|Home | About | Journals | Submit | Contact Us | Français|
Src homology region 2-containing protein tyrosine phosphatase-2 (SHP-2)4 is required for full activation of Ras/ERK in many cytokine and growth factor receptor signaling pathways. In contrast, SHP-2 inhibits activation of human natural killer (NK) cells upon recruitment to killer cell Ig-like receptors (KIR)4. To determine how SHP-2 impacts NK cell activation in KIR-dependent or KIR-independent signaling pathways, we employed knockdown and overexpression strategies in NK-like cell lines and analyzed the consequences on functional responses. In response to stimulation with susceptible target cells, SHP-2-silenced NK cells had elevated cytolytic activity and IFN-γ production, whereas cells overexpressing wild type or gain-of-function mutants of SHP-2 exhibited dampened activities. Increased levels of SHP-2 expression over this range significantly suppressed microtubule organizing center (MTOC)4 polarization and granzyme B release in response to target cells. Interestingly, NK-target cell conjugation was only reduced by overexpressing SHP-2, but not potentiated in SHP-2-silenced cells, indicating that conjugation is not influenced by physiological levels of SHP-2 expression. KIR-dependent inhibition of cytotoxicity was unaffected by significant reductions in SHP-2 levels, presumably because KIR were still capable of recruiting the phosphatase under these limiting conditions. In contrast, the general suppressive effect of SHP-2 on cytotoxicity and cytokine release was much more sensitive to changes in cellular SHP-2 levels. In summary, our studies have identified a new, KIR-independent role for SHP-2 in dampening NK cell activation in response to tumor target cells in a concentration-dependent manner. This suppression of activation impacts MTOC-based cytoskeletal rearrangement and granule release.
In humans, natural killer (NK) cells comprise 10–15% of peripheral blood lymphocytes, and they serve as critical sentinels protecting against tumor and virus-infected cells (1, 2). NK cells are controlled by a fine balance between signals generated from adhesion receptors (e.g. integrins), activating receptors, and inhibitory receptors (3–5). Upon initial contact with a sensitive target cell, integrins promote enhanced intercellular conjugation, thereby stabilizing the cell-cell interaction (6, 7). Subsequently, both the actin and the microtubule-based cytoskeleton polarize toward the NK-target cell interface, a region referred to as the NK immune synapse (NKIS)4 (8–11). Consequently, activating receptors (e.g. NKG2D, NKp44, CD16) aggregate at the NKIS, and Src family protein tyrosine kinases (PTKs) phosphorylate the intracellular domains associated with these aggregated receptors to recruit Syk family kinases (Syk and ZAP-70) and adaptor proteins (e.g. SLP-76) (3, 12, 13). Polarization of the microtubule organizing center (MTOC) toward the NKIS facilitates the trafficking of cytolytic granules to the cell membrane and their subsequent release toward the target cell (14, 15). These cytolytic granules contain proteins that rupture the target cell membrane (e.g. perforin) and activate caspase-dependent apoptosis (e.g. granzymes) (16, 17). NK cell activation also leads to the production of cytokines (especially IFN-γ), which are important in both tumor/viral clearance and lymphocyte recruitment in vivo (18).
NK cell activation is dominantly suppressed if the NK cell inhibitory receptors engage with major histocompatibility complex class I (MHC-I) molecules on normal target cells at the NKIS (19, 20). The main inhibitory receptor family expressed by human NK cells is the killer cell Ig-like receptors (KIRs), which mediate the suppression of NK cell activation through ITIMs [(I/V)xYxx(L/V)] in the cytoplasmic domain (12, 21, 22). When inhibitory KIRs engage with MHC-I at the inhibitory NKIS, the ITIMs are phosphorylated by Src family PTKs (23, 24), which creates docking sites for the protein tyrosine phosphatases, SHP-1 and SHP-2 (25–27). SHP-1 and SHP-2 exhibit distinct requirements for binding to the KIR ITIMs. SHP-1 recruitment requires the phosphorylation of both the N- and C-terminal ITIM motifs of KIR, while SHP-2 can bind to KIR with only the N-terminal ITIM phosphorylated and can even bind weakly to the same ITIM in the unphosphorylated state (28–31). Substantial evidence indicates that the recruitment of SHP-1/2 is necessary for KIR function, since elimination of both ITIM motifs or expression of dominant negative SHP-1 or SHP-2 abolishes all inhibitory function (27, 32, 33). SHP-1/2 recruitment to the NKIS blocks many of the key steps leading to cytolysis, such as: a) the phosphorylation of activating receptors, b) the recruitment of Src and Syk kinases to the NKIS, c) NK-target cell conjugation, d) the accumulation of the cytoskeleton at the NKIS and e) the release of cytolytic vesicles (22). Although the direct substrates of SHP-2 in KIR signaling are not yet defined, available data suggest that SLP-76 and Vav1 are direct substrates of SHP-1 (34, 35)
Depending upon the context of cell type and signaling pathway, SHP-2 can act as an activator or inhibitor in various signaling pathways (36). As previously mentioned, SHP-2 can inhibit cellular activation through recruitment to a number of inhibitory receptors (e.g. KIRs, CD31, CTLA4), where the phosphatase is thought to dephosphorylate key players of cellular activation (37, 38). In sharp contrast, SHP-2 is also well known to function as an activator of the Ras/ERK signaling cascades downstream of many receptor tyrosine kinases (e.g. EGFR, PDGFR) and cytokine receptors (e.g. IL-2) (39–44). In this context, SHP-2 may mediate activation of this pathway either by inhibiting the Src kinase inhibitor Csk, allowing for Src-dependent activation of Ras/ERK (45), or by inhibiting RasGAP (46, 47), which catalyzes the transition from GTP-bound, active Ras to GDP-bound, inactive Ras. SHP-2-mediated activation of Ras/ERK may also involve inhibition of Sprouty proteins, a small family of molecules involved in the negative regulation of Ras (48).
Tight regulation of SHP-2 function is important for human health, since too much or too little SHP-2 can be detrimental to cellular development and function. Severe gain-of-function SHP-2 mutations are associated with cancer (e.g. juvenile myelomonocytic leukemia (JMML), acute myelogenous leukemia) (49–51). Less severe gain-of-function mutations cause Noonan syndrome, a fairly common autosomal dominant disorder typified by an irregular face, short stature, cardiac abnormalities and an increased cancer risk (52, 53). Many of the abnormalities associated with Noonan syndrome can be linked to the inappropriate over-proliferation of cells during development. Noonan syndrome mutations are found throughout the protein sequence, although most map to the N-terminal SH2 domain (e.g. Y63C, E76D, Q79P) or phosphatase domain (e.g. I282V, N308D) (52, 54–56). The vast majority of these gain-of-function mutations disrupt the interaction between the N-terminal SH2 domain and the phosphatase domain, which constitutively suppresses catalytic activity of wildtype SHP-2 (36).
Our previous work utilized dominant negative SHP-2 to demonstrate the role of SHP-2 in KIR-dependent inhibition of NK cell function (27, 31). In the current report, we extended our studies of SHP-2 in NK cells by performing shRNA knockdown and overexpression of SHP-2 to address the role(s) of the phosphatase in KIR-dependent and KIR–independent processes. Our findings demonstrate that SHP-2 is an inhibitor of both cytolytic activity and IFN-γ secretion by NK cells and that this function is independent of the role of SHP-2 in KIR signaling.
All cell culture was preformed at 37°C in a 7% CO2 humidified atmosphere. The IL-2-dependent NK-like cell lines KHYG-1 (kindly provided by Dr. Masato Yagita, Kijano Hospital, Osaka, Japan through the Japan Health Science Research Resources Bank, #JCRB0156) and NKL (a gift from Dr. Marco Colonna, Washington University, St. Louis, MO) were maintained in α-MEM medium (Life Technologies, Rockville, MD) containing 10% heat inactivated FBS (HyClone, Logan, UT), 10% heat inactivated horse serum (Invitrogen, Carlsbad, CA), 2 mM L-glutamate (Mediatech, Herndon, VA), 100 I.U./ml penicillin (Mediatech), 100 µg/ml streptomycin (Mediatech), 1 mM sodium pyruvate (Sigma-Aldrich, St. Louis, MO), 200 µM myoinositol (Sigma-Aldrich), 125 µM folic acid (Sigma-Aldrich), 1X non-essential amino acids (Mediatech), 100 µM 2-ME (Fisher Scientific, Pittsburgh, PA) and supplemented with 2% culture supernatant of J558L cells transfected with the human IL-2 gene (a gift from Dr. Antonio Lanzavecchia, Institute for Research in Biomedicine, Bellinzona, Switzerland). Cells were passed with fresh IL-2 and medium every 3–4 days. The MHC-I deficient EBV-transformed B cell line, 721.221, or those transduced to express HLA-B*5101 were cultured in RPMI-1640 medium (Mediatech) containing 10% heat inactivated FBS, 2 mM L-glutamate, 100 I.U./ml penicillin, 100 µg/ml streptomycin, 50 mM Hepes (Fisher Scientific) and 50 µM 2-ME. The 721.221 cells were passed into fresh medium every 3–4 days. Since a subset of KHYG-1 cells endogenously express KIR3DL1 (57), the parental KHYG-1 cells were sorted for lack of this receptor using the DX9 mAb (BD PharMingen). These sorted cells stably lacked KIR3DL1 for at least 1 month in culture. KHYG-1 cells lacking the endogenous KIR3DL1 were used for all experiments.
Four SHP-2 shRNAs were designed using Oligoengine software (www.oligoengine.com): shRNA #1 (targeting gattcagaacactggtgat, in N-terminal SH2 domain; a gift from Drs. Benjamin Neel, Ontario Cancer Institute, Ontario, Canada and Frank David, previously of Beth Israel Deaconess Medical Center, Boston, MA), #2 (gaatcctatggtggaaaca, in C-terminal SH2 domain), #3 (caggaactgaaatacgacg, in C-terminal SH2 domain) and #4 (gctgagaccacagataaag, just preceding the phosphatase domain). To generate double stranded shRNA, 3 µg sense and 3 µg antisense strands incorporating an intervening hairpin and terminal Bgl II and Hind III restriction site overhangs were annealed together and ligated into pSuperior.retro.neo or .puro vectors (Oligoengine, Seattle, WA) according to the manufacturer’s instructions. SHP-2-WT cDNA (a gift from Dr. Benjamin Neel through Addgene; plasmid #8329) was mutated to the Noonan Syndrome missense mutations, E76D and N308D (54), using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) and cloned into pBMN-IRES-EGFP vector (a gift from Dr. Garry Nolan, Stanford University, Stanford, CA). KIR3DL1-WT and –YF have been described previously (27). All manipulations and duplication of pSuperior and pBMN plasmids were done at 30°C in Stbl2 bacteria (Invitrogen, Carlsbad, CA) to prevent recombination at retroviral LTRs.
The retroviral transduction protocol was previously described (58, 59). Briefly, 4 µg of each pSuperior or pBMN-IRES-EGFP construct was transfected with Lipofectamine and Plus Reagent (Invitrogen) into the retroviral packaging cell line, Phoenix-ampho (also a gift from Dr. Garry Nolan). After two days, viral supernatant was harvested and used to transduce NK cell lines with Lipofectamine and Plus Reagent. Cells transduced with SHP-2 shRNAs were selected with 1.25 mg/ml G418 and/or 2.5 µg/ml puromycin for 5 days (starting 2 days after viral infection). SHP-2 protein levels decreased over time after selection, becoming stably knocked down 7 days after antibiotic selection was completed. Cells were used for a maximum of 40 days after drug selection. Data from multiple transduced NK cell populations are shown for all shRNA experiments and some overexpression experiments.
KHYG-1 cells were mock-treated or stimulated with pervanadate for 10 min, and then lysed for 30 min in lysis buffer containing 20 mM Tris (pH 7.4) and 1% Triton X-100 (Pierce, Rockford, IL) as described (60). To generate whole cell lysates, 5×105 NK cells were washed once with HBSS, boiled in 3× SDS sample buffer for 5 min and probe-sonicated for 10 sec. Samples were separated in 10% SDS-PAGE gels under reducing conditions and transferred onto PVDF membranes (Millipore, Billerica, MA). After transfer, membranes were air-dried, reactivated with methanol, washed once with dH2O and blocked with 5% nonfat milk (ACME, Salt Lake City, UT) in PBS. Blocked membranes were incubated initially with mouse anti-GAPDH mAb (Chemicon, Temecula, CA), mouse anti-SHP-2 mAb and rabbit anti-SHP-1 pAb (Santa Cruz Biotechnology, Santa Cruz, CA), washed in PBS 0.1% Tween-20, followed by probing with anti-mouse-IR dye-680 and anti-rabbit-IR dye-800 Abs (LI-COR Biosciences, Lincoln, NE). Blots were washed with PBS 0.1% Tween-20, and then with PBS alone. Proteins were visualized with an Odyssey Infrared Imaging scanner (LI-COR Biosciences) by scanning at 700 and 800 nm. Fluorescence of GAPDH, SHP-1 and SHP-2 bands were quantified using Odyssey software.
KHYG-1 or NKL cells (harvested on day 1 or 2 after IL-2 stimulation, respectively) were tested for direct cytolytic activity against 721.221 cells in a 51Cr-release assay in 200 µl medium/well (complete α-MEM lacking IL-2). 721.221 targets were labeled with 100 µCi 51Cr (Perkin Elmer, Waltham, MA) in 200 µl 100% heat inactivated FBS for 60–90 min at 37°C, washed 3 times with RPMI-1640 and resuspended in complete α-MEM lacking IL-2. Labeled targets and NK cells were mixed in V-bottom 96-well plates (Costar, Cambridge, MA; 1×104 targets/well), pelleted at 1300–1500 rpm for 3 min, and incubated for 2–5 hr at 37°C and 7% CO2 atmosphere. Plates were then centrifuged and 100 µl of culture medium was removed from each well and γ-counted. Spontaneous release and total release of 51Cr were determined by incubating target cells in medium alone or containing 1% Triton X-100, respectively. Each assay condition was always performed at least in triplicate. The percentage of specific target cell lysis was determined as: [(average cpm experimental release − average cpm spontaneous release)/(average cpm total release − average cpm spontaneous release)] × 100.
NK cells (2.5–5×105 cells/well) were stimulated for 24 hrs in a flat-bottom 96-well plate with an equal number of 721.221 cells, plate-bound antibodies, or medium alone. All antibodies were pre-absorbed to plates at 25°C for 3 hrs or overnight at 4°C in 100 µl of antibody coating buffer (e-Biosciences, San Diego, CA)/well at the following concentrations: 0.5–1 µg/ml anti-NKp44 (3.43.13, a gift from Dr. M. Colonna), 0.5–1 µg/ml anti-NKG2D (R&D Systems, Minneapolis, MN) or 1 µg/ml anti-CD11a (a subunit of LFA-1; Biolegend, San Diego, CA). Following stimulation with targets, medium or antibodies, culture supernatants were harvested, and then tested for the presence of IFN-γ by ELISA according to the manufacturer’s instructions (e-Biosciences).
Up to 3×106 NK and target cells were washed once with GIBCO OPTI-MEM serum free medium and resuspended in the same medium containing 4 µM Cell Tracker Blue (Invitrogen) or 5 µM Cell Tracker Orange (Invitrogen), respectively. Cells were stained for at least 10 min at 37°C in the dark, pelleted and incubated in fresh OPTI-MEM for at least 15 min more. Following staining, cells were resuspended in cold α-MEM and kept on ice. 1–5×104 NK and target cells were combined in pre-chilled FACS tubes on ice at a 1:1 ratio, pelleted briefly (<1 min) at 500 rpm and either immediately fixed with 200 µl cold 0.5% PFA (0 min of conjugation) or moved to 37°C water bath for 5, 10 or 15 min. Pellets were resuspended by vigorous agitation, and 3–10×105 cells were analyzed by FACS. The percentage of NK cells in conjugates was determined as: average (% conjugates at t min/(% conjugates at t min + % unconjugated NK cells at t min)) − average (% conjugates at 0 min/(% conjugates at 0 min + % unconjugated NK cells at 0 min)). All time points were assayed in triplicate and only samples in which the E:T ratio was similar (± 0.25) to control cells were compared. In live cell assays, cells were treated as above except that NK cells were stained with 0.4 µM CTGreen (Invitrogen). Instead of fixing cells following centrifugation, cell pellets were moved to 37°C for 3, 5, 10, 15, 30 or 60 min, when they were resuspended by vigorous flicking and immediately analyzed by FACS.
A granzyme B Elispot kit (R&D Systems) was used to measure the secretion of granzyme B from NK cells. NKL or KHYG-1 cells alone or mixed at 1:1 ratio with 721.221 cells, were incubated in PVDF membrane plates at 37°C for 4–5 hrs. Plates were developed according to manufacturer’s instructions. Spots were acquired using a CTL IMMUNOSPOT Analyzer (Immunospot, Cleveland, OH) with CTL Image Acquisition v4.6 and the data were analyzed using CTL 5.0 Academic software (Immunospot). The mean spot number (from 2–3 wells/plate) for unstimulated NK cells was subtracted from the mean spot number (from 2–3 wells/plate) for target cell-stimulated NK cells to generate the mean change in spot number.
6×105 NKL cells and 3×105 721.221 cells were mixed together without centrifugation for 30 min at 37°C in a 15 ml conical tube, allowed to sediment onto poly-L-lysine-coated slides (BD Pharmingen) for 10 min at 37°C and fixed with 3% PFA + 0.1% Triton X-100 for 15 min at RT. Slides were blocked with PBSS (PBS, 1% BSA, 0.1% saponin, 0.4% sodium azide) for at least 10 min, and then stained with rabbit anti-pericentrin pAb (Abcam) and mouse anti-perforin mAb (BD Biosciences) for 1 hr at RT. Subsequently, slides were washed and stained with Alexa Fluor 350–conjugated highly cross-adsorbed goat anti–rabbit IgG (Invitrogen), Alexa Fluor 488–conjugated highly cross-adsorbed goat anti–mouse IgG (Invitrogen) and Alexa Fluor 635–conjugated phalloidin (Invitrogen) for 45 min at RT. Slides were rinsed and covered with 0.16 mm thick glass coverslips (Fisher) using Fluoromount-G (Southern Biotech, Birmingham, AL). Cell conjugates were visualized using a Nikon TE300 inverted microscope fitted for phase contrast and epifluorescence (Nikon Instruments, Melville, NY) including a ProScan II filter/shutter/objective z-step controller (Prior Scientific, Cambridge, UK) and CoolSnap HQ CCD camera (Photometrics, Tucson, AZ). The NK cells in each conjugate were identified by the presence of perforin staining. Only NK cells in contact with a single target cell, with both cells in the same x-y plane, were analyzed. For three-dimensional analysis of MTOC distance, a minimum of 25 images in 0.5 µm increments through z-space were collected for each conjugate. The resulting image stack for each fluorescent channel was max projected into single-plane images, and then each fluorescent channel was overlayed into a single image. The max projection of pericentrin staining was thresholded and the distance from the centroid of the MTOC(s) to the center of the NKIS was measured using MetaVue software version 6.2r6 (Molecular Devices, Sunnyvale, California). For cells with multiple MTOCs, the distances were averaged. At least 15 conjugates per slide were randomly chosen and analyzed in this fashion for each experiment.
Comparisons of groups of data were performed using the pairwise, two-tailed Student's t test.
To examine the role of SHP-2 in NK cell function, four SHP-2 small hairpin RNAs (shRNAs) were designed and tested. The shRNAs (denoted #1–4) map to the N-terminal SH2 domain (#1), the C-terminal SH2 domain (#2 and #3), and a region that lies N-terminal of the phosphatase domain (#4) of SHP-2. These shRNAs were expressed in NK-like cell lines from pSuperior.retro (puro or neo) retroviral expression vectors. SHP-2 shRNA expressing cells were selected with antibiotic treatment for five days following retroviral transduction. The ability of the shRNAs to decrease SHP-2 protein levels was tested, both individually (Figures 1A and 1B) and in combination (Figure 1C), in the IL-2 dependent human NK-like cell line, KHYG-1 (61). As summarized in Figure 1B, all four shRNAs decreased SHP-2 protein levels to varying degrees, with #1 and #4 being the most effective. Co-transduction of shRNAs #1 and #4 (shRNA 1&4) further improved the knockdown of SHP-2 to <10% of wild-type levels (Figure 1C and 1D). This combination was used in all subsequent experiments unless otherwise indicated, and cells expressing these constructs will be referred to as SHP-2-silenced cells.
SHP-2 silencing was specific and shRNA-dependent, since cells transduced to express empty pSuperior vector had similar SHP-2 levels as the untransduced control cells (Figure 1C), and the levels of another closely related phosphastase, SHP-1, were not significantly altered (Figure 1E). The expression of SHP-2 shRNAs led to a similar loss of SHP-2 in two other NK-like cell lines, NKL and NK-92 (Figure 1F and data not shown, respectively). In contrast, SHP-2 levels were greatly increased when cells were transduced to express wildtype (WT) or gain-of-function Noonan mutants of SHP-2 (E76D or N308D), but not empty retroviral expression vector (pBMN) alone (Figure 1C and 1F).
To determine the effect of SHP-2 knockdown on NK cell function, we first compared untransduced (control) and SHP-2-silenced cells in a direct cytotoxicity assay against an EBV-transformed MHC-I-deficient B cell line, 721.221 (62). Notably, SHP-2-silenced KHYG-1 and NKL cells killed the target cells significantly better than control cells (Figure 2). The enhanced cytotoxicity was observed with distinct SHP-2 shRNAs, demonstrating that the impact on cytotoxicity was not due to off-target effects of individual shRNAs (Figure 2A and data not shown). Furthermore, this increased cytolytic activity correlated with decreasing levels of SHP-2 in a concentration-dependent manner (Figures 2A and 2B). This phenotype was highly reproducible across multiple independent transductions and was specific to the SHP-2 shRNAs, since the empty vector control did not significantly impact upon cytolytic activity (Figure 2E). In accordance with the increased cytotoxicity seen in SHP-2-silenced cells, overexpression of wildtype SHP-2 decreased the killing of 721.221 targets in a concentration-dependent manner compared to untransduced cells or compared to cells transduced with vector alone (Figure 2C–F). Overexpression of gain-of-function mutants of SHP-2 (E76D or N308D that are commonly found in Noonan syndrome) further decreased cytolytic activity (Figure 2E and 2F). Together these data indicate that SHP-2 restrains the cytolytic activity of two distinct NK cell lines in a concentration-dependent manner.
We attempted to knockdown SHP-2 in primary human NK cells by introducing these shRNA constructs with retroviral transduction and with the Amaxa Nucleofector system. Primary NK cells transduced/nucleofected with SHP-2 shRNAs, however, never survived drug selection. Cell death was not due to nucleofection alone or drug toxicity, since cells transfected with pMax control vector (Amaxa) survived and primary NK cells demonstrated the same drug sensitivity as KHYG-1 cells (data not shown). These data suggest that acute silencing of SHP-2 in human primary NK cells may be toxic.
We next addressed whether cytokine production by the NK-like cell lines could be impacted by varying levels of SHP-2. IFN-γ, a type I cytokine, is readily secreted by NK cells after activation (18). The 721.221 target cells were used to stimulate either parental NKL or KHYG-1 cells or the same lines expressing either SHP-2 shRNAs or SHP-2 cDNAs, and then the resulting IFN-γ production was quantified by ELISA (Figures 3A and 3B). Silencing of SHP-2 significantly increased IFN-γ production as compared to untransduced control NK cells, while the overexpression of SHP-2-WT or – E76D significantly suppressed the production of IFN-γ (Figure 3A and 3B). Therefore, similar to the affect on cytotoxicity, SHP-2 also suppressed IFN-γ production in a concentration-dependent manner in response to encounter with transformed target cells.
Conjugation with tumor targets results in the ligand engagement of multiple receptors on the NK cell surface, and co-engagement of both adhesion receptors (e.g. LFA-1) and activating receptors (e.g. NKp44 and NKG2D)] results in optimal NK activation (5, 63). LFA-1 is an integrin that is important in mediating adhesion of NK cells to target cells (64). NKp44 and NKG2D are activating receptors that promote IFN-γ production through the DAP12/Src family kinase/Syk family kinase cascade (65) and DAP10/phosphatidylinositol 3-kinase/Grb2-mediated signaling (3), respectively. To determine if SHP-2 acts through one or all of these specific signaling pathways, we stimulated SHP-2-silenced and SHP-2 overexpressing KHYG-1 cells with plate bound antibodies against each receptor alone or in combination and measured IFN-γ production by ELISA (Figure 3C). Surprisingly, in contrast to the collective engagement of multiple activating receptors that occurs when NK cells are stimulated by target cells, alterations in SHP-2 levels did not influence IFN-γ production in response to stimulation through antibody-mediated engagement of the NKp44, NKG2D, and/or LFA-1 receptors.
A wealth of data indicates that the recruitment of SHP-1 and SHP-2 to KIR blocks the tyrosine kinase-mediated activation events in NK cells. Previously, we demonstrated that transiently over-expressing catalytically inactive SHP-2 (DN-SHP-2) in NK-92 cells significantly suppressed KIR-mediated inhibition of cytotoxicity (27). To determine whether KIR function was similarly affected in SHP-2-silenced cells, KHYG-1 cells were transduced to express KIR3DL1 in either wild-type (WT) or an ITIM mutant (YF) form. Mutation of the COOH-terminal ITIM tyrosine (Y) to phenylalanine (F) in the YF mutant results in a receptor that was previously shown to only mediate inhibition through recruitment of SHP-2, while wildtype KIR3DL1 functions through recruitment of both SHP-2 and SHP-1 (27). Control and SHP-2-silenced NK cells expressing these receptors were compared in a direct cytotoxicity assay with 721.221 target cells lacking or expressing the KIR3DL1 ligand, HLA-B51. Consistent with our results in Figure 2, SHP-2 silenced cells exhibited enhanced cytotoxicity whether KIR were engaged or not (Figure 4A–C). In contrast to our previous findings with DN-SHP-2, however, KIR-dependent inhibition of cytotoxicity was intact in SHP-2-silenced cells expressing either WT or YF receptors (Figure 4A–C). Moreover, the degree of inhibition by KIR engagement in control vs. SHP-2 silenced cells expressing KIR-WT (50.8% vs. 44.0%, p = 0.32; n = 10 experiments) or -YF (31.2% vs. 33.4%, p = 0.76; n = 10 experiments) was essentially indistinguishable. To clarify our current and previous findings, we performed immunoprecipitation of KIR from lysates of pervanadate-stimulated cells and found that a detectable amount of the remaining SHP-2 was still recruited to phosphorylated KIR3DL1-WT or -YF in the SHP-2-silenced cells (Figure 4D). Consequently, we conclude that either SHP-2 is dispensable for KIR-mediated inhibition of cytotoxicity or only very low levels of SHP-2 protein are necessary to mediate effective inhibitory KIR signaling. Despite the considerable degree of SHP-2 silencing using our currently available tools, we cannot distinguish between these two possibilities.
We hypothesized that SHP-2 may suppress NK cell cytotoxicity by diminishing their capacity to conjugate with target cells. To test this hypothesis, we analyzed the capacity of NK cells with varying levels of SHP-2 to bind to 721.221 target cells using two-color FACS analysis (Figure 5). Knockdown of SHP-2 with shRNAs #1 and #4 alone or in combination, however, did not significantly impact the conjugation of KHYG-1 cells to 721.221 target cells (Figure 5B and C). Similarly, SHP-2 silencing did not potentiate the degree of conjugation of NKL cells to 721.221 target cells (Figure 5D). Results were similar when conjugates were fixed just prior to FACS analysis (Figure 5) or analyzed freshly over a time course of 60 min (data not shown). These data demonstrate that the enhanced cytotoxicity observed in SHP-2 silenced cells is independent of effects on adhesion. Surprisingly, in contrast to the effect observed in SHP-2 silenced cells, overexpression of wildtype SHP-2 in KHYG-1 or NKL cells significantly decreased extent of conjugation to 721.221 target cells (Figure 5C and 5D). Therefore, the impact of SHP-2 on target cell conjugation is only evident when the phosphatase is expressed at higher than physiological levels.
NK cells can kill tumor or virus-infected cells through the targeted exocytosis of lytic granules, which releases proteins that compromise the target cell membrane (e.g. perforin) and activate caspase-mediated apoptosis (e.g. granzyme B). The blockade of vesicle exocytosis with inhibitors or mutations in the exocytic machinery decreases target cell killing (66). Similarly, loss of perforin or granzymes either blocks or decreases the killing of target cells, respectively (67, 68). We hypothesized that SHP-2 might suppress granule release in NK cells, which would thereby dampen cytolytic activity. To test this hypothesis, SHP-2-silenced, control and SHP-2 overexpressing NK cells were incubated with or without 721.221 target cells for 4 hrs, and the resulting granzyme B release was quantified by ELISPOT (Figure 6). Substantially greater spontaneous granzyme B release was observed in KHYG-1 cells, as compared to NKL cells (Figure 6A), which is consistent with a previous report that KHYG-1 cells have pre-docked granules (69). Nonetheless, following stimulation with target cells, granzyme B release was significantly elevated in both NK cell lines as compared to unstimulated cells (Figure 6A). Consistent with our cytotoxicity results, SHP-2-silencing significantly enhanced the target cell-mediated release of granzyme B, while overexpression of wild type or Noonan mutants of SHP-2 significantly suppressed granzyme B release in both cell lines (Figures 6B and 6C). These data indicate that SHP-2 suppresses cytolytic granule release in response to target cell conjugation.
We next hypothesized that the SHP-2-mediated suppression of granzyme B release could be caused by altered polarization of the cytolytic apparatus toward the target cells. To kill target cells, NK cells first polarize the cytoskeleton toward the target cell, which moves the cytolytic granules to the NK immune synapse (NKIS) for release at the interface with the target cell membrane (8). Cytolysis can be abolished by blocking cytoskeletal rearrangement with drugs or inhibitors (55, 70, 71), and genetic mutations in NK cells that prevent granule polarization toward the target cell can severely decrease cytotoxicity (66). Therefore, we analyzed MTOC polarization in NK-target cell conjugates, since MTOC polarization provides a defined focal point for measurement and is a critical initiating event for orienting the cytolytic granules toward the target cell (8). SHP-2-silenced and SHP-2 overexpressing (WT or E76D) NKL cells were conjugated with 721.221 target cells for 40 min, then fixed and stained for the NKIS (F-actin), granules (perforin) and the MTOC (pericentrin). The conjugates were imaged by fluorescence microscopy, and the average distances between the MTOC and the center of the NKIS were then measured within the NK cells (Figure 7A). Within each population, we observed a wide distribution in MTOC distance, since each population encompassed a heterogeneous group of NK cells at various stages of conjugation. Although, all three populations of NK cells exhibited a similar range in MTOC distances (1–13 µm), the mean MTOC distance was significantly shorter in SHP-2-silenced cells as compared to SHP-2-overexpressing cells (Figure 7D). A concentration response relationship was observed in which a greater level of wildtype or gain-of-function SHP-2 correlated with reduced cytolytic activity and a concomitant further distance of the MTOC (Figure 7B and 7C). In fact, when the fractions of individual measurements within each cell population were compared (Figure 7D), a clear concentration-dependent shift toward more cells exhibiting greater MTOC-NKIS distance was observed: shRNA-silenced < SHP-2-WT-expressing < SHP-2-E76D-expressing NK cells. These data demonstrate that SHP-2 suppresses polarization of the cytolytic machinery toward the NKIS in a concentration-dependent manner.
In this study, we describe a previously unknown KIR-independent role for the SHP-2 phosphatase in suppressing the general responsiveness of NK cells toward tumor target cells. Using shRNA silencing and overexpression of wildtype and gain-of-function SHP-2 Noonan mutants in NK-like cell lines, we found that SHP-2 suppresses both cytolytic activity and cytokine production in a concentration-dependent manner. This role for SHP-2 was observed in two distinct NK-like cells lines, indicating that SHP-2 likely plays a general role in suppressing the activation potential of NK cells. Furthermore, SHP-2 silencing increased cytotoxicity even when KIR were engaged (Figure 4), indicating that the depletion of cellular levels of SHP-2 reduces the general NK cell activation threshold in a KIR-independent manner. Since knockdown and overexpression of SHP-2 affected MTOC polarization and granzyme B release in both NK cell lines tested, we conclude that SHP-2 is playing an inhibitory role at the level of pathways controlling cytoskeleton rearrangement and granule exocytosis. Alternatively, we observed suppressed conjugation when SHP-2 was overexpressed, but no affect when SHP-2 was silenced. This result indicates that SHP-2 may also inhibit the pathways controlling adhesion to target cells, but only when overexpressed at non-physiological concentrations. SHP-2, through control of FAK and Rho, has been shown to both promote and inhibit remodeling of the actin cytoskeleton (72). Therefore, we postulate that SHP-2 suppresses NK function through blocking the polarization of the cytoskeleton in response to target cell conjugation. Our data demonstrating greater impacts when expressing the gain-of-function Noonan mutants in most assays also suggest that SHP-2 catalytic activity plays a role in the suppression of cytotoxicity (Figure 2), granule release (Figure 6), and MTOC polarization (Figure 7).
Our previous work using dominant negative SHP-2 constructs identified a role for SHP-2 in KIR-dependent inhibition of NK cell activation (27). However, results from those studies were not definitive since over-expressed DN-SHPs would competitively block the binding of any other inhibitory effector molecules to phosphorylated ITIMs of KIR. Using shRNA-mediated silencing of SHP-2 in our current study, we were surprised to still observe potent inhibition through a mutant KIR (YF) that was previously shown to rely solely on SHP-2 recruitment for function (Figure 4B–C). Our biochemical evidence, however, showed that SHP-2 can still be recruited to tyrosine phosphorylated KIR in these shRNA-transduced cells, despite significant reductions in expression of the phosphatase (Figure 4D). Therefore, it is plausible that only minimal levels of SHP-2 protein are required to mediate KIR inhibition. Our previous work showed that SHP-2 is constitutively, albeit weakly, associated with unphosphorylated KIR (27), which may reflect a mechanism by which low cellular levels of SHP-2 can still efficiently contribute to inhibitory function by the receptor. However, we cannot at this time rule out the possibility that SHP-2 is dispensable for KIR-mediated inhibition of cytotoxicity and that another effector enzyme is contributing to KIR-mediated inhibition, particularly inhibition by the YF mutant, which cannot recruit SHP-1. It is important to note that we used NK cells that lacked KIR expression for all experiments assessing general NK cell functions (Figure 2, Figure 3, Figure 5–Figure 7), further demonstrating that the SHP-2 effects are KIR-independent.
We also found that higher SHP-2 levels reduced IFN-γ responses by NK cells in response to target cells, but not in response to crosslinking of the ITAM-dependent activating receptor, NKp44, the DAP10-dependent receptor, NKG2D, or the adhesion receptor, LFA-1, with plate-bound antibodies (Figure 3). Although the phosphoprotein targets of SHP-2-dependent suppression of NK cell activation are unknown, these data suggest that the ITAM- or DAP10-mediated receptor signaling pathways are not directly impacted. Multiple NK cell receptors, including adhesion and activating receptors, are engaged upon target cell conjugation, and these receptors activate a wide array of signaling pathways and biological events. We attempted co-engaging both activating and adhesion receptors to mimic target cell binding, but IFN-γ production in response to co-engagement was still not significantly impacted by changes in SHP-2 levels (Figure 3C). It must be stressed, however, that antibody-mediated engagement is biophysically distinct from the dynamic receptor-ligand engagement occurring during target cell conjugation. The NKIS is known to functionally segregate signaling molecules at the target cell interface (8, 10), which may not be mimicked by immobile plate-bound antibodies. The disparate results also provide further evidence that SHP-2 imparts significant impact on NKIS-directed cytoskeletal rearrangement in response to target cell conjugation (as in Figure 7), whereas antibody-mediated stimulation may not require such a directed response and may therefore be insensitive to alterations in SHP-2 levels. We cannot rule out the possibility that an inhibitory receptor other than KIR is facilitating SHP-2-mediated inhibition, but if such a receptor exists, it would be independent of MHC-I ligand, which is lacking on the 721.221 target cells used in our experiments. Dissection of the receptor pathways and biochemical events through which SHP-2 is acting will be the focus of our future studies.
We were surprised that NK-target cell conjugation was not enhanced by SHP-2 silencing in either cell line (Figure 5), since this phosphatase has been reported to modulate integrin-dependent adhesion in other cell types (73, 74). In contrast to SHP-2 silencing, SHP-2 overexpression significantly reduced conjugation of KHYG-1 and NKL cells to target cells. These results indicate that physiological levels of SHP-2 are not negatively influencing target cell conjugation, whereas these levels are blunting cytotoxicity, IFN-γ production, MTOC polarization, and granzyme B release. The literature describing impacts of SHP-2 on integrin functions contains inconsistencies, which may be due to the expression of dominant negative or truncated forms of SHP-2 in many of these studies (75, 76). Dominant negative SHP-2 lacks catalytic activity, but may still act as an adaptor to recruit SHP-2 substrates/binding partners without dephosphorylating them. Furthermore, some early studies used SHP-2 knockout mouse embryonic fibroblasts in which a truncated form of SHP-2 protein lacking the N-terminal SH2 domain was still expressed (77), thereby potentially disrupting the autoinhibited state to generate a gain-of-function mutation (78). In view of these findings, these early studies may have established mutations with unanticipated biological impacts. In our studies, we generated a battery of NK cells expressing a range of SHP-2 concentrations from <10% of normal (shRNA expression) through >500% of normal expression of either wildtype or gain-of-function forms of the phosphatase to demonstrate a significant impact only under overexpressed conditions. We conclude that SHP-2 does not suppress NK cell adhesion to susceptible tumor target cells under physiological expression conditions.
Overall, this report demonstrates a novel, KIR-independent role for SHP-2 as a general inhibitor of NK cell responsiveness. SHP-2 suppressed MTOC polarization and granzyme B release in response to target cells, and this directly correlated with decreased cytolytic activity in a concentration-dependent manner. In support of previous observations by Wülfing et al. (11), we show a direct connection between MTOC polarization and cytotoxicity by NK cells. MTOC polarization is required for target-directed granule release, indicating that SHP-2 likely suppresses granzyme B release indirectly due to an earlier (more upstream) block of MTOC polarization. In addition, this work indicates that therapeutic agents that limit SHP-2 levels or catalytic function may be beneficial to stimulate NK cell activity in patients with cancer or viral infections. Furthermore, our data with the gain-of-function mutants suggest that NK cells may exhibit suppressed function in Noonan syndrome patients, where these mutations where originally identified.
We thank Drs. Glenn Rall, Maureen Murphy, Shahjahan Miah, Alexander MacFarlance IV, Alana O-Reilly and Ginger Young (Fox Chase Cancer Center, Philadelphia, PA) for constructive criticism and comments during preparation of this manuscript. We thank the following research facilities at FCCC for reagents and technical support: DNA sequencing, DNA synthesis, cell culture, and cell sorting. We also thank Drs. Jordan Orange and Pinaki Banerjee (The Children’s Hospital of Philadelphia, Philadelphia, PA) for discussion and technical help with the NK-target cell conjugation and MTOC polarization assays. Finally, we thank Dr. Matthias Schnell and Ms. Celestine Wanjalla (Thomas Jefferson University, Philadelphia, PA) for use of the CTL IMMUNOSPOT apparatus, Dr. John Taylor (FCCC) for use of the Odyssey Infrared scanner, Drs. Ben Neel (Ontario Cancer Institute, Ontario, Canada) and Frank David (previously of Beth Israel Deaconess Medical Center, Boston, MA) for providing the WT SHP-2 cDNA and pSuperior.retro.puro.SHP-2 shRNA #1 construct, and Drs. Garry Nolan and Marco Colonna for additional reagents.
1This work was supported by grants CA083859 (K.S.C), CA009035-32 (A.K.P) and partially by Centers of Research Excellence grant CA06927 from the National Institutes of Health. The research was also supported in part by an appropriation from the Commonwealth of Pennsylvania. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the National Cancer Institute.
4Abbreviations: KIR, killer cell Ig-like receptor; MHC-I, major histocompatibility complex class I, MTOC, microtubule organizing center; NKIS, NK immune synapse; SHP-2, src homology region 2-containing protein tyrosine phosphatase-2.
Conflict of Interest Disclosure:
The authors declare no conflict of interest or financial interests.