|Home | About | Journals | Submit | Contact Us | Français|
Natural killer (NK) cells eliminate cancer and virus-infected cells through cytolytic activity. The last step in NK cell cytotoxicity, resulting in exocytosis of granule content, requires fusion of lytic granules with the plasma membrane. Proteins from the SNARE family mediate membrane fusion events in the cell. Here we show that NK cells express all members of the R-SNARE subgroup. Two of these R-SNARE proteins, VAMP4 and VAMP7, co-localize with lytic granules during cytotoxic interactions. However, only VAMP7 associates with perforin-containing granules in non-activated cells, indicating that the two VAMPs have different functions in exocytosis. Using both the tumor NK cell line, YTS, and peripheral NK cells we show that disruption of expression of either VAMP4 or VAMP7 inhibits release of lytic granules and severely impairs NK cell cytotoxic activity. Furthermore, VAMP7 but not VAMP4 is involved in IFNγ secretion in NK cells, indicating that VAMP7 is involved in many fusion processes and thus plays a more general function in NK cell activity than VAMP4.
Natural killer (NK) cells are able to directly remove pathogens through their cytotoxic activity and contribute to the adaptive immune response through production of cytokines and chemokines, as well as activation of dendritic cells and macrophages [1, 2].
Activity of NK cells is regulated by a variety of germline-encoded inhibitory or activating receptors. Ligation of inhibitory receptors negatively regulates NK cell activity. Engagement of activation receptors results in activation of NK cells and triggers a complex response leading to the death of a target cell [3, 4]. Killing is achieved by localized secretion from lytic granules containing perforin and granzymes at the cell-cell contact site, known as the immunological synapse [5–9].
Membrane fusion events are mediated directly by soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) proteins [10–12]. Based on the presence of arginine or glutamine in the central position of the SNARE motif, SNARE proteins are divided into two subgroups: R-SNAREs found on vesicular membranes and Q-SNAREs found on target membranes. Functional SNARE complexes require 4 SNARE motifs and are comprised of one R-SNARE and two or three Q-SNARE proteins [10, 12]. They play critical roles in secretion of inflammatory mediators, cytokines as well as phagocytosis in many cells of the immune system, including granulocytes, platelets, plasma cells and macrophages , and thus are perfect candidates to function in release of the lytic granules. From almost 40 SNARE proteins expressed in humans only seven belong to the R-SNARE group. Here, we show that human NK cells express all known R-SNARE proteins. Two of the R-SNAREs, vesicle associated membrane protein VAMP4 and VAMP7, co-localize with perforin-containing lytic granules in activated cells. However, in resting cells they have distinct localizations on different vesicles: only VAMP7 co-localizes with lytic granules. Both VAMP4- and VAMP7-expressing vesicles are essential for proper NK cell activity and play a role in the release of lytic granules. Additionally, VAMP7-positive vesicles, in contrast to VAMP4-containing vesicles, appear to have a broader function in NK cell activity.
To examine which SNARE proteins might be involved in lytic granule exocytosis in NK cells this study focused on vesicle-associated R-SNARE proteins. Expression of different R-SNARE proteins was evaluated in NK cells. RT-PCR using the total RNA isolated from either ex vivo NK cells or the tumor NK cell line, YTS, revealed that NK cells express all seven known human R-SNARE proteins (Fig. S1).
To investigate whether a particular R-SNARE protein co-localized with lytic granules, the cellular localization of different R-SNARE proteins in relation to perforin was studied next. VAMP7 and VAMP4 showed substantial co-localization with lytic granules during NK cell cytolytic interactions (Fig. 1A and C). Interestingly, while perforin showed very strong co-localization with VAMP7 (Manders' co-localization coefficient M=0.81 for YTS and M=0.80 for NK), the ability of VAMP4 to overlap with perforin was less pronounced (YTS: M=0.67, NK: M=0.59; Fig. 1A, C). Although VAMP4 was present on a substantial portion of lytic granules during cytolytic interactions, granules clearly devoid of VAMP4 were also present, often close to the immune synapse area (Fig. 1A, C inserts and Fig. 3B), suggesting that VAMP4, contrary to VAMP7, is not constantly associated with lytic granules. Indeed, VAMP7 co-localized with lytic granules in NK cells in the absence of target cells (Fig. 1B, D). Co-localization of VAMP7 with the majority of perforin during cytotolytic and non-cytolytic interactions is in agreement with and expands previous data showing VAMP7 and granzyme B co-localization in resting cells [13, 14]. Surprisingly, VAMP4 showed a very low level of association with lytic granules in the majority of non-activated NK cells and its ability to overlap with perforin was very low (YTS: M=0.17, NK: M=0.13; Fig. 1B, D). Thus, while VAMP7 is likely a constitutive component of the lytic granules VAMP4 may interact with lytic granules at some stage(s) of exocytosis.
The localization of both proteins in relation to each other was also investigated (Fig. S2). Without the target cells, the overlap between VAMP4 and VAMP7 was minimal. VAMP4 and VAMP7 showed co-localization in YTS cells activated by mixing with target cells, mostly in the area adjacent to the immune synapse indicating that both VAMP4 and VAMP7 can be present on lytic granules at the same time. The degree of co-localization between VAMP4 and VAMP7 in the other regions of the NK cells was limited and both proteins displayed areas specific only for one of them. Thus, while pools of VAMP4 and VAMP7 exist in separate cellular compartments, they are also able to share location on the lytic granules.
In view of the present and earlier findings [11, 12], it is unlikely that VAMP4 and VAMP7 are components of the same functional SNARE complex. The data suggest the hypothesis that VAMP7 may be involved in the terminal event of fusion of lytic granules with the cell membrane, while VAMP4 engages in fusion of vesicles containing different components of cytotoxic granules. VAMP4 had been shown to associate with the trans-Golgi network (TGN) and mediate maturation of secretory granules in endocrine and exocrine cells [15–17].
VAMP1, VAMP3 or VAMP8 did not display strong co-localization with perforin (Fig. 2), indicating that these proteins were likely not involved in regulation of exocytosis in NK cells. In contrast to observations reported here, VAMP3 was shown to accumulate at the IS in activated T cells, likely because of its involvement in TCR trafficking . Also, VAMP8 has been shown to be involved in granule exocytosis from T cells in mice  and from mast cells  but not HeLa cells . The difference in observations is likely due to cell-specific differences. The present experiments suggest that VAMP8 is not involved in exocytosis in human NK cells, although it could be involved in this process in T cells.
To assess the role of the two SNARE proteins in NK cell activity, RNAi was used to disrupt VAMP4 or VAMP7 expression in YTS NK cells. Introduction of VAMP4 or VAMP7 RNAi caused specific reduction of, respectively, VAMP4 or VAMP7 expression and severely inhibited YTS NK cell cytotoxicity toward tumor target cells (Fig. 3A and S4). The knock-down of VAMP4 or VAMP7 had no effect on cell-cell conjugation (Fig. S3) and polarization of lytic granules toward the immunological synapse (Fig. 3B), demonstrating that the defect in cytotoxicity was not due to impaired adhesion or granule transport to the cell-cell interface.
Since SNARE proteins mediate fusion of vesicles with the plasma membrane, the effect of disruption of SNARE expression on the ability of NK cells to release lytic granules was investigated next. Analysis of the induced expression of CD107a on the cell surface of YTS cells after mixing with target cells revealed that both VAMP4 and VAMP7 RNAi caused downregulation of CD107a expression on the cell surface (Fig. 3C and S5). Knock-down of either VAMP did not affect the expression of CD107a in YTS cells (Fig. S6A). These results demonstrate that not only VAMP7, but unexpectedly also VAMP4 is critical for NK cell degranulation and without either of these proteins NK cells are not able to efficiently exocytose their lytic granules.
NK cells are known for their ability to produce cytokines and secrete large amounts of IFNγ. IFNγ production was assessed to investigate whether VAMP4 and VAMP7 are required in secretion of this cytokine, in addition to granule release. VAMP7 RNAi cells showed decreased production of IFNγ in response to different stimuli, while knock-down of VAMP4 had no effect on secretion of IFNγ (Fig. 3D). None of the cells had decreased levels of intracellular IFNγ (Fig. S6B), indicating that the ability of RNAi-transfected cells to synthesize IFNγ was not affected. Thus, VAMP7 but not VAMP4 contributes to IFNγ secretion in NK cells.
Our results indicate that VAMP4 plays a specialized role in NK cell activity and is required at some step in the release of lytic granules. Disruption of expression of VAMP4 by RNAi rendered NK cells unable to kill target cells. This defect in cytotoxicity was not due to impaired granule transport to the cell-cell interface, as perforin polarized normally to the immunological synapse. Rather, knock-down of VAMP4 caused a significant decrease in NK cell degranulation, indicating the importance of VAMP4 in lytic granule release in NK cells, and providing the likely explanation for inhibited cytotoxicity of VAMP4 RNAi cells. This finding is the first demonstration of an essential role of VAMP4 in lytic granule release in NK cells, and contributes to the idea that VAMP4 is involved not only in intracellular transport but also plays an important role in regulated exocytosis, as recently evidenced by demonstration of the role of VAMP4 in regulated exocytosis of enlargosomes in neuroendocrine cells . Knock-down of VAMP7 had a similar effect on NK cell degranulation and cytotoxicity. However, the involvement of VAMP7 in secretion of different types of granules indicates that this R-SNARE protein is involved in multiple processes and plays a broader function in NK cell activity than VAMP4. These data, together with other reports showing VAMP7 participation in secretion of mediators from multiple granules in human granulocytes  and involvement in phagocytosis in macrophages , demonstrate the widespread function of this protein in hematopoietic cells.
This report provides an important step in understanding the mechanisms underlying the last, effector stage of NK cell activity, by highlighting the role of VAMP4 and VAMP7, two R-SNARE proteins indispensable for proper NK cell function.
Detailed description of the experimental procedures, antibodies and reagents used is provided in Supplementary Information. The following is a brief description.
Human NK, YTS and 721.221 were cultured as described previously . The VAMP RNAi YTS cells were grown in RPMI1640 medium with puromycin (2 µg/ml).
Vector-based RNAi was used to mediate knock-down of VAMP4 and VAMP7. Oligos designed to produce either VAMP4 or VAMP7 or an appropriate scramble siRNA hairpin were inserted into the lentiviral pLKO1.puro vector. Stable VAMP4 or VAMP7 knock-down cell lines were maintained as described above.
Total RNA was isolated from cells using the PureLink Micro-to-Midi RNA purification system (Invitrogen), and used in a RT-PCR reaction (SuperScript III One-Step Sysytem; Invitrogen). PCR products were verified by sequencing.
For immunoblotting, total cell lystates were resolved on 10 – 20% PAGEr Gold gels, transferred to a PVDF membrane and followed by immunoblotting with anti-VAMP4 antibody.
YTS cell cytotoxicity was evaluated by the 51Cr release assay as described previously .
YTS cell degranulation ability was evaluated by the expression of CD107a on the cell surface using flow cytometric analysis. The specific percentage of CD107a+ cells from the CD56+ population was calculated using the following formula: (% CD107a+ stimulated CD56+ cells) – (% CD107a+ unstimulated CD56+ cells).
The ability of YTS cells to secrete IFNγ was determined using Human IFNγ ELISA Ready-SET-Go kit.
Cells stained with appropriate antibodies were visualized by a Zeiss LSM 510 laser-scanning confocal microscope. Co-localization of VAMP proteins and perforin was assessed using BioImage XD v0.9 software . Co-localization coefficients according to Manders were chosen because of their representation of the true degree of colocalization .
We thank Prachi Trivedi and Matthew Nicotra for reading the manuscript and helpful discussions. This work was supported by NIH research grant RO1 AI050207-06.
CONFLICT OF INTEREST
None of the authors has a conflicting interest.