then washed, pulsed with VLP_HIV, and transferred to multiwell plates (cat. no. 81156, Ibidi) for live imaging as described above.

MDDCs stained with CellTracker™ Red CMTPX dye were treated for 2 h at 37°C with 100 nM wortmannin PI3K inhibitor (cat. no. PHZ1301, ThermoFisher), 25 µM 2-bromopalmitate palmitoyl-acyl transferase inhibitor (2-BP; cat. no. 238422, Merck), or 100 µM 5-(N-Ethyl-N-isopropyl) of the specific macropinocytosis inhibitor 5-(N-ethyl-N-isopropyl) amiloride (EIPA; cat. no. A3085, Merck). Then, MDDCs were pulsed with saturating levels of VLP_HIV for 15 min on ice, washed with PBS, resuspended in 1 ml RPMI, and incubated for 6 h at 37°C and 5% CO₂. In parallel, treated cells were incubated with either labelled transferrin at 25 µg/µL (Tfn-Alexa Fluor 488; cat. no. T23366, ThermoFisher) for 30 min or with labelled recombinant cholera toxin subunit B at 1 µg/µl (Ctxβ-Alexa Fluor 488; cat. no. C34775, ThermoFisher) for 1 h. After the indicated incubation time, MDDCs were transferred to glass coverslips coated with poly-L-lysine solution (cat. no. P8920, Merck) and fixed in PFA 4% for 20 min (cat. no. P6148, Merck). PFA was then replaced with PBS-0.5% BSA w/v (cat. no. A966-506, Sigma-Aldrich) and stored at 4°C.

MDDCs were treated for 1 h with permeabilization buffer (5 mg/ml BSA (cat. no. A7906 Merck) and 1 mg/ml saponin (cat. No. SAE0073, Merck). Subsequently, cells were incubated for 1 h with 20 ng/ml of anti-CD169 antibody clone #6H9 (7) and NHS conjugated with Abberior STAR RED-NHS dye (STRED-0002-1MG, Abberior GmbH). When indicated, actin fibers were stained for 1 h with 165 nM Phalloidin-Alexa Fluor 555 (cat. no. A30106, ThermoFisher) after treatment with permeabilization buffer. Nuclei were stained with 300 nM DAPI solution in PBS (cat. no. D1306, ThermoFisher) for 5 min and then washed with PBS. Coverslips were mounted onto glass slides with ProLong Glass Antifade mounting media (cat. no. P36982, ThermoFisher) and sealed with nail polish. Fixed samples were then imaged using Andor Dragonfly 505 spinning disk confocal microscope.

Sac-like compartment diameter was measured by tracing a line with the line tool in Fiji/ImageJ software over the sac-like compartment in a projection of 3 XY slices comprising the sac-like compartment. To determine sac-like compartment volume, we applied the “Volume Calculator” tool in Fiji/ImageJ software using the eGFP channel signal corresponding to VLP_HIV in the sac-like compartment. 3D reconstruction was performed with the plugin “3D viewer” in Fiji/ImageJ software. Orthogonal views in the Z axis were obtained by tracing a line in the XY plane across the cell and applying the Reslice-Z tool in Fiji/ImageJ software.

To distinguish VLP_HIV distribution, we utilized a script based on the circularity of the eGFP channel. To systematize the analysis, we applied the “Subtract Background” and “Convert to Mask” commands to generate a script based on tools from Fiji/ImageJ software, which uses the cell tracker channel to determine the regions of interest (ROI) corresponding to cell boundaries. We set a minimum sac-like compartment size of 1 micron; therefore, cells with no particles were assigned to random binding. We used circularity in “Analyze Particles” to determine whether cells were polarizing VLP_HIV (circularity value = 0 - 0.7) or whether cells formed sac-like compartment (circularity value = 0.7 - 1). Analysis of internalization of Tfn and Ctxβ was based on the cell tracker channel. This was used to determine the ROI and analyze the mean fluorescence intensity of the green channel, which corresponds to these cargoes, inside ROIs. Image analysis scripts are provided in Supplementary Table 1 .

We analyzed cholesterol coalescence by determining an ROI corresponding to the membrane area where VLP_HIV coalesces to measure the mean fluorescence intensity (MFI) of the lipid probe in the VLP_HIV ROI. In parallel, we determined the whole membrane ROI manually to measure the MFI of the lipid probe in the whole membrane. A ratio between the two MFI measured for each ROI was calculated as an indicator of the fold change in the cholesterol probe concentration.

We performed a pixel-by-pixel colocalization test to compare colocalization in VLP_HIV and CD169. This enabled us to calculate the Pearson correlation coefficient for both fluorescence signals. The Pearson correlation coefficient ranges from -1 to +1, with higher values indicating a better correlation.

All the statistical tests were performed with GraphPad software. In the analysis of sac-like compartment size and volume, 25 cells from two different donors were measured. The results are represented using a violin plot graph, indicating median and quartiles. Average and standard deviation were calculated for volume and maximal axial size.

To analyze cholesterol coalescence, 78 cells from three different blood donors were measured. The results are represented using a violin plot graph, indicating median and quartiles. A paired t test was used to analyze the statistical significance of the increases in the cholesterol probe in the VLP_HIV coalescence regions of the cell membrane.

In the experiments with 2-BP, wortmannin, and EIPA inhibitory treatments, groups of at least 100 cells for each condition were counted and analyzed. MDDCs were isolated from 9, 7, and 4 blood donors, respectively. The Wilcoxon test was used to analyze the statistical significance of the changes in the phenotype distribution described for sac-like compartment formation. In control experiments to measure endocytosis of Tfn and Ctxβ upon treatment with 2-BP and wortmannin, groups of at least 100 cells for each condition were counted and analyzed. MDDCs were isolated from seven blood donors. The results are represented using a violin plot graph, indicating median and quartiles. The Wilcoxon test was performed in GraphPad to analyze the statistical significance of changes in the proportion of cells internalizing the cargoes.

Every endocytosis mechanism produces transport intermediates with distinct sizes and morphologies. To investigate which endocytosis mechanism drives sac-like compartment formation, we measured the size, volume, and temporal dynamics of sac-like compartments in activated MDDCs. MDDCs were pulsed with HIV-1 virus-like particles expressing recombinant Gag-eGFP protein (VLP_HIV) and imaged using confocal microscopy. MDDCs incubated in a culture chamber and imaged in vivo ( Figure 1A ) showed behavior similar to that of MDDCs pulsed with VLP_HIV and then fixed ( Figure 1B ), thus validating live cell imaging to study sac-like compartment formation. While the sac-like compartment is not perfectly spherical, we determined that the average of two donors of maximal axial size of the sac-like compartments was 2.9 ± 0.7 μm ( Figure 1C ), and the volume, measured after 3D reconstruction, was found to be 20.0 ± 9.9 μm³ ( Figure 1D ). These findings provide significant insight into the morphology and dimensions of sac-like compartments in MDDCs exposed to VLP_HIV, shedding light on the intracellular dynamics of viral uptake and compartmentalization within DCs.

We conducted live-cell confocal microscopy imaging to determine whether VLP_HIV internalize simultaneously or accumulate after trespassing cell boundaries. Our observations from 3D reconstruction and Z-Slice projection of the corresponding microscope images revealed that bound VLP_HIV remained localized on the outer membrane surface until polarization in a distinct region within a timeframe of less than 5 min. Then, they internalized to form a single compartment, where all VLP_HIV were densely packed together ( Figure 1E ; Supplementary Videos S1 , S2 ).

The formation of this unique compartment involves extensive internalization of VLP_HIV attached to the cellular membrane into a compartment in the cytoplasm. This dynamic imaging approach provides valuable insights into the temporal and spatial dynamics underlying internalization of VLP_HIV, shedding light on the mechanisms of viral entry and compartmentalization within the cell.

To investigate the involvement of actin in sac-like compartment biogenesis, we pulsed MDDCs with VLP_HIV at different time points, fixed them, and stained actin fibers using phalloidin-AF555. This enabled us to measure changes in actin distribution and identify potential intermediate phenotypes during the sac-like compartment formation process. Based on the distribution of VLP_HIV and in the presence or absence of actin, we categorized five different phenotypes ( Figure 2A ). We observed that actin fibers in MDDCs were concentrated within dendrites and that the actin cortex localized beneath the plasma membrane, as compared to the rest of the cytoplasm. Polarization of VLP_HIV was observed in some cells, with actin fibers present in membrane-associated regions, whereas in other cells, the actin cortex was absent in regions displaying polarized VLP_HIV ( Figure 2B ). Some cells with internalized VLP_HIV did not exhibit colocalization with actin, whereas others displayed fully formed sac-like compartment that colocalized with actin fibers ( Figure 2C ).

When we quantified the dynamic incorporation of VLP_HIV and its colocalization with actin fibers in MDDCs from different donors, we observed that VLP_HIV polarized in a single pole phenotype is more abundant than a sac-like compartment fully formed by internalized VLP_HIV at early time points ( Figure 2D ).

The proportion of cells with VLP_HIV internalized in sac-like compartments colocalizing with actin increased from 10% at earlier time points to over 50% of total evaluated cells after 16 h of incubation. This suggests that actin colocalization represents the final stage in the sac-like compartment formation process. Concurrently, the polarized phenotype colocalizing with actin decreased from approximately 50% to less than 20% of total evaluated cells. Interestingly, phenotypes where actin does not colocalize with VLP_HIV were more frequent at intermediate time points, suggesting they are transient states during sac-like compartment formation. Moreover, the orthogonal view of sac-like compartments lacking actin colocalization revealed discontinuities in the actin cortex, forming channel-like structures ( Figure 2E ). Colocalization analysis of CD169 and VLP_HIV in these structures showed a Pearson correlation index of 0.85, indicating a high degree of colocalization degree in both ( Figure 2E ). All sac-like compartments phenotypes exhibited colocalization with CD169 expression ( Figure 2E ), thus reinforcing the role of CD169 in mediating viral uptake. To further validate the abundance of these phenotypes, MDDCs from four additional blood donors were incubated with VLP_HIV for 2h to maximize the visualization of intermediate polarized and internalized phenotypes ( Supplementary Figure S1 ).

In conclusion, these data suggest that the actin cytoskeleton plays a role in the early stages of polarization of VLP_HIV, making it crucial to preserve the structural integrity of sac-like compartments in the final phase. The data also indicates that rearrangement of cortical actin accompanies internalization of HIV at early stages of sac-like compartment formation, suggesting the involvement of a MEND-like mechanism.

We subsequently used live confocal microscopy with a fluorescent cholesterol analogue probe to investigate whether the MEND pathway accompanied by cholesterol coalescence could contribute to sac-like compartment formation ( Figure 3 ). Remarkably, the fluorescent signal of the cholesterol analogue increased notably in cell membrane regions containing VLP_HIV, exhibiting both polarized and internalized distribution phenotypes ( Figure 3A ). We quantified the fluorescence intensity of this cholesterol probe in membrane regions where viral particles accumulated before sac-like compartment formation and compared it to the average fluorescence intensity of the cellular membrane. Our analysis focused solely on cells displaying the polarization phenotype, as sac-like compartment formation involves membrane folding that enhances the cholesterol fluorescence signal ( Figure 3B ). We observed that the cholesterol fluorescence intensity at regions where sac-like compartments were forming was twice as high as in the rest of the membrane. This trend persisted over time, confirming the redistribution and coalescence of cholesterol, with accumulation of VLP_HIV in the cellular membrane ( Figure 3C ; Supplementary Video S3 ).

As a control, we simultaneously used a fluorescent dipalmitoyl phosphatidylethanolamine analogue probe (DPPE) to assess the lipid bilayer independently of cholesterol lipid rafts ( Figure 3D ). In contrast to the cholesterol analogue, this lipid molecule exhibited a uniform distribution in the cell membrane, with no localized increase in membrane regions where VLP_HIV accumulated.

These findings demonstrate that cholesterol coalescence in the cell membrane is associated with the accumulation of VLP_HIV prior to sac-like compartment formation, supporting the notion that MEND mechanisms play a crucial role in sac-like compartment formation in myeloid cells.

In accordance with the current definition of MEND, protein palmitoylation initiates the mechanism that facilitates protein coalescence and activates MEND. To assess the significance of protein palmitoylation in sac-like compartment formation in MDDCs, we used the palmitoylation inhibitor 2-bromopalmitate (2-BP). 2-BP abrogates the internalization of VLP_HIV into the sac-like compartment, halting them at the polarization stage on the cell membrane ( Figure 4A ). The potential impact of 2-BP on conventional endocytosis pathways was also investigated by separately pulsing MDDCs with transferrin (Tfn) and cholera toxin subunit β (Ctxβ)–labelled reporters, which are commonly used cargoes for endocytosis assays. No obvious differences were observed in the cell uptake of Tfn or Ctxβ after 2-BP treatment, thus ruling the role of palmitoylation in this endocytosis process ( Figure 4A ). Interestingly, zoomed-in representative cells revealed that VLP_HIV failed to complete internalization upon treatment with 2-BP. Additionally, CD169 expression remained unaltered following treatment with 2-BP, indicating that arrest of uptake cannot be attributed to a lack of CD169 ( Figure 4B ).

Quantitative data demonstrated that the proportion of MDDCs arrested at the polarization stage increased from 25% to 60% following treatment with 2-BP compared to untreated MDDCs ( Figure 4C ). Inversely, treatment with 2-BP drastically reduced sac-like compartment formation (25%) compared to controls (60%), arresting the VLP_HIV uptake process at the polarization stage ( Figure 4C ). MDDCs treated with 2-BP were concurrently pulsed with Tfn- and Ctxβ-labelled reporters to evaluate any potential side effects on classical endocytosis pathways ( Figure 4D ). We observed that the internalization patterns of transferrin remained unaltered upon treatment and that internalization of Ctxβ is minimally affected compared with sac-like compartment formation.

In conclusion, these data suggest that sac-like compartment formation is dependent on protein palmitoylation, which is another essential regulatory mechanism for MEND.

To assess the reliance of sac-like compartment formation on (PI(4,5)P₂) phosphorylation by PI3K, MDDCs from various donors were treated with the PI3K paninhibitor wortmannin prior to the VLP_HIV pulse. Wortmannin halts sac-like compartment formation at the VLP_HIV polarization stage in the cell membrane ( Figure 5A ). To exclude side effects of wortmannin on conventional endocytosis mechanisms, cell uptake of Tfn or Ctxβ was assessed as described above, with no evident differences observed for either after treatment with wortmannin ( Figure 5A ). This incomplete VLP_HIV polarization phenotype is characterized by a semicircular distribution of VLP_HIV, which do not converge at a single membrane region for internalization. Importantly, treatment with wortmannin does not interfere with either CD169 expression or the attachment of VLP_HIV ( Figure 5B ).

Quantitatively, the proportion of cells arrested at the polarization stage after treatment with wortmannin increased from 30% to 50% ( Figure 5C ). Moreover, the percentage of MDDCs with internalized VLP_HIV forming sac-like compartments decreased from over 60% to 40%, halting the VLP_HIV uptake process at the polarization stage. As a control, we evaluated internalization of Tfn and Ctxβ, which was comparable between wortmannin-treated and untreated MDCCs, indicating that wortmannin does not impair conventional endocytosis pathways in this cellular model ( Figure 5D ).

As PI3K metabolism also plays an important role in macropinocytosis, we used the specific macropinocytosis inhibitor 5-(N-ethyl-N-isopropyl) amiloride (EIPA) to evaluate whether the effect observed by wortmannin could be explained by inhibition of macropinocytosis ( Figures 5E, F ). EIPA functions by inhibiting Na⁺/H⁺ exchangers, thereby preventing the activation of intracellular macropinocytosis regulators. Pretreatment with EIPA did not show any significant effect on sac-like compartment formation, confirming that macropinocytosis is not involved in sac-like compartment formation.

Collectively, these findings suggest that when PI3K metabolism is inhibited, internalization of virus in MDDCs is diminished at the polarization stage. This inhibition of PI3K has been reported to be an additional regulatory mechanism of MEND.