Transmission of HIV-1 via intercellular connections has been estimated as 100-1000 times more efficient than a cell-free process, perhaps in part explaining persistent viral spread in the presence of neutralizing antibodies. Such effective intercellular transfer of HIV-1 could occur through virological synapses or target-cell filopodia connected to infected cells. Here we report that membrane nanotubes, formed when T cells make contact and subsequently part, provide a new route for HIV-1 transmission. Membrane nanotubes are known to connect various cell types, including neuronal and immune cells, and allow calcium-mediated signals to spread between connected myeloid cells. However, T-cell nanotubes are distinct from open-ended membranous tethers between other cell types, as a dynamic junction persists within T-cell nanotubes or at their contact with cell bodies. We also report that an extracellular matrix scaffold allows T-cell nanotubes to adopt variably shaped contours. HIV-1 transfers to uninfected T cells through nanotubes in a receptor-dependent manner. These data lead us to propose that HIV-1 can spread using nanotubular connections formed by short-term intercellular unions in which T cells specialize.

We report that two classes of membrane nanotubes between human monocyte-derived macrophages can be distinguished by their cytoskeletal structure and their functional properties. Thin membrane nanotubes contained only F-actin, whereas thicker nanotubes, i.e., those > approximately 0.7 microm in diameter, contained both F-actin and microtubules. Bacteria could be trapped and surf along thin, but not thick, membrane nanotubes toward connected macrophage cell bodies. Once at the cell body, bacteria could then be phagocytosed. The movement of bacteria is aided by a constitutive flow of the nanotube surface because streptavidin-coated beads were similarly able to traffic along nanotubes between surface-biotinylated macrophages. Mitochondria and intracellular vesicles, including late endosomes and lysosomes, could be detected within thick, but not thin, membrane nanotubes. Analysis from kymographs demonstrated that vesicles moved in a stepwise, bidirectional manner at approximately 1 microm/s, consistent with their traffic being mediated by the microtubules found only in thick nanotubes. Vesicular traffic in thick nanotubes and surfing of beads along thin nanotubes were both stopped upon the addition of azide, demonstrating that both processes require ATP. However, microtubule destabilizing agents colchicine or nocodazole abrogated vesicular transport but not the flow of the nanotube surface, confirming that distinct cytoskeletal structures of nanotubes give rise to different functional properties. Thus, membrane nanotubes between macrophages are more complex than unvarying ubiquitous membrane tethers and facilitate several means for distal interactions between immune cells.

We present evidence that nanotubular highways, or membrane nanotubes, facilitate a novel mechanism for intercellular communication in the immune system. Nanotubes were seen to connect multiple cells together and were readily formed between a variety of cell types, including human peripheral blood NK cells, macrophages, and EBV-transformed B cells. Nanotubes could be created upon disassembly of the immunological synapse, as cells move apart. Thus, nanotubular networks could be assembled from transient immunological synapses. Nanotubes were seen to contain GFP-tagged cell surface class I MHC protein expressed in one of the connected cells. Moreover, GPI-conjugated to GFP originating from one cell was transferred onto the surface of another at the connection with a nanotube. Thus, nanotubes can traffic cell surface proteins between immune cells over many tens of microns. Determining whether there are physiological functions for nanotubes is an intriguing new goal for cellular immunology.

NK cells eliminate virus-infected and tumor cells by releasing cytotoxic granules containing granzyme B (GrzB) or by engaging death receptors that initiate caspase cascades. The orchestrated interplay between both cell death pathways remains poorly defined. Here we simultaneously measure the activities of GrzB and caspase-8 in tumor cells upon contact with human NK cells. We observed that NK cells switch from inducing a fast GrzB-mediated cell death in their first killing events to a slow death receptor–mediated killing during subsequent tumor cell encounters. Target cell contact reduced intracellular GrzB and perforin and increased surface-CD95L in NK cells over time, showing how the switch in cytotoxicity pathways is controlled. Without perforin, NK cells were unable to perform GrzB-mediated serial killing and only killed once via death receptors. In contrast, the absence of CD95 on tumor targets did not impair GrzB-mediated serial killing. This demonstrates that GrzB and death receptor–mediated cytotoxicity are differentially regulated during NK cell serial killing.