This story was originally published in PLOS One Journal:
URL: https://journals.plosone.org.
This content has not been altered
Licensed under Creative Commons Attribution (CC BY) license .
url:https://journals.plos.org/plosone/s/licenses-and-copyright
(C) PLOS One [1]
------------



Degranulation enhances presynaptic membrane packing, which protects NK cells from perforin-mediated autolysis

['Yu Li', 'Department Of Pathology', 'Immunology', 'Baylor College Of Medicine', 'Houston', 'Texas', 'United States Of America', 'Department Of Pediatrics', 'Vagelos College Of Physicians', 'Surgeons']

Date: None

Natural killer (NK) cells kill a target cell by secreting perforin into the lytic immunological synapse, a specialized interface formed between the NK cell and its target. Perforin creates pores in target cell membranes allowing delivery of proapoptotic enzymes. Despite the fact that secreted perforin is in close range to both the NK and target cell membranes, the NK cell typically survives while the target cell does not. How NK cells preferentially avoid death during the secretion of perforin via the degranulation of their perforin-containing organelles (lytic granules) is perplexing. Here, we demonstrate that NK cells are protected from perforin-mediated autolysis by densely packed and highly ordered presynaptic lipid membranes, which increase packing upon synapse formation. When treated with 7-ketocholesterol, lipid packing is reduced in NK cells making them susceptible to perforin-mediated lysis after degranulation. Using high-resolution imaging and lipidomics, we identified lytic granules themselves as having endogenously densely packed lipid membranes. During degranulation, lytic granule–cell membrane fusion thereby further augments presynaptic membrane packing, enhancing membrane protection at the specific sites where NK cells would face maximum concentrations of secreted perforin. Additionally, we found that an aggressive breast cancer cell line is perforin resistant and evades NK cell–mediated killing owing to a densely packed postsynaptic membrane. By disrupting membrane packing, these cells were switched to an NK-susceptible state, which could suggest strategies for improving cytotoxic cell-based cancer therapies. Thus, lipid membranes serve an unexpected role in NK cell functionality protecting them from autolysis, while degranulation allows for the inherent lytic granule membrane properties to create local ordered lipid “shields” against self-destruction.

Despite there being a number of potential mechanisms contributing to NK cell resistance to perforin, the question as to how they are protected from autolysis upon degranulation remains unanswered. In light of the charge-sensitive nature of hydrophobic interactions, we considered particular properties of the presynaptic membrane that might make NK cells more resistant to perforin upon degranulation. We demonstrate that perforin binding to NK cell membranes is inhibited by them having high lipid ordering caused by the coalescence of high order lipid membrane domains (or rafts—a known property of the presynaptic membrane) [ 18 – 20 ]. Furthermore, we identify a specific feature of lytic granules in that they have endogenously densely packed membranes. This further enhances lipid packing and potential protection against perforin at the specific sites of degranulation where NK cells face the highest perforin concentrations. Additionally, we find that an aggressive breast cancer cell line utilizes the same strategy to evade NK cell–mediated killing by assembling a densely packed postsynaptic membrane to block perforin attack. Disrupting these protective membrane packing mechanisms reverses NK cell protection against autolysis and, in the case of the breast cancer cell line, effectively switches it to an NK-susceptible state.

Protection from perforin permeabilization is not appreciated as an intrinsic property of NK cell membranes. Early studies have demonstrated fratricide between NK cells, implying that protection from perforin is not a default but is potentially acquired during the target killing process [ 12 ]. Several hypotheses have been proposed to explain this relative and potentially augmented resistance of NK cells during target engagement. These include lytic granule–derived cell surface cathepsin B providing self-protection for cytotoxic lymphocytes [ 13 ]. However, the universal protective function of cathepsin B is questionable since CTLs from cathepsin B knockout mice have normal survival post-degranulation, indicating that this protease is not exclusive for perforin inactivation [ 14 ]. Another hypothesis has involved a protective role for LAMP-1, frequently used as a marker for degranulation, which has been shown to facilitate the delivery of perforin to the target cell membrane [ 15 ]. It was proposed that the steric hindrance and negative charges of heavily glycosylated LAMP-1 could prevent the binding of perforin to the membrane [ 16 ]. Recent work [ 17 ], however, has demonstrated that under some circumstances, LAMP-1 is dispensable for protection against perforin-mediated autolysis and has shown the relevance of the cytotoxic T cell membranes themselves. Thus, a protective mechanism relying on membrane-bound proteins might be contributory to resistance in some cases but possibly locally overcome by secreted perforin under other circumstances.

Due to the nonspecific nature of the hydrophobic interaction between perforin and a target cell membrane, the NK cell membrane theoretically shares the same vulnerability as that of target cell. That said, NK cells are generally not susceptible to the released cytotoxic contents of their lytic granules. Even though both the NK and the target cell membranes are exposed to perforin within the confined space of synaptic cleft, less than 5% of NK cells undergo autolysis during target engagement [ 9 ]. This unidirectional killing is presumably necessary for the immune system to preserve NK cells within a diseased environment for additional killing and host defense. NK cells are quite effective in serial killing and on average can mediate approximately 6 separate kills within a 24-h period [ 10 ]. It is presently unclear, however, as to why NK cells maintain a survival advantage upon their degranulation relative to that of the target cells. That said, perforin-mediated pore formation does occur predominantly in the target cell membrane [ 11 ]. Thus, there must be a membrane-protective mechanism of some kind to ensure the resistance of NK cells to their own perforin upon its release into the synaptic cleft.

Among the destructive molecules contained within the lytic granules, perforin serves a critical nonredundant function and is an absolute requirement for NK cell cytotoxicity [ 6 ]. Perforin is a pore-forming protein that penetrates the target cell membrane and allows for the uptake of proapoptotic serine proteases, such as granzymes A and B that are also extruded into the synaptic cleft from the lytic granule. Although perforin is a potent pore-forming effector, its activity is strictly inhibited inside the lytic granule owing to the fact that the internal environment of the granule is both acidic and hypocalcemic [ 7 ]. Once secreted into the nonacidic synaptic cleft, perforin attains its functionality in membrane attack via its calcium-activated C2 domain, which mediates the hydrophobic interaction between the perforin molecule and a target membrane [ 8 ].

Immune defense against intracellular pathogens and tumors by cytotoxic effector cells, which include natural killer (NK) cells and cytotoxic T lymphocytes (CTLs), is fundamental to human health. NK cells in particular have an inherent ability to recognize and destroy infected and transformed cells. They express high levels of cytolytic effector molecules without the need for clonal expansion and use these to kill diseased cells. Upon recognition of a target, an NK cell establishes an immunological synapse to initiate a series of tightly regulated downstream events [ 1 ]. These include the mobilization of lysosome related organelles, known as lytic granules that are specialized for secretion. During the progression of the synapse, these lytic granules initially converge to the microtubule organizing center (MTOC) and then polarize along with the MTOC to the synapse formed with the target cell [ 2 ]. Once delivered to the synapse, the lytic granules navigate a meshwork of filamentous actin and find openings through which they may access the NK cell plasma membrane [ 3 ]. This allows the lytic granules to dock and fuse with the plasma membrane and then utilize myosin-II to extrude their cytotoxic contents into a protected synaptic cleft between the NK cell and target cell [ 4 , 5 ]. Therein, the lytic effector molecules can affect the destruction of the target cell.

Results

NK cells assemble presynaptic membranes with high lipid order during target cell engagement NK cells form immunological synapses with target cells to initiate a series of tightly regulated downstream events that eventually release perforin and other cytolytic molecules into synaptic clefts [21]. Despite the close range between NK and target cell membranes defining the synaptic cleft, perforin-mediated pore formation predominantly occurs in the target cell membrane [11]. Ultimately, this results in the death of the target and survival of the NK cell upon its degranulation. In considering this unidirectional attack of perforin, it is interesting that the membrane targeting of perforin to the target cell does not rely on any particular receiving ligand and is solely mediated by hydrophobic interaction [22]. Since these hydrophobic interactions are charge sensitive, we postulated that coalescence of high order lipid membrane domains, which can greatly affect the charge distribution on presynaptic membrane [23], might prevent perforin binding. If true, NK cells could form high lipid order membranes to inhibit perforin activity and allow for their unidirectional attack against target cells. To test our hypothesis, we first attempted to measure the lipid ordering of the NK cell membrane in living cells using 2 fluorescent sensors: Di-4-ANEPPDHQ and Laurdan, whose emission spectra are dependent upon lipid packing (S1A and S1B Fig). Because these 2 sensors take advantage of different mechanisms to probe lipid packing density [24] and were applied across multiple NK–target cell pairs, our measurements should be technically reduced in bias and also not dependent upon any particular cell line. The lipid packing of NK cell membranes was quantified using previously established ratiometric analysis [25] and represented as generalized polarization (GP) values (S1C and S1D Fig). The lipid packing density of the entire NK cell plasma membrane was initially measured and specifically evaluated relative to the site of contact with target cells. NK cells (YTS and NK92) were prelabeled with sensors and then conjugated to target cells (721.221 and K562, respectively) and imaged using live cell confocal microscopy. In both YTS and NK92 cells, more densely packed membranes were observed at the immunological synapse, while the nonsynaptic regions displayed relatively lower packing (consistent with previous findings using indirect methods or in fixed cells) [20,26]. The increased membrane lipid packing at the synapse was significant for both YTS and NK92 cells across repeated experiments (Fig 1A–1C). Nonsynaptic regions had a GP value that was similar to that of the plasma membranes in unconjugated NK cells (S2A and S2B Fig), indicating that the uneven distributions of membrane packing in conjugated NK cells was likely due to an increase of lipid ordering at the synapse. This is also consistent with previously published results from CTLs and antigen-presenting cells (APCs) showing lipid raft aggregation at the immunological synapse [27]. Thus, during target cell engagement, NK cells form a presynaptic membrane characterized by having high lipid order. PPT PowerPoint slide

PowerPoint slide PNG larger image

larger image TIFF original image Download: Fig 1. NK cells assemble presynaptic membranes with high lipid order during target cell engagement. The lipid packing of NK cell membranes in fluorescent packing sensor labeled YTS (A, D) and NK92 (B, E) conjugated with calcein green–loaded 721 or K562 target cells, respectively, was measured using live cell confocal microscopy (left image). 7KC-treated and untreated NK cells were overlaid with a pseudocolor scale (A, B, D, E–right image) to demonstrate differences in packing. Membrane lipid density was quantified as a GP value, which was measured in distinct conjugates (n = 12 for untreated YTS, n = 14 for 7KC-treated YTS, n = 12 for untreated NK92, n = 12 for 7KC-treated NK92) performed over 4 independent experiments and the images shown were representative. The individual GP values for synaptic and nonsynaptic regions were compared as paired observations from individual cells in aggregate for YTS (C left, p < 0.001) and NK92 (C right, p < 0.001) cells using a paired t test and the difference was significant, but not when YTS (F left, p > 0.05, ns) and NK92 (F right, p > 0.05, ns) cells were treated with 7KC. The lipid packing of the NK cell presynaptic membrane was further measured using TIRF microscopy on activating glass surfaces (anti-CD18 with anti-CD28 for YTS), and 4 representative images from untreated (G) and 7KC-treated YTS cells (H) were overlaid with a pseudocolor scale to allow visualization of packing differences (G vs H). Packing density was quantified as a GP value in 32 distinct cells from 4 independent experiments, and an unpaired t test was used for the comparison of the means (I) and the difference was significant p < 0.0001. Scale bar: 10 μm. GP, generalized polarization; NK, natural killer; ns, not significant; TIRF, total internal reflection fluorescence; 7KC, 7-ketocholesterol. https://doi.org/10.1371/journal.pbio.3001328.g001 To evaluate the potential role of increased lipid ordering in the NK cell presynaptic membrane, we utilized 7-ketocholesterol (7KC). 7KC is effective in disrupting lipid packing and functions by steric hindrance resulting from an additional ketone group it possesses. It has been shown to be effective in cytotoxic lymphocytes as it can disrupt lipid ordering in the plasma membrane of CTL cells without interfering other biological properties such as viability and proximal signaling [28]. In order to modulate membrane packing density in NK cells, we pretreated NK cells with 7KC, labeled them with lipid packing sensors, and then conjugated them to target cells. We then measured their membrane packing distribution during target engagement using live cell confocal microscopy. Using this approach, 7KC treatment reduced the presence of lipid ordering at the immunological synapse (Fig 1D–1F). Furthermore, the membranes of NK cells in conjugates displayed evenly distributed lipid packing after 7KC treatment. To further study the lipid packing on the presynaptic membrane with higher spatial resolution, en face images of the immunological synapse were acquired using total internal reflection fluorescent (TIRF) microscopy. TIRF provides increased contrast by confining sample illumination to a depth of approximately 150 nm from the plane of the coverslip. NK cells were lipid sensor prelabeled and seeded on glass surfaces coated with antibodies against activating receptors to generate immunological synapses at the plane of the glass and visualized using TIRF imaging. NK cells had densely packed membranes at the glass surface, corresponding to the presynaptic membrane, indicated by the high GP values (Fig 1G and 1I). Importantly, this high degree of ordering was completely eliminated after 7KC pretreatment of the NK cells (Fig 1H and 1I). Thus, during target cell engagement, NK cells form a densely packed presynaptic membrane, which can be effectively disrupted by 7KC.

7-Ketocholesterol induces NK cell death during target killing Since NK cells demonstrate a highly ordered synaptic lipid membrane, we asked if it could serve a role in NK cell survival during the cytolytic process. Thus, we next evaluated the influence of lipid packing on NK cell survival during cytotoxicity. NK cells were pretreated with or without 7KC, 51Cr labeled and incubated with paired target cells for 4 h, after which the survival of NK cells was evaluated via the release of 51Cr from dying NK cells. 7KC treatment of NK cells did not affect the lytic capability of NK cells, and both YTS and NK92 cells killed their respective target cells similarly to those not treated with 7KC (Fig 2A and 2B). 7KC-pretreated YTS and NK92 cells, however, had significantly elevated cell death rates compared to those in untreated cells (Fig 2A and 2B). The similarity of finding between 2 disparate NK cell lines suggests that the finding represents a consistent property of NK cells. That said, we also evaluated freshly isolated ex vivo human NK cells. As was found with the NK cell lines, 7KC pretreatment did not affect the lytic capability against, but did also lead to increased death of the NK cells during incubation with susceptible target cells (Fig 2C, individual donors shown separately in S3 Fig). Thus, 7KC induces NK cell death during target cell exposure, while not compromising the ability of NK cells to recognize and destroy a target cell. This suggests that lipid ordering in NK cells is required for their optimal survival during cytotoxicity. PPT PowerPoint slide

PowerPoint slide PNG larger image

larger image TIFF original image Download: Fig 2. 7KC treatment of NK cells induces NK cell death during target killing. Cytotoxic function (A, left, ns) and survival (A, right, p < 0.05) of untreated or 7KC-pretreated YTS cells incubated with 721.221 target cells were measured using 4-h 51Cr release assay. Similarly, cytotoxic function (B, left, ns) and survival (B, right, p < 0.05) of untreated or 7KC-pretreated NK92 cells incubated with K562 cells were measured by 4-h 51Cr release assay. To measure cytotoxic function in each case, target cells were 51Cr labeled (A, B, left), whereas to measure survival of NK cells, the NK cells themselves were 51Cr labeled (A, B, right), and for both, the 51Cr release was measured after 4 h of incubation. To validate the elevated death of 7KC-pretreated YTS cells, the cytotoxic function (C, left, ns) and survival (C, right, p < 0.05) of primary human purified eNK cells were measured after their labeling with 51Cr and when incubated with K562 cells with or without 7KC pretreatment. A correlation between NK cell survival and cytotoxicity after 7KC treatment was evaluated by varying the duration of incubation for assays measuring cytotoxicity (D) and survival (E) of 7KC-pretreated YTS cells incubated with 721.221 target cells using 51Cr-labeled NK or target cells, respectively. All values presented represent averages of 3 independent experiments (NK cell lines) or 5 distinct healthy donors and error bars display ± SD (A−E). Wilcoxon signed rank test was used for the comparison between curves. eNK, ex vivo NK; NK, natural killer; ns, not significant; 7KC, 7-ketocholesterol. https://doi.org/10.1371/journal.pbio.3001328.g002 Although the standard time frame used for an in vitro cytotoxicity assay is 4 h, target cell killing occurs within much shorter time frames and also continues after the 4-h time point. This represents a combination of NK cell access to target cells as well as eventually serial killing activity (which would theoretically benefit from NK cell survival). By varying the time of the cytotoxicity assay using 7KC-treated NK cells, we wanted to ask if the extent of target cell killing was related to that of autolysis. Thus, 7KC-pretreated YTS cells were incubated with 721.221 cells for different durations. The cytotoxicity and survival of YTS cells were measured based on the release of 51Cr from 721.221 cells and YTS cells, respectively. As expected, the killing of 721.221 target cells increased proportionally to the time of incubation with YTS cells in the cytotoxicity assay (Fig 2D; with the exception of the 16-h time point, which we presume to be a feature of extended 51Cr labeling of 721.221 cells). The death of NK cells, however, increased markedly during extended coincubations with target cells and was generally increased in concert with the time of exposure to 721.221 target cells (Fig 2E). The parallel increases in target cell exposure and NK cell death in these cytotoxicity assays could imply a degranulation-coupled mechanism underlying the 7KC-induced NK cell autolysis. To validate the role of lipid packing and further rule out any unknown effect of 7KC, we also attempted an alternative experimental approach to manipulate membrane properties. We utilized Nystatin, which disrupts cholesterol-rich microdomains on plasma membrane without removing cholesterol from the membrane [29]. Thus, its effect would be expected to be distinct from 7KC in changing lipid composition. YTS cells were treated with Nystatin in varying concentrations and then evaluated for their ability to kill 721.221 target cells or for their survival when paired with 721.221 cells. Albeit with lower molar efficiency compared to 7KC, Nystatin treatment exhibited dose-dependent decreases in NK cell survival without affecting their ability to kill 721.221 target cells (S4A Fig). This extends the findings using 7KC and suggests a vital role for lipid packing in NK cell self-protection.

7KC-induced NK cell death requires target cell-induced NK cell degranulation Although 7KC has been reported to have no significant impact on T cell viability and signaling [28], we wanted to be certain that NK cell death was not due to any intrinsic toxicity of 7KC. Thus, YTS cells were pretreated with different concentrations of 7KC, and then cultured alone or with 721.221 cells for 4 h. Over a wide range of 7KC concentrations, the viability of YTS cells in the absence of target cells was unaffected by the drug or their activation state (Figs 3A, S4B and S4C). Other routine parameters such as cell size, number and performance in experiments were not affected by treatment. In the presence of target cells, however, 7KC caused a dose-dependent NK cell death (Fig 3B). Thus, it is likely that the completeness of blockade of lipid ordering in the NK cell presynaptic membrane is directly related to the degree of autolysis. Furthermore, this indicated that the death of 7KC-pretreated NK cells was linked to the presence of target cells and either to target cell-induced activation of NK cells or to the process of cytotoxicity itself. PPT PowerPoint slide

PowerPoint slide PNG larger image

larger image TIFF original image Download: Fig 3. 7KC-induced NK cell death requires target cell–induced NK cell degranulation. (A) Survival of YTS cells after a 4-h incubation in the absence of target cells over a range of 7KC concentrations was measured based on their uptake of PI by flow cytometry to evaluate any direct toxicity of 7KC. (B) The same range of 7KC concentrations was also used as pretreatment of 51Cr-labeled YTS cells that were incubated with 721.221 target cells over a range of effector to target cell ratios for 4 h. The release of 51Cr was measured in both sets of experiments at 4 h. Cytotoxicity (C, p < 0.05) and survival (D, p < 0.05) of 7KC-pretreated YTS cells, with or without EDTA (10 mM) calcium chelation, were measured after a 4-h incubation with 51Cr-labeled 721.221 target cells. Cytotoxicity mediated by (E, p < 0.05) and survival (F, p < 0.05) of 7KC-pretreated YTS cells were measured after a 4-h incubation with nontriggering K562 cells or triggering KT86 cells. Viability (G) of untreated and 7KC-pretreated YTS cells with or without up-regulated LAMP-1 after a 4-h incubation with 721.221 cells was evaluated using flow cytometry with PI used to detect cell death. All data represent averages from 3 independent experiments, and error bars show ± SD (A−E). Wilcoxon signed rank test was used for the comparison between curves. NK, natural killer; PI, propidium iodide; 7KC, 7-ketocholesterol. https://doi.org/10.1371/journal.pbio.3001328.g003 To more specifically investigate the mechanism of 7KC-induced NK cell death, especially its linkage with the activation induced by target cells and ultimately degranulation, we measured the effect of 7KC treatment under calcium-depleted conditions. Calcium is critical in the signaling required for NK cell activation and is also necessary for lytic granule release as well as perforin activation and membrane binding [22]. Thus, we depleted calcium from medium using the chelator EDTA at concentrations previously defined as effective to inactivate perforin [30]. In the presence of EDTA, the cytotoxicity of YTS cells was completely abolished when evaluated by 51Cr release assays against 721.221 target cells, as has been previously reported for NK cells (Fig 3C). Interestingly, when YTS cells were labeled with 51Cr and combined with 721.221 target cells, EDTA also completely blocked the death of the 7KC-treated NK cells (Fig 3D). Thus, NK cell autolysis was prevented by chelation suggesting that NK cell activation, degranulation, or lytic molecule function is required for NK cell death in the presence of 7KC. This also largely excludes any direct NK cell toxicity of 7KC pretreatment as the autolysis of NK cells can be blocked by EDTA. To further evaluate the dependency of 7KC-induced NK cell death upon activation by target cells, we utilized both triggering and nontriggering target cell pairs along with YTS cells. Unlike 721.221 target cells, K562 cells lack the surface expression of activation ligands needed by YTS cells to promote degranulation. YTS cells do conjugate with and adhere to K562 cells resulting in the generation of some signals, but the signals required for degranulation and cytotoxicity are not present. Thus, YTS cells form noncytolytic conjugates with K562 cells (S5 Fig) [31]. The expression in K562 cells of a ligand sufficient for triggering the degranulation of YTS, CD86, makes the K562 target cell susceptible to YTS killing. A K562 cell line stably expressing CD86 (KT86) are easily killed by cytotoxicity by YTS cells [31] and were used in comparison to K562 cells to further evaluate the mechanism of 7KC-induced NK cell autolysis. Cytotoxicity assays with 51Cr-labeled K562 or KT86 cells demonstrated the ability of YTS cells to kill the latter but not the former, neither of which were affected by 7KC pretreatment of the NK cells (Fig 3E). When the YTS cells were 51Cr labeled, only those pretreated with 7KC were killed and only when incubated with the KT86 cells (Fig 3F). This suggests that adhesion to a target cell was insufficient to cause NK cell autolysis in the presence of disrupted lipid ordering and that triggering for cytotoxicity was required. This, in concert with calcium chelation, suggests that active cytotoxicity with degranulation is prerequisite for autolysis and that lipid ordering might be protective. Furthermore, since the only difference between these experiments was the presence of the cytotoxicity triggering ligand, it is again unlikely that 7KC is in anyway intrinsically toxic to NK cells. Since only a fraction of 7KC-pretreated NK cells underwent cell death when exposed to a susceptible target cell, we wanted to compare the dying NK cells to those that were still alive to try and gain insights into the mechanism underlying 7KC-induced NK cell death. Given that cytotoxicity and calcium were required for 7KC-induced NK cell autolysis, we speculated that NK cell degranulation would be associated. To assess this, we performed flow cytometry–based cytotoxicity assays in which 7KC-pretreated YTS cells were coincubated with 721.221 cells in the presence of medium containing propidium iodide (PI) as a viability indicator. After incubation, CD56+ cells were profiled via flow cytometry, and the percentage of cell surface LAMP-1, which denotes degranulation, was determined in the population of either viable or dying cells as marked by the absence or presence of PI, respectively. We found that when NK cells were pretreated with 7KC, those that expressed LAMP-1 were more frequently found to be dying (Fig 3G). In those that were not 7KC pretreated, LAMP-1+ NK cells were generally alive. As LAMP-1 is a marker of degranulation, and NK cells mediate cytotoxicity predominantly via directed secretion of perforin and proapoptotic proteases, it is likely that 7KC-induced NK cell death is dependent on the NK cell’s own killing machinery.

7KC-induced NK cell death is dependent on perforin The NK cell death resulting from 7KC pretreatment appears to be autolysis owing to its dependence upon both a cytotoxicity-triggering target cell and calcium as well as its correlation with LAMP-1 up-regulation. Therefore, we wanted to evaluate if it was indeed the NK cell lytic machinery that was responsible for autolysis after the blockade of presynaptic lipid ordering via 7KC. To evaluate this, we initially attempted to interfere with the function of the NK cell lytic granule by treating cells with Concanamycin A (CMA). CMA deacidifies the granules and depletes perforin by promoting its degradation [32]. Pretreatment of NK cells with CMA for 60 min depleted their perforin as demonstrated by the detection of protein via flow cytometry (S6 Fig). To determine if CMA treatment would have an impact on the 7KC-induced NK cell autolysis, we treated NK cells with CMA and measured cytotoxicity against 51Cr-labeled target cells, or NK cells that had also been treated with 7KC. As expected, CMA treatment of NK cells reduced target cell killing. There was also a similar reduction in 7KC-induced NK cell death after CMA treatment (Fig 4A). The incomplete reduction in cytotoxicity by CMA could be a feature of the residual, granule-independent perforin that would not be affected by granule deacidification [33]. CMA treatment, however, suggests a dependence of NK cell autolysis after 7KC treatment upon the lytic machinery itself. PPT PowerPoint slide

PowerPoint slide PNG larger image

larger image TIFF original image Download: Fig 4. 7KC-induced NK cell death is perforin dependent. Cytotoxic function mediated by (A, left) and survival of (A, right) untreated or CMA-treated NK92 cells, when incubated with K562 targets, were measured using 4-h 51Cr release assays. Cytotoxic function mediated by (B and C, left) and survival of (B and C, right) YTS cells, with or without knockdown of perforin induced by siRNA (B) or shRNA (C), were measured using 4-h 51Cr release assay. Sensitivity of untreated or 7KC-pretreated YTS cells (D) or 721.221 (E) to free perforin were compared based on the uptake of viability dye using flow cytometry. Data represent mean values from 3 independent experiments, and error bars display ± SD (A−E). CMA, Concanamycin A; NK, natural killer; shRNA, short hairpin RNA; siRNA, small interfering RNA; 7KC, 7-ketocholesterol. https://doi.org/10.1371/journal.pbio.3001328.g004 Since perforin degradation via the deacidification of lytic granules may be associated with additional consequences [32], we utilized alternative approaches to consider a dependence of 7KC-induced autolysis upon the lytic machinery. We employed 2 different RNA interference techniques to specifically deplete perforin from NK cells and evaluate the impact of reducing perforin expression upon autolysis. PRF1-targeting small interfering RNA (siRNA) in YTS cells affected substantial decreases in perforin protein as measured by western blot and fluorescence microscopy analyses (S7A–S7D Fig). The same was found using PRF1-short hairpin RNA (shRNA)-transduced YTS cells when measured by flow cytometry and fluorescence microscopy analyses (S7E–S7H Fig). As would be expected, both of these approaches resulted in a complete loss of cytotoxicity against 721.221 target cells (Fig 4B and 4C). When YTS cells were 51Cr labeled and pretreated with 7KC, however, the perforin siRNA-treated or shRNA-transduced cells survived at levels similar to the YTS cells that were not 7KC treated (Fig 4B and 4C). Irrespective, 7KC-induced autolysis was dependent upon perforin and thus likely represents a susceptibility to perforin released during cytotoxicity mediated through the lytic synapse. While perforin was necessary for 7KC-induced NK cell autolysis, we next asked if it was also sufficient on its own to kill NK cells having lipids that were disordered by 7KC treatment. To directly assess the impact of perforin upon the NK cell membrane and the effect of lipid ordering, we measured the sensitivity of NK cells to free perforin with or without 7KC pretreatment. YTS cells that were either 7KC pretreated or untreated were cultured in PI containing medium with various concentrations of perforin. Here, perforin-mediated transmembrane pore formation allows for the uptake of impermeable PI dye, and, thus, the sensitivity to perforin can be evaluated based on the percentage of PI-positive NK cells. When incubated with purified perforin, YTS cells demonstrated minimal PI uptake, which increased slightly with increasing perforin concentrations. When YTS cells were pretreated with 7KC, however, there was an increase in the perforin-induced PI uptake at all concentrations of perforin tested, which became increasingly noticeable at higher concentrations (Fig 4D). This difference was not observed in the same experiment performed with a susceptible target cell, 721.221 (Fig 4E), which demonstrated similar PI uptake at given concentrations of free perforin irrespective of 7KC treatment. Thus, 7KC treatment and presumably the resulting disordering of the otherwise increased ordered lipid in the NK cell membrane lead to a lack of protection against perforin.

Membrane binding of perforin is blocked by densely packed lipid membranes To understand how 7KC treatment might contribute to sensitizing NK cells to perforin attack, we wanted to directly evaluate if the 7KC-induced disruption of densely packed lipid membranes could underlie the inhibition of perforin binding. Thus, we utilized a synthetic cell-free system and measured the binding of perforin to model membranes having different lipid packing densities. Liposomes were generated using dioleoyl phosphatidylcholine (DOPC) and dipalmitoyl phosphatidylcholine (DPPC). Both lipids share identical head groups, and, thus, any potential head group specificity for perforin binding can be excluded. The difference between the 2 lipids, however, is in their tails. DOPC contains 2 unsaturated fatty acid sidechains that generate a more fluid, disordered lipid membrane, and the saturated sidechains of DPPC tend to form densely packed, high lipid order membranes (at 37°C). By adjusting the ratio of DOPC/DPPC, liposomes with a range of packing densities were prepared, which were validated using fluorescent packing sensors (S8A and S8B Fig). To determine the impact of membrane lipid composition on perforin binding ability, differently packed liposomes were incubated with perforin, and then membranes were precipitated by centrifugation. These liposome preparations were then evaluated by western blot analysis to determine the amount of membrane-bound perforin. Because perforin activation and membrane binding is dependent upon calcium, it was intentionally depleted in control groups to assess any nonspecific binding of perforin. When compared to liposomes constructed of 100% DPPC, which had the most packed and ordered membrane, liposomes with higher percentages of DOPC exhibited higher perforin binding (Fig 5A and 5B). The pure DPPC membrane retained the least amount of perforin, and a 1:1 mixture of the 2 lipids was intermediate. Thus, highly ordered membranes tend to be resistant to perforin binding. PPT PowerPoint slide

PowerPoint slide PNG larger image

larger image TIFF original image Download: Fig 5. Perforin membrane binding is prevented by the presence of densely packed lipid. Differently packed liposomes were created by either pure or a ratio of DOPC and DPPC, which were coincubated with perforin. Levels of membrane-bound perforin were evaluated by western blot analysis (A, representative of 3 independent repeats). Since membrane binding of perforin requires the presence of Ca++, this was either added at 20 mM concentration or depleted as specified in the blot and quantification. As a control, pure perforin (PF) was evaluated without liposomes. Three independent experiments were performed, and the quantitative values for each via densitometry were plotted with mean ± SD shown (B). Using Student t test, the mean values were different from DOPC Ca++ p < 0.0001, **** (C) Differently packed PC:SM:CHO liposomes were created with 1:1 PC:SM and increasing percentages of CHO (as specified above each lane) and coincubated with perforin in the presence of Ca++. Levels of membrane-bound perforin were evaluated by western blot analysis using anti-perforin antibody (clone D48) (representative result shown). Three independent experiments were performed, and the quantitative values for each via densitometry were plotted with mean ± SD shown (D). Using Student t test, the mean values were different from 10% CHO p < 0.0001, ****. CHO, cholesterol; DOPC, dioleoyl phosphatidylcholine; DPPC, dipalmitoyl phosphatidylcholine; PC, phosphatidylcholine; SM, sphingomyelin. https://doi.org/10.1371/journal.pbio.3001328.g005 Although the identical polar head group is a benefit of the DOPC/DPPC model membranes system, these are physically quite distinct from cell membranes [34]. In order to better mimic the plasma membrane of NK cells, liposomes were constructed that consisted of phosphatidylcholine (PC), sphingomyelin (SM), and cholesterol. Using this more physiological system, the amount of cholesterol could be varied to change the degree of lipid ordering. To validate this approach, liposomes were generated using a constant amount of PC and SM, but with cholesterol ranging from 10% to 50%. The degree of lipid ordering was measured using a fluorescent membrane sensor and the degree of lipid packing correlated with the percentage of added cholesterol (S8C and S8D Fig). Perforin was then added to these liposomes, membranes precipitated by centrifugation, and the amount of perforin retained, evaluated by western blot analysis. The cholesterol concentration inversely correlated with the degree of perforin retention by the liposomes (Fig 5C and 5D). In concert with our findings from the DOPC/DPPC model membrane, this suggested that the degree of membrane lipid ordering governed the protection from perforin insertion. Thus, the mechanism of NK cell protection from perforin is likely due to a blockade of perforin binding by densely packed presynaptic membranes.

Lytic granule membranes reinforce local membrane packing at sites of NK cell degranulation Increased lipid ordering at the presynaptic membrane was found to be critical for NK cell protection against secreted perforin during degranulation. That said, the concentration of perforin is likely to be highest where the release of perforin occurs. Thus, we wanted to evaluate the lytic granule itself in consideration of special membrane properties it might possess that could potentially contribute to the protection of NK cells from autolysis. In particular, we wanted to consider the possibility that the previously described fusion of the lytic granule–cell membrane with that of NK cell membrane that occurs during degranulation [35] may itself alter the composition and packing status of presynaptic membrane. To investigate potential influence of lytic granule–cell membrane fusion on presynaptic membrane properties, we first evaluated the lipid properties of the lytic granule membranes themselves. We initially used live cell confocal microscopy and NK cells incubated with a cell permeable packing sensor (Laurdan) to define lipid packing along with lysotracker to identify lytic granules. Organelles that fluoresced in the presence of lysotracker interestingly appeared to have densely packed membranes (Fig 6A). When quantified, the lytic granule membranes had higher lipid order than the NK cell presynaptic membrane as indicated by their comparative GP values (Fig 6B). This suggested that the lytic granules of NK cells have endogenous highly ordered lipid. Since these analyses were performed in resting YTS cells, it suggests that the lytic granules might be characterized by an inherently distinct lipid composition. In order to evaluate this, we isolated lytic granules from YTS cells and compared those to isolated total YTS cell membranes (which also include those of lytic granules). Both membrane preparations were then analyzed with liquid chromatography–mass spectrometry. The lipid profiles from the membranes indicated a dramatic enrichment of SM and its derivatives as well as other liquid ordered membrane components in the lytic granules (Fig 6C, S1 Table). SM and derivatives are known to facilitate aliphatic tail alignment and thereby enhance the packing density of lipid membrane [36]. Of note, monosialodihexosylganglioside (GM3), a high melting point lipid enriched in lipid-ordered membranes [37,38], was increased over 20-fold in the granules (S1 Table). Thus, the lytic granule appears to be defined by a highly specialized lipid membrane. PPT PowerPoint slide

PowerPoint slide PNG larger image

larger image TIFF original image Download: Fig 6. Lytic granule membranes are highly ordered and enhance local membrane packing at sites of NK cell degranulation. (A) Lipid packing densities of the cell membranes of nonstimulated YTS cell were measured with the fluorescent packing sensor Laurdan (pseudocolor scale, middle), and lytic granules were identified using lysotracker dye (red, left). The overlay of the 2 fluorescent channels was used to identify lytic granules (using an object identifier in ImageJ software–outlined in teal), which allowed for the measurement of lipid packing density at the location of the lytic granules (pseudocolor scale). An enlargement (right) shows the lipid packing of lytic granule membranes and cell membrane with an expanded pseudocolor scale using a more refined range to amplify lipid packing differences. (B) Lipid packing density was quantified in the region of the lytic granules as determined by lysotracker colocalization as the GP value of this colocalized region and was compared to the GP value of the entire cell membrane using a paired t test (n = 15 cells from 3 independent experiments, p < 0.001). (C) Lipid composition of whole YTS cell membranes and isolated YTS lytic granule membranes were determined by LC–MS and were plotted in pie charts. The results represent 2 independent experiments, and individual lipids are shown in different colors with the following major lipids depicted FC, PE, CE, SM, dhSM, MG, PEp, PC, PI, PCe, PS, and Cer. (D) YTS-LAMP1-pHluorin cells were used to identify the sites of degranulation at the NK cell membrane on an activating glass surface (coated with anti-CD18 and anti-CD28 antibodies) via live TIRF microscopy. Images from a representative time lapse were selected to denote the presence of pHluorin (green, top), which signifies the site of degranulation. For reference, the entirety of the cell on the glass is depicted in gray. The lipid packing signal using a pseudocolor scale is shown in the image below. (E) The emergence of pHluorin signal was monitored using TIRFm in cells that had been preloaded with the Laurdan lipid packing sensor, and local membrane packing at degranulation sites was quantified and compared with randomly selected non-degranulation sites. The mean values for 10 independently measured cells are shown ± SD. (F) Cytotoxicity mediated by and survival of 7KC-pretreated YTS cells as measured by Cr release assays in the presence of LAMP-1 blocking antibody. (G) Cytotoxicity mediated by and survival of 7KC-pretreated YTS cells in which LAMP-1 expression was reduced by prior siRNA introduction. Cytotoxicity data represent averages of 3 replicated wells (i.e., technical repeats), and error bars display ± SD of these technical replicates. (F, G). CE, cholesteryl ester; Cer, ceramide; dhSM, dihyodrosphingomyelin; FC, free cholesterol; GP, generalized polarization; LC–MS, liquid chromatography–mass spectrometry; MG, monoglycerol; NK, natural killer; PC, phosphatidylcholine; PCe, alkylacyl phosphatidylcholine; PE, phosphatidylethanolamine; PEp, phosphatidylethanolamine-based plasmalogen; PI, phosphatidylinositol; PS, phosphatidylserine; siRNA, small interfering RNA; SM, sphingomyelin; TIRF, total internal reflection fluorescent; 7KC, 7-ketocholesterol. https://doi.org/10.1371/journal.pbio.3001328.g006 Since the lytic granules had highly ordered membrane with SM-enriched composition, it was possible that lytic granule–cell membrane could reinforce the packing density of plasma membrane at sites of degranulation via membrane fusion. To determine if there were degranulation-induced changes in local synaptic membrane packing, we utilized NK cells expressing LAMP1–pHluorin as a degranulation indicator to define the precise site of lytic granule fusion [39] and prelabeled the cells with a lipid packing sensor. These NK cells were incubated on an activating glass surface containing antibodies against NK cell activation receptors that promote degranulation and imaged by TIRF microscopy. Local lipid packing changes were recorded at specific sites of lytic granule fusion as detected by the emergence of pHluorin fluorescence. At the time pHluorin signal was detected, a rapid rise of lipid order was also found at the degranulation site (Fig 6D). Interestingly, this increase of lipid ordering was sustained and spatially restricted to the degranulation site without a sign of diffusion over a 5-min time frame (Fig 6E). This may be explained by the low lateral diffusion rate of lipid molecules in a densely packed presynaptic membrane [40] and is consistent with the previous definition of prolonged granule transmembrane molecule patches in the NK cell synaptic membrane [39]. Given the protection provided by ordered membrane, this further increased ordering upon degranulation may provide enhanced defense at degranulation sites where NK cells face maximum local concentrations of secreted perforin. Thus, lytic granules themselves have an endogenously densely packed membrane, and after degranulation contribute to focal reinforcement of presynaptic membrane lipid packing through lytic granule–cell membrane fusion to potentially provide additional protection where it is likely needed most. Given the enhanced protection against autolysis provided by degranulation itself, we wanted to evaluate the possibility that protective receptors externalized by lytic granules could be contributing. LAMP-1 comprises 50% of all lysosomal membrane proteins [41], is externalized upon degranulation, and has been previously proposed to provide protection against perforin attack under certain circumstances [16]. To consider an independent and protective role of LAMP-1 externalization, YTS cells were untreated or 7KC pretreated and coincubated with 721.221 cells in the presence of medium containing LAMP-1-neutralizing antibodies. Blockade of LAMP-1 did not interfere with NK cell killing of 51Cr-labeled 721.221 target cells, nor did it cause autolysis of 51Cr-labeled YTS cells in the presence of 721.221 target cells (Fig 6F). Pretreatment with 7KC did not change these results, suggesting that lipid order was the primary influence upon degranulation-induced autolysis. Since LAMP-1-bound antibodies could in theory dampen perforin binding by steric hindrance effects, we utilized siRNA as an alternative approach to substantially reduce LAMP-1 expression in NK cells globally and inside the lytic granules specifically (S9 Fig). Despite the reduction in LAMP-1 expression by siRNA treatment, there were no changes in NK cell survival by 51Cr release assay using labeled NK cells (Fig 6G). Furthermore, there was no amplification of autolysis in 7KC-treated YTS cells that had also been subject to LAMP-1 siRNA treatment. Thus, while LAMP-1 is abundant in lytic granules and is externalized in the region of a degranulation, the lipid ordering of the presynaptic membrane appears to be a more impactful protective mechanism utilized by NK cells against secreted perforin.

[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3001328

(C) Plos One. "Accelerating the publication of peer-reviewed science."
Licensed under Creative Commons Attribution (CC BY 4.0)
URL: https://creativecommons.org/licenses/by/4.0/


via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.