Cell Lines and Ex Vivo NK Cells
NK-92 and GFP-actin expressing NK-92 cell lines were a kind gift from K. Campbell and were maintained in Myelocult (StemCell) media supplemented with 100 U/mL penicillin and streptomycin (Gibco) and 100 U/mL IL-2 (Hoffman-La Roche). mCherry-actin, Citrine-actin, and pHluorin-LAMP1 expressing cells were generated by retroviral transduction of NK-92 cells as described 
. Briefly, 2–4 µg of plasmid DNA were transfected into the Phoenix packaging line using Fugene (Roche) lipofection reagent. Supernatant was harvested on day 2 post-transfection. NK-92 cells, Polybrene (Sigma), and supernatant were mixed and spun in a well of a 6 well plate at 1,000×g for 90 min at 32°C. Following overnight incubation at 32°C, cells were spun down and resuspended in supplemented Myeolocult. Cells were grown for 3 d prior to the introduction of puromycin (2 µg/mL) (InvivoGen) or hygromycin B (150 µg/mL) (Cellgro). mCherry-actin and Citrine-actin expressing cells were sorted for high expression by the University of Pennsylvania Cell Sorting Facility. Ex vivo NK cells were prepared from concentrated whole blood as described 
Generation of Plasmids
The pHluorin-LAMP1 retroviral plasmid was generated by BioMeans, Inc. by inserting the sequence for pHluorin (a kind gift from G. Miesenböck) between the signal sequence and the transmembrane domain of IL-2Rα linked to the cytoplasmic tail of LAMP1 (a kind gift from M. Marks). A flexible GS linker was added between pHluorin and the transmembrane domain sequences. The entire construct was subsequently cloned into the MIGR1-puromycin vector.
The mCherry-actin retroviral plasmid was generated by PCR amplifying mCherry-actin from a pmCherry plasmid with 5′ BglII and 3′ EcoRI restriction site overhangs. The PCR product was digested and ligated into the pMSCV-Hygromycin plasmid (a kind gift from W. Pear), which had an EcoRI site in the Hygromycin resistance gene sequence eliminated by site-directed mutagenesis. The Citrine-actin retroviral plasmid was generated by amplifying the Citrine sequence from the pRSET-b Citrine plasmid (a kind gift from R. Tsien) with 5′ and 3′ BglII overhangs. The product was digested and ligated into a similarly digested mCherry-actin retroviral plasmid, effectively removing the mCherry sequence and inserting the Citrine sequence. Proper orientation of insert was verified by DNA sequencing by the Children's Hospital of Philadelphia Research Institute sequencing core facility.
Flow cytometry was performed to verify pHluorin-LAMP1 expression. Cells were untreated or treated with phorbol myristate acetate (PMA, 100 ng/mL, Sigma) and Ionomycin (1 µg/mL, Sigma) for 30 min or Concanamycin A (CMA, 100 nM, Sigma) for 90 min and samples were run on a BD FACSCalibur.
Total Internal Reflection Fluorescence Microscopy
Cells were washed and resuspended in supplemented Myelocult prior to use. For imaging of lytic granules, cells were incubated with 100 nM LysoTracker Red DND-99 (Molecular Probes) for 30 min at 37°C, washed once, and resuspended in supplemented Myelocult. ΔT dishes (Bioptechs) were coated with 5 µg/mL anti-NKp30 (Beckman-Coulter) and 5 µg/mL anti-CD18 (Clone IB4) for 1 h at 37°C, washed with PBS, and prewarmed prior to imaging with 1 mL dye free R10 (dye free RPMI 1640 (Gibco), 10% fetal bovine serum (Atlanta Biologicals), 10 mM HEPES (Gibco), 100 U/mL penicillin and streptomycin, 100 µM MEM nonessential amino acids (Gibco), 1 mM sodium pyruvate (CellGro), and 2 mM L-glutamine (Gibco). 4×105 cells were added to the dishes, which were maintained at 37°C with a heated stage and lid (Bioptechs).
For live cell imaging of actin dynamics following inhibitor treatment, cells were activated as above for 10 min before addition of media containing DMSO or jasplakinolide (1 µM, Calbiochem). Following 5 min of incubation, media containing DMSO or latrunculin A (10 µM, Sigma) was added to the dish. After 5 min of further incubation, cells were imaged.
For fixed cell experiments, 1×105
cells were adhered to No. 1 glass coverslips coated with antibody as described above. Samples were fixed and stained with Alexa Fluor 488 phalloidin or 568 phalloidin (Molecular Probes) as described 
. For experiments with inhibitor treatments, cells were activated on coverslips for 10 min or 20 min, treated with inhibitors for 5 min, and then fixed and stained. Cytochalasin D (Sigma) was used at 10 µM.
Samples were imaged through a 1.49 NA, oil immersion, 60×, APO N TIRFm objective or a 1.45 NA, oil immersion, 100×, PlanApo TIRFm objective (Olympus) when noted. 488 nm (Spectra-Physics) and 561 nm (Cobalt) diode lasers were launched through a two-line combiner of an LMM5 (Spectral Applied Research) into a rear mounted TIRF illuminator (Olympus) on an Olympus IX-81. Lasers were aligned for total internal reflection prior to each experiment. Images were captured using Volocity (PerkinElmer) to control a C9100 EM-CCD camera (Hamamatsu).
Mel1190 cells were plated into ΔT dishes 1 d prior to use. Cells were stained with CellMask Deep Red (Invitrogen) according to manufacturer's instructions just prior to imaging. GFP-actin expressing NK-92 cells were added to the dishes, which were maintained at 37°C, and imaged for up to 1 h. Cells were imaged through a 63×1.4 NA Plan-APOCHROMAT objective (Zeiss) on a Zeiss Observer.Z1 using a C10600 ORCA-R2 camera (Hamamatsu). The microscope was equipped with a CSU10 spinning disk system (Yokogawa). 491 nm (Cobalt) and 655 nm (CrystaLaser) diode lasers were launched through an LMM5 (Spectral Applied Research).
NK-92 cells were activated on antibody coated glass coverslips as described above for 30 min and then fixed and stained with rabbit anti-human Pericentrin (Abcam), Alexa Fluor 488-phalloidin, and anti-Perforin-Alexa Fluor 647 (Biolegend). The secondary antibody to anti-Pericentrin was a goat anti-mouse Pacific Blue (Molecular Probes). Cells were imaged in three dimensions on a spinning disk confocal Olympus DSU IX-81 microscope.
Stimulated-Emission Depletion (STED) Microscopy
Mel1190 cells were grown in a monolayer overnight on No. 1.5 glass coverslips. Citrine-actin expressing NK-92 cells (106) were resuspended in media and incubated on Mel1190 targets for 30′ at 37°C. Cells were fixed with 2% paraformaldehyde and mounted with ProLong antifade reagent (Invitrogen). Cells were imaged at the plane of the interface between NK and target cells. Separately, for imaging of granules, Citrine-actin expressing NK-92 cells were immobilized to bound antibody as described above, then fixed, permeabilized, and stained with anti-perforin Alexa Fluor 488 (Biolegend).
For visualization of actin and perforin in NK-92 cells, cells were immobilized to bound antibody as described above, then fixed, permeabilized, stained with Alexa Fluor 488 phalloidin and anti-perforin Alexa Fluor 647, and imaged at the plane of the glass or 0.5–1 µm above it.
All samples were mounted with Prolong anti-fade reagent (Invitrogen). Cells were imaged through a 100×1.4 NA HCX APO objective on a Leica TCS STED CW system controlled by Leica AS AF software. Alexa Fluor 488 and Citrine were excited using a 488 nm Argon laser and STED depletion was achieved using a 592 nm continuous wave fiber laser. Alexa Flour 647 was excited using a HeNe 633 laser and imaged using the laser scanning confocal modality of the system. Fluorescence was detected with HyD detectors (Leica).
Platinum Replica Electron Microscopy
Cells were washed, resuspended in supplemented Myelocult, and allowed to adhere to 0.15 glass coverslips coated with antibody as described above. For imaging of the actin network alone, samples were prepared following a modified protocol described in 
. After a period of incubation at 37°C, samples were washed once in PBS, dipped into a 1
3 dilution of PEM buffer (0.1 M PIPES (Sigma), pH 6.9, 1 mM EGTA (Sigma), 1 mM MgCl2
(Sigma)) in dH2
0 for 10 s, sonicated in PEM for 1–2 s at a 45° angle to the probe, and incubated with 1% Triton X-100 (Sigma) in PEM for 1–2 min before fixation in 2% glutaraldehyde (Fluka) in PEM. For imaging of granules, cells were “unroofed” by applying a nitrocellulose membrane that had been wet in PEM to the coverslip for 45–60 s, removing it, and then fixing the sample in 2% glutaraldehyde in PEM. All samples were processed for EM as described 
. Platinum replicas were imaged on a JEM 1011 transmission electron microscopy (JEOL USA) at 100 kV. Images were captured using an ORIUS 835.10W CCD camera (Gatan) and are presented in inverted contrast.
Immulon 4HBX 96 well flat bottom plates (Thermo) were coated with murine IgG (BD), anti-human NKp30, anti-human CD18, or both anti-NKp30 and anti-CD18 at 5 µg/mL in PBS overnight at 4°C. Plates were washed twice with PBS and blocked for 1 h at room temperature with R10 media. 1×105 cells were added to wells and plates were spun at 1,000 rpm for 2 min before incubation at 37°C. For the timecourse degranulation assay, supernatants from spontaneous and activated wells were harvested at indicated times. For inhibitor treatments, media containing inhibitor or vehicle (DMSO) was added at indicated times. Thapsigargin (Calbiochem) was used at 1 µM. Supernatants were harvested after 60 min. For total release, cells were lysed in 0.5% Nonidet P40 (Accurate Chemical and Scientific).
Supernatants were assayed by mixing 20 µL supernatant with 200 µL of a solution containing PBS, 9.8 mM HEPES (Gibco), 196 µM Z-L-Lys-SBzl hydrochloride (BLT, Sigma), and 218 µM 5,5′-dithiobis(2-nitrobenzoic acid) (DTNB, Sigma). Samples were incubated for 30 min at 37°C and absorbance was measured immediately at 405 nm. Percent total release was measured by subtracting the spontaneous release value from activated release values (A–S) and the total release value (T–S), and then dividing (A–S) by (T–S).
Images and timelapse series were analyzed using either Volocity or the FIJI package of ImageJ (http://pacific.mpi-cbg.de
). Using Volocity, fluorescently tagged actin footprints were identified using a classifier that identifies objects above a selected standard deviation above the mean intensity (usually 0–1). LysoTracker and pHluorin positive events were similarly identified (4–10 standard deviations above mean intensity), with the additional exclusion of events smaller than 0.05 µm2
. For actin fluorescence ratio measurements, the MFI at the location of granule approximation or degranulation was divided by the MFI of the actin footprint. Maximum and minimum potential actin intensities were also measured by dividing the maximum and minimum pixel intensities of the actin footprint by the MFI of the footprint. For the “adjacent points” measurement, the region of interest (ROI) that was identified for the degranulation (pHluorin) signal was moved to four neighboring locations. The actin intensities were measured, averaged, and then divided by the MFI of the actin footprint to generate a single value.
Distance of granules from the MTOC was determined as described 
by inputting the coordinates of the MTOC centroid and a granule centroid into the Pythagorean equation (a2
, where c is the distance from the MTOC).
FIJI was used to generate radial intensity profiles using the radial profile plugin (http://rsbweb.nih.gov/ij/plugins/radial-profile.html
). For a profile of the entire cell footprint an ROI circle was drawn around the cell and data from running the plugin were exported to Excel (Microsoft). For profiles of degranulation events images were first split into red and green images and then a circle with a radius of 8 pixels (1.08 µm) was drawn around each event. Data were generated for 6 radii from the center of each event. To determine the radial intensity change, an outer radial intensity value was subtracted from an inner radial intensity value. Thus a negative value indicates that the outer circle has higher mean intensity than the inner circle.
To measure changes in the actin network over time, sequential images were imported into FIJI, a line was drawn across the center of the cell, and line intensity profile data were generated for each time point at the same location within the cell. Data were exported to Excel and standard deviations calculated. Surface plots were generated using the surface plot function in FIJI.
Images of granules taken using STED microscopy were analyzed in Volocity and diameters measured by drawing lines across the center of the granules. STED microscopy images of the actin network were imported into FIJI and processed before analysis. Background was subtracted using the Rolling Ball Subtraction algorithm with the radius set to 150 pixels and then pixel intensities were squared twice. An ROI was drawn around the interior of the cell and clearances were identified using the default autothreshold with “dark background” unchecked. Clearance areas were sorted, grouped, and counted in Excel based on size. Dividing the number of clearances per cell by the area measured normalized the values. Area cutoffs were implemented by using granule diameter as a reference. The smallest two sizes calculated assume that granules are uniformly spherical and require a clearance that has an area that would accommodate the equatorial area of the granule. The larger size categorizes all clearances that are larger than most granules.
Colocalization analysis was performed in Volocity using the “Find objects” tool to identify granules and a fixed intensity threshold, which was adjusted to generate interfilament or intercellular black space on a per field basis, to identify the actin network. The colocalized area of granule and actin network staining was divided by the area of the granule to yield a value denoting the percent of the granule area colocalized with actin. Five granules from 10 cells over 2 experiments were measured (N
50 granules). Line profiles of granules and the actin network were performed in FIJI after channels were separated. All STED images were measured in their raw form, but shown after processing in Leica AS AF with the median noise reduction feature.
Images from platinum replica electron microscopy were inverted and linearly contrast enhanced using Photoshop (Adobe) and imported into FIJI for processing and analysis. Background was subtracted from each image using the Rolling Ball Subtraction algorithm with the radius set to 25 pixels. Pixel intensities were subsequently squared. Cells were identified using the default autothreshold and cell contact area and cell centroid were measured with “include holes” checked. ROIs were drawn around the interior of the cell to more accurately identify filaments and to avoid debris. Filaments were identified using the default automatic threshold with “dark background” checked. Clearances in the filamentous network were identified as mentioned above. Distance from the cell centroid was determined by inputting the coordinates of the cell centroid and a clearance centroid into the Pythagorean equation as described above for granule to MTOC distance.
All data were plotted using Prism (Graphpad).
Statistical significance was determined using Prism to perform one- or two-sample, unpaired or paired, two-tailed Student's t tests. Unless otherwise indicated all tests were two-sample, two-tailed, and unpaired. Where noted, n.s. denotes differences that have p values >0.05 and therefore considered not significant.