IL-2 in the tumor microenvironment is necessary for Wiskott-Aldrich syndrome protein deficient NK cells to respond to tumors in vivo

To kill target cells, natural killer (NK) cells organize signaling from activating and inhibitory receptors to form a lytic synapse. Wiskott-Aldrich syndrome (WAS) patients have loss-of-function mutations in the actin regulator WASp and suffer from immunodeficiency with increased risk to develop lymphoreticular malignancies. NK cells from WAS patients fail to form lytic synapses, however, the functional outcome in vivo remains unknown. Here, we show that WASp KO NK cells had decreased capacity to degranulate and produce IFNγ upon NKp46 stimulation and this was associated with reduced capacity to kill MHC class I-deficient hematopoietic grafts. Pre-treatment of WASp KO NK cells with IL-2 ex vivo restored degranulation, IFNγ production, and killing of MHC class I negative hematopoietic grafts. Moreover, WASp KO mice controlled growth of A20 lymphoma cells that naturally produced IL-2. WASp KO NK cells showed increased expression of DNAM-1, LAG-3, and KLRG1, all receptors associated with cellular exhaustion and NK cell memory. NK cells isolated from WAS patient spleen cells showed increased expression of DNAM-1 and had low to negative expression of CD56, a phenotype associated with NK cells exhaustion. Finally, in a cohort of neuroblastoma patients we identified a strong correlation between WASp, IL-2, and patient survival.

NK cells integrate signals from the environment by forming two types of immunological synapses; one inhibitory synapse mediated by inhibitory receptors and one activating lytic synapse meditated by activating receptors 15 . NK cells from Wiskott-Aldrich syndrome (WAS) patients have decreased polarization of actin, MTOC, and lytic vesicles in the synapse interface to target cells 16,17 . The tumor incidence in WAS is estimated to be 13-22% with a poor prognosis and most frequently associated with lymphoreticular tumors including non-Hodgkin lymphoma (76% of the total tumors associated with WAS), Hodgkin disease, and Burkitt lymphoma [18][19][20][21][22] .
WASp knockout (KO) mice bred with tumor-prone mice have accelerated onset of tumor growth and B16 melanoma cells are more metastatic in WASp KO mice 23 . In another study, breast carcinoma cells had similar tumor growth in WT and WASp KO mice 24 , however, WASp KO mice had decreased metastatic spread 24 . Thus, the data from these two studies are somewhat contradictory and the extent of WASp KO NK cell dysfunction may depend on the tumor context. Importantly, the cytolytic defect of WAS patient NK cells can be rescued by addition of exogenous IL-2 17,25 that induces phosphorylation of WAVE2 and actin polymerization 17 . This has prompted initiation of clinical trials for IL-2 treatment of WAS patients as described for the first treated patient 17 . The efficacy of IL-2 treatment in WASp deficiency relies on that NK cells develop normally, are educated correctly, and that they are responsive to IL-2 treatment in vivo. Moreover, the outcome of IL-2 treatment for NK cell-mediated killing of lymphoreticular tumors, including lymphomas that is a major cause of death in WAS patients, remains unknown.
Tumorigenesis is difficult to study in WAS patients since it develops late in life and the vast majority of patients rapidly undergo hematopoietic stem cell transplantation today. We took the approach to investigate tumorigenesis in WASp knockout (KO) animals. We found that WASp KO NK cells developed normally but had increased expression of DNAM-1, KLRG1, and LAG-3. WASp KO NK cells had decreased formation of immune synapses with lymphoma target cells and reduced killing capacity of MHC class I negative hematopoietic grafts in vivo. We identify that IL-2, either exogenously provided or produced naturally by lymphoma cells, restored killing capacity by WASp KO NK cells in vivo. NK cells isolated from WAS patient spleen cells showed increased expression of DNAM-1 and had low to negative expression of CD56, a phenotype associated with NK cells exhaustion. Finally, in cohorts of neuroblastoma patients and patients with diffuse large B cell lymphoma we identified a strong correlation between WASp, IL-2, and patient survival.

Reduced rejection of MHC class I-hematopoietic grafts in WASp KO NK mice.
To examine the tumor rejection capacity in WASp KO mice, allogeneic YAC-1 T cell lymphoma cells were injected intravenously and monitored at 5 h by in vivo imaging (IVIS). WT and WASp KO mice showed similar growth of YAC-1 cells (Fig. 1A,B). To address the role of NK cell-mediated tumor rejection in WASp KO mice, we performed a competitive assay in which we injected T cell lymphoma cells expressing MHC class I (RMA) or with reduced expression of MHC class I (TAP −/− ; RMA-S), labeled with different concentrations of CFSE (Fig. 1C). Both wildtype and WASp KO C57Bl/6 mice could efficiently reject RMA-S T cell lymphoma cells (Fig. 1D). We next performed the competitive assay using wildtype syngenic splenocytes together with syngenic MHC class I negative (β2m −/− ) splenocytes, labeled with different concentrations of CFSE (Fig. 1C). In wildtype mice, β2m −/− splenocytes will be rejected by NK cells due to missing-self recognition. While up to 80% of β2m −/− cells were rejected in wildtype mice, WASp KO mice had decreased capacity to reject β2m −/− cells (Fig. 1E). This data suggested that WASp KO NK cells failed to reject hematopoietic grafts lacking MHC class I molecules. However, if the tumor cells express multiple activating ligands and/or cytokines, such as YAC-1 and RMA-S lymphoma cells, WASp KO mice controlled tumor growth similar to wildtype mice.  26 . Moreover, WASp KO NK cells developed normally in the spleen as examined by profiles for the final maturation stages; CD27 − CD11b − , CD27 + CD11b − , CD27 + CD11b + , and CD27 − CD11b + NK cells (Fig. 2C). The inhibitory receptors expressed on the surface of NK cells educate them to recognize self MHC class I molecules and inhibit NK cell activation during an effector phase, when self MHC class I molecules are expressed. By determining the expression of self specific and non-self specific inhibitory receptors on NK cells, education can be assessed. We first examined the frequency of NK cells expressing self-MHC-I-specific inhibitory receptors Ly49C (C), Ly49I (I) or NKG2A (N) or any combination of the three (C + I + N + CI + CN + IN + CIN) versus NK cells expressing non-self-specific inhibitory receptors Ly49A (A) or Ly49G2 (G) or any combination thereof (A + G + AG). CIN + NK cells are educated in mice expressing K b D b MHC class I molecules such as C57BL/6 mice whereas AG + NK cells do not encounter a strong MHC-I ligand in mice of the C57BL/6 background and are therefore considered to be non-educated. A high proportion of NK cells (50%) from wildtype and WASp KO mice expressed the inhibitory receptors Ly49C, Ly49I, and NKG2A that recognize self H-2 b MHC class I molecules in C57BL/6 mice (CIN, Fig. 2D and supplementary Fig. 1C). Less than 10% of the NK cells from both wildtype and WASp KO mice expressed any of the non-self specific inhibitory receptors Ly49A and Ly49G2 that lack MHC class I ligand in C57BL/6 hosts (AG, Fig. 2D). One of the key features of an educated NK cell receptor repertoire in MHC class I +/+ mice is a shift towards a repertoire where many NK cells express only few (1-2) inhibitory receptors, while in MHC class I −/− mice a lot of NK cells express many (3)(4)(5) inhibitory receptors 27 . This phenomenon can be explained by the fact that an NK cell with only 1 or 2 inhibitory receptors is actually the most competent one in protecting against transformation or infection. Wildtype and WASp KO NK cells had similar frequency of cells expressing 1-2 inhibitory receptors (Fig. 2E) Fig. 2). We next determined the expression of activating and inhibitory receptors on NK cells by assessing mean fluorescence intensity by flow cytometry. Wildtype and WASp KO NK cells had similar expression of activating receptors NKp46, NK1.1, NKG2D, 2B4, FAS-L, LFA-1, and ICOS (Fig. 3A), and of inhibitory receptors Ly49A, Ly49C, Ly49G2, Ly49I, NKG2A, CTLA-4, and PD-1 (Fig. 3B). WASp KO NK cells in the spleen had a slight but significant increase of the early activation receptor CD69 when compared to wildtype NK cells (Fig. 3A). WASp KO NK cells had higher expression of the activating receptor DNAM-1 (Fig. 3A) that associates with LFA-1 on the NK cell surface and potentiates activating signals 28 . When compared to wildtype NK cells, WASp KO NK cells also showed higher expression of the inhibitory receptors LAG-3 and KLRG1 (Fig. 3B), associated with terminally differentiated cells and cellular exhaustion 10 . KLRG1 is preferentially expressed on the most mature NK cells and the majority of KLRG1-expressing cells in WASp KO mice were mature CD27 − CD11b + NK cells (supplementary Fig. 3A). Since NK cell development depends on expression of the transcription factors T-box transcription factor (T-bet) and Eomesodermin (Eomes) 29 , we examined whether WASp KO NK cells had increased T-bet and Eomes expression as an indication of a terminally differentiated and/or exhausted phenotype and found no evidence for this (supplementary Fig. 3B,C). Another feature of exhausted CD8 + T cells is increased intracellular content of Granzyme B and perforin 30 . Naïve wildtype and WASp KO NK cells had similar content of granzyme B and perforin ( Supplementary Fig. 3D,E). Together, this data shows that the receptor repertoire in WASp KO NK cells was skewed towards increased expression of DNAM-1, LAG-3, and KLRG1.

WASp KO NK cells fail to accumulate F-actin in the lytic synapse. Peripheral blood NK cells from
WAS patient have decreased F-actin 31 . However, murine WASp-deficient NK cells had normal quantity of total F-actin as determined by flow cytometry (Fig. 4A). To visualize the lytic synapse, wildtype and WASp KO NK cells were co-incubated with YAC-1 lymphoma cells and analyzed by ImageStream analysis that combines flow cytometry with microscopy ( Fig. 4B,C). The absolute number of conjugates between YAC-1 lymphoma cells and wildtype or WASp KO NK cells were similar as determined by flow cytometry (Fig. 4B) and microscopy ( Fig. 4C). Lytic synapses between WT NK cells and YAC-1 cells were characterized by F-actin in the lytic synapse interface and polarized localization of granzyme B towards the lytic synapse (Fig. 4C). In contrast, WASp KO NK cells showed reduced capacity to accumulate F-actin and granzyme B towards the lytic synapse interface with YAC-1 tumor cells (Fig. 4C). To examine the capacity of NK cells to accumulate F-actin towards an interface surface in greater detail, wildtype and WASp KO NK cells were stimulated on glass surfaces coated with antibodies to the activating receptor NKp46 and imaged by stimulated emission depletion (STED) microscopy. Wildtype NK cells accumulated F-actin towards the interface surface (Fig. 4D), while WASp KO NK cells lacked F-actin towards the interface surface (Fig. 4D).  The cytolytic defect of WAS patient NK cells can be rescued by addition of exogenous IL-2 17,25 . Wildtype and WASp KO NK cells had similar expression of the IL-2 receptor complex and activation of NK cells with IL-2 rapidly induced phosphorylation of STAT5 (supplementary Fig. 4A,B). Moreover, addition of IL-2 to wildtype and WASp KO NK cells induced phosphorylation of WAVE2 (supplementary Fig. 4C). We examined if addition of IL-2 could restore the capacity to polarize actin towards activating receptors. WASp KO NK cells treated with IL-2 for 48 hrs showed normal polarization of polymerized actin towards anti-NKp46-coated glass surfaces (Fig. 4E). The lower capacity of WASp KO NK cells to organize the actin cytoskeleton and form lytic synapses could be completely rescued by stimulation with exogenous IL-2.

IL-2 restores degranulation and IFNγ production by WASp KO NK cells.
To investigate the functional outcome of decreased synapse formation by WASp KO NK cells, NK cells were stimulated with antibodies towards the activating receptors NKp46 and NK1.1, followed by measurement of IFNγ production and CD107a, as a marker for degranulation. Wildtype NK cells stimulated with anti-NKp46 or anti-NK1.1 antibodies degranulated, and produced IFNγ (Fig. 5A). In contrast, WASp KO NK cells had reduced proportion of CD107a + IFNg + double positive cells upon NKp46 or NK1.1 stimulation (Fig. 5A). This was also associated with lower intracellular content of Granzyme B and perforin in WASp KO NK cells upon NKp46 activation ( Supplementary Fig. 3D,E). We examined if the decreased degranulation and IFNγ response in WASp KO NK cells could be related to KLRG1 expression but found that KLRG1 − , KLRG1 + , and KLRG1 + CD27 − CD11b + NK cells all showed a decreased response upon anti-NKp46 stimulation (supplementary Fig. 5). We next addressed if the impaired degranulation response of WASp KO NK cells could be rescued by addition of IL-2. Wildtype and WASp KO NK cells were pre-treated with IL-2 for 48 hours and then stimulated with anti-NKp46 or anti-NK1.1 antibodies to measure degranulation and IFNγ production. Pre-treatment with IL-2 before anti-NKp46 or anti-NK1.1 activation completely restored the degranulation capacity and IFNγ production by WASp KO NK cells (Fig. 5B). Moreover, pre-treatment of WASp KO NK cells with IL-15 and/or IL-18 also restored degranulation and IFNγ production to the response of wildtype NK cells (Fig. 5C), albeit not as efficiently as IL-2. Seeing that IL-2 restored the functionality of WASp KO NK cells, at least in vitro, we examined if IL-2 stimulation would influence expression of KLRG1. IL-2 pre-treatment for 48 hrs had no effect on KLRG1 expression by wildtype and WASp KO NK cells (supplementary Fig. 6A).

IL-2 restores tumor rejection capacity by WASp KO NK cells. We next examined if IL-2 production by lymphoma cells resulted in activation of WASp KO NK cells in vivo.
When compared to spleen cells stimulated for 4 hours with PMA and Ionomycin (Spleen+P/I), all lymphoma cell lines used here (RMA, RMA-S, YAC-1 and A20 cells) produced high quantity of IL-2 without prior stimulation, with the A20 cells producing the highest quantity by intracellular staining and by measuring secretion of IL-2 ( Fig. 6A and supplementary Fig. 6B). In vivo, A20 lymphoma tumors stained positively for IL-2 whereas B16 melanoma tumors were negative for IL-2 (Fig. 6B). To determine the response of WASp KO NK cells to lymphoma cells with high production of IL-2, we injected A20 lymphoma cells labeled with the DiR fluorescent dye into wildtype and WASp KO Balb/c mice and monitored tumor growth by IVIS during 6 days (Fig. 6C,D). WASp KO mice showed similar control of A20 tumor growth when compared to wildtype mice (Fig. 6C,D). We next tested if IL-2 pre-treatment ex vivo could restore functional capacity of WASp KO NK cells in vivo. Wildtype and WASp KO NK cells were stimulated and expanded with IL-2 for 96 hours ex vivo and then injected into mice that had received syngenic wildtype and MHC class I negative (β2m −/− ) splenocytes, labeled with different concentrations of CFSE, 24 hours earlier. WASp KO NK cells stimulated ex vivo with IL-2 rejected β2m −/− splenocytes similarly to wildtype NK cells (Fig. 6E). Transfer of WASp KO NK cells without prior IL-2 stimulation led to increased rejection capacity of β2m −/− cells in WASp KO mice although the difference did not reach significance when compared to WASp KO mice that did not receive NK cells (Fig. 6E). This suggests that only increasing the number of WASp KO NK cells could rescue at least partly, the defect in rejection of b2m −/− cells in WASp KO mice. Together, this data shows that IL-2 rescued dysfunction of WASp KO NK cells in vivo.

WAS patient spleen NK cells have altered receptor expression. WAS patient NK cells isolated from
peripheral blood mononuclear cells have decreased capacity to form the lytic synapse and kill target tumor cells in vitro 25,31 . To assess if these defects were also present in spleen, we examined spleen NK cells from a WAS patient and an age-matched healthy control. When compared to peripheral blood mononuclear cells with both CD56 bright and CD56 dim NK cells, spleen NK cells almost exclusively consisted of CD56 dim NK cells (Fig. 7A). When compared to healthy spleen NK cells, WAS patient NK cells had lower expression of CD56, leading to a large population of CD56 dim/negative NK cells (Fig. 7A). When compared to healthy spleen cells, WAS patient NK cells had lower expression of CD69, and increased expression of DNAM-1 and granzyme B (Fig. 7B). Perforin expression was similar in WAS patient and control spleen NK cells (Fig. 7B). KLRG1 was not detected on spleen NK cells. WAS patient spleen NK cells had decreased degranulation response as indicated by lower proportion of CD107a + IFNγ + NK cells upon PMA+Ionomycin stimulation and when compared to healthy donor spleen NK cells (Fig. 7C). Since we were not able to obtain tumor tissue from WAS patients, we next wanted to examine if we could correlate expression of WASp and IL-2 with tumor outcome in cancer patients. NK cells can kill neuroblastoma cells in a process dependent on DNAM-1 32 . Since both WASp KO NK cells and WAS patient NK cells had increased expression of DNAM-1, we examined correlation between expression of WASp and IL-2 with survival in a cohort of neuroblastoma patients in the R2 database (R2: Genomics Analysis and Visualization Platform; http://r2.amc.nl). By analysing a non-hematopoietic derived tumor, neuroblastoma, we focused on gene expression of WASp and IL-2 by tumor-infiltrating hematopoietic cells. In this cohort, expression of WASp correlated with expression of IL-2 (Fig. 7D). When comparing neuroblastoma patients with high expression of WASp with those that had low or absent expression of WASp, high expression of WASp correlated with better survival Scientific RepoRts | 6:30636 | DOI: 10.1038/srep30636 (Fig. 7E). Using the same analysis, high expression of IL-2 correlated with better survival in the neuroblastoma patient cohort (Fig. 7E). We next analysed a cohort of patients with diffuse large B cell lymphoma using the R2 database. Since diffuse large B cell lymphoma is of hematopoietic cell origin and expresses WASp, it was not surprising that high expression of WASp if anything led to increased mortality (Fig. 7E). Interestingly, high expression of IL-2 correlated with better survival also in the cohort of diffuse large B cell lymphoma patients (Fig. 7F). Significance was assessed with the Mann-Whitney test. *P < 05, **P < 01, ***P < 001, ns = not significant.  24 . Catucci et al. shows that defective NK cell killing could be a secondary cause of decreased cross-talk between WASp KO dendritic cells and NK cells, required for the priming of NK cells and polarized cytokine release 23,33 . The study by Ishihara et al. show that tumor-associated macrophages from WASp KO mice showed reduced migration resulting in decreased tumor metastasis 24 . The previous studies describe tumorigenesis in WASp KO mice to non-hematopoietic derived tumors, B16 melanoma and breast cancer cells 23,24 . Since almost 90% of the malignancies in WAS patients are of lymphoreticular origin, we took the approach to study the killing response by WASp KO NK cells to lymphoma cells and hematopoietic grafts. We found that WASp KO  NK cells had decreased capacity to kill MHC class I negative hematopoietic grafts. In light of this finding and the marked defects in the in vitro capacity to degranulate and produce IFNγ, it was surprising that WASp KO NK cells could control growth of lymphoma cells such as A20, YAC-1, and RMA-S cells in vivo. We reason that this could be due to that tumor cells of lymphoid origin express multiple activating ligands and/or express high quantity of cytokines such as IL-2 that rescue dysfunction of WASp KO NK cells (this study) 17,25 . This may explain why WAS patients have relatively late onset of tumor development in adolescence and adulthood [18][19][20][21][22]34 .
WASp KO NK cells showed normal distribution of the inhibitory receptor repertoire but had decreased capacity to reject MHC class I negative grafts. This suggests that despite having normal education in terms of inhibitory receptor repertoire, loss of MHC class I molecules was not enough to trigger activation of WASp KO NK cells in the wildtype:β2m −/− splenocyte rejection assay. The exact details on where NK cells are educated and which host cells they interact with remains enigmatic. It has been suggested that the localization of activating and inhibitory receptors to specific membrane microdomains steer education so that when activating and inhibitory receptors cluster together NK cells become uneducated and hypo-responsive 35 . If instead clustering of activating and inhibitory receptors occurs in separate membrane microdomains, NK cells become educated and responsive 35,36 . Our data suggests that the synapses formed during education and effector stages are regulated by different signaling molecules and that the education synapse occurs independently of WASp while the lytic synapse requires WASp for optimal signaling. This is similar to the requirement for WASp during development and function of B and T cells, where immune synapse formation is critical both during development and in cell-to-cell communication in peripheral organs. During development of B and T cells, WASp is largely dispensable and another WASp family member, neuronal(N)-WASp, acts redundantly with WASp 37,38 . It is possible that N-WASp serves redundant function with WASp or has a unique role in development and education of NK cells.
CD69 is an early activation receptor upregulated on NK cells and T cells in the early immune response. Murine WASp KO NK cells showed higher expression of CD69, while WAS patient spleen NK cells instead showed lower expression of CD69. In the murine setting, we examined CD69 on NK cells from naïve mice. In contrast, WAS patient spleen cells were probably subjected to a state of chronic inflammation as evidenced by a large proportion of WAS patient NK cells having low or absent expression of CD56 when compared to CD56 dim NK cells in spleen cells from the healthy control. CD56 negative NK cells is a rare population in healthy individuals but are expanded in chronically infected individuals 39 . When compared to CD56 dim NK cells, CD56 negative NK has lower capacity to degranulate and produce cytokines such as IFNγ 39 . We show here that WAS patient spleen NK cells showed decreased capacity to degranulate and produce IFNγ in response to PMA and Ionomycin. The limited WAS patient spleen sample did not allow us to examine receptor-mediated degranulation and IFNγ production. Murine WASp KO NK cells degranulated and produced IFNγ upon PMA and Ionomycin activation. Together with the mild defect in WAS patient spleen NK cell in response to PMA and Ionomycin, this suggest that WASp is required for proximal receptor signaling leading to degranulation and IFNγ production.
Of all the activating receptors tested in our panels, WASp KO NK cells and WAS patient NK cells had increased expression of DNAM-1. DNAM-1 is a key molecule in differentiation of memory NK cells and blocking DNAM-1 suppresses NK cell memory formation during cytomegalovirus infection 40 . Interestingly, when comparing DNAM-1 + with DNAM-1 − NK cells and their tumor rejection capacity and in vitro killing, DNAM-1 + NK cells showed superior rejection and tumor killing capacity of B16 melanoma cells and RMA-S lymphoma cells 41,42 . It is possible that increased DNAM-1 expression on WASp KO and WAS patient NK cells leads to better killing of tumor cells that express DNAM-1 ligands, which are members of the Nectin/Nectin like family of adhesion molecules 43 , such as B16 melanoma cells and RMA-S lymphoma cells. Of all the inhibitory receptors tested here, only KLRG1 and LAG-3 were markedly increased on WASp KO NK cells. Expression of KLRG1 and LAG-3 have been associated with an exhausted phenotype in CD8 + T cells and in NK cells 10,44 , implying that WASp KO NK cells may show signs of exhaustion. However, WASp KO NK cells had capacity to reject A20 and RMA-S lymphoma cells, and controlled growth of YAC-1 cells. Moreover, WASp KO NK cells remained responsive to IL-2 stimulation as determined by STAT5 signaling, WAVE2 activation, and rescued ability to polarize actin, degranulate, and produce IFNγ in response to IL-2. WASp KO NK cells showed normal expression of the transcription factors T-bet and Eomes and had similar content of granzyme B and perforin, suggesting that they did not have an exhausted phenotype 29,30 . Interestingly, a recent study shows that terminally matured CD27 low KLRG1 + NK cells are needed for effective anti-tumor immunity 14 . WASp KO NK cells had high expression of KLRG1 in mature CD27 low CD11b + NK cells when compared to wildtype mice. Together, our data suggests that WASp KO NK cells were not exhausted but that they need increased signaling to reach activation threshold and that loss of MHC class I alone was insufficient to trigger a killing response by WASp KO NK cells. The analysis of WASp KO NK cells implies that increased expression of KLRG1, LAG-3 and DNAM-1 was associated with normal tumor rejection capacity, at least when multiple ligands and cytokines such as IL-2 was produced by tumor cells.
Several strategies have been tested to target NK cells in cancer therapy 3 . All strategies aim at skewing the signaling balance towards activating and away from inhibitory input using cytokine treatment and/or transfer of ex vivo expanded NK cells 3,45,46 . We provide evidence that both these strategies may be viable to rescue NK cell dysfunction in WAS patients. IL-2 treatment of WASp KO NK cells completely restored degranulation capacity and IFNγ production in vitro, and IL-2 treatment of WASp KO NK cells ex vivo stimulated killing capacity of MHC class I negative hematopoietic grafts in vivo. Our data is promising in the light of that the responsiveness of NK cells remains plastic once the cells reach maturity and can be reset when the NK cells are exposed to an environment in which MHC class I expression differs 47 . NK cells are sensitive to the balance of the inhibitory and activating ligands, but once activation is reached, killing is an all-or-nothing response 15,48 . Targeted cytokine therapy may be an option to restore cellular dysfunctions in WAS as has been shown for many primary immunodeficiency diseases 49 . We show here that IL-2, IL-15 and IL-18, that signal via the common γ chain, restores functionality of murine WASp KO NK cells. Subcutaneous administration of IL-2 is currently in clinical trial for WAS patients and has shown positive outcome in one WAS patient 17  to the high affinity IL-2 receptor α chain, a modified version of IL-2 capable of functioning independently of the α chain has been used successfully to activate NK cell killing of MHC class I negative tumors 45,46,50 . A more feasible option may be ex vivo culture of NK cells in the presence of IL-2 where the gain is two-fold, rescue of NK cell dysfunction, such as in WASp deficiency, and expansion of NK cells 3,51 .  Technologies) and Ly49I (YLI-90) and NKG2A (20d5) (eBioscience). The 4LO3311 (Ly49C) hybridoma was a kind gift from Suzanne Lemieux. All surface staining was performed in phosphate-buffered saline (PBS) after blocking unspecific staining via FcγRII/III with purified anti-CD16/32 (2.4G2). Dead cells were excluded from the gating with the Aqua dead cell dye (Life Technologies). For intracellular staining of IL-2 and IFNγ, cells were fixed and permeabilized with the Cytofix/Cytoperm kit (BD) and then stained with IL-2 (JES6-5H4, Biolegend) and IFNγ (XMG1.2, BD Biosciences) antibodies. For pWAVE2 and pSTAT5 analysis, cells were stimulated with IL-2 for the indicated time, fixed with Lyse/Fix (BD Biosciences), permeabilized with Perm Buffer III (BD Biosciences) and stained with pWAVE2 (WP1821 Tyr-150, ECM Biosciences) and pSTAT5 (47/Stat5(pY694), BD Biosciences) antibodies. Secretion of IL-2 was determined during 45 minutes using the IL-2 secretion assay detection kit (Miltenyi). All flow cytometric analyses were performed on an LSRII or LSRFortessa X-20 (Becton Dickinson) and data analysis was performed using the FlowJo software (Tree Star). For immunohistochemistry, tissue sections were stained with IL-2-biotin (JES6-5H4, Biolegend) and Streptavidin (Life Technologies) antibodies. Images were collected with a Leica DM IRBE confocal laser scanning microscope (Leica Microsystems) equipped with 1 argon and 2 HeNe lasers, using an HC PL APO lens at 10x/0.40 CS and 90% glycerol (MP Biomedicals) and processed with Adobe Photoshop CS5 (Adobe Systems) and ImageJ software (National Institutes of Health).

Tumor cell injection and In vivo imaging. A20 B cell lymphoma cells or YAC-1 T cell lymphoma cells
were labelled with the DiR dye (Life Technologies) and 1 × 10 6 cells were injected subcutaneously or intravenously, respectively, into Balb/c WT and WASp KO mice. In vivo imaging was performed immediately after injection, and at Day 1, 2, 4 and 6 for the A20 experiments and after 5 h for the YAC-1 experiments. In vivo imaging of fluorescence was performed with a CCD camera, mounted in a light-tight chamber (IVIS Spectrum CT, Caliper LifeSciences). Anesthesia was induced by 4% isofluorane and maintained by 2.3% isofluorane throughout the imaging procedure. Regions of interest were localized around the injections sites and intensity was quantified as Radiant Efficiency in (p/sec/cm 2 /sr)/(mW/cm 2 ). Analysis was performed using the Living Image ® software version 4.1 (Caliper Life Sciences). For immunohistochemistry of A20 and B16 melanoma cells, 1 × 10 6 tumor cells were injected subcutaneously in Matrigel (BD Biosciences) and removed on day 7 for analysis of IL-2 expression.
In vivo rejection assay. Spleens from WT and β2m −/− mice were mechanically disrupted, erythrocyte depleted, and the single cell suspensions were labelled with a high (β2m −/− ) or low (WT) dose of CFSE at a 1:1 ratio 48,53 . RMA and RMA-S (a TAP-2 deficient variant of RMA with low expression of MHC class I molecules) T cell lymphoma cells were grown as ascites in irradiated C57Bl/6 mice and collected to be labelled with a high (RMA-S) or low (RMA) dose of CFSE. The cells were mixed in a 1:1 ratio and injected intravenously at 30 × 10 6 cells into WT or WASp KO C57Bl/6 mice. After 2 days, spleens from these mice were collected and the amount of rejected β2m −/− or RMA-S cells was assessed by flow cytometry and calculated as follows: 100% − (β2m −/− or RMA-S cells in sample/WT or RMA cells in sample)/(β2m −/− cells or RMA-S in inoculate/WT or RMA cells in inoculate).
In vitro NK cell stimulation. Splenocytes from WT and WASp KO C57Bl/6 mice were mechanically disrupted, erythrocyte depleted, and the cells were added to a plate pre-coated with 20 μg/ml anti-NKp46 antibody (R&D Systems), or 50 μg/ml anti-NK1.1 antibody (BD Biosciences) for 4 h at 37 °C. Cells were incubated in complete RPMI medium, supplemented with 10 mg/ml Brefeldin A (GolgiPlug; BD Biosciences) and anti-CD107a. Positive controls were stimulated with 20 ng/ml PMA (Sigma) and 50 ng/ml Ionomycin (Sigma). Cells were then harvested and stained for surface markers and intracellular IFNγ after fixation and permeabilization with the Cytofix/Cytoperm kit (BD Biosciences) and analyzed by flow cytometry. Degranulation was assessed by the surface expression of CD107a.
IL-2 treatment. Splenocytes from WT and WASp KO C57Bl/6 mice were obtained as described above and cultured in complete RPMI, at 2 × 10 6 cells/ml with 1000 U/ml IL-2 (Peprotech) for 48 h (in vitro experiments) or 96 h (in vivo experiments).
Synapse Formation and Imaging Flow Cytometry. Splenic NK cells from WT or WASp KO C57Bl/6 mice were purified by MACS sorting with CD49b beads (Miltenyi Biotech). NK cells were incubated with YAC-1 lymphoma cells at a 5:1 effector:target ratio for 1 h at 37 °C. YAC-1 cells were pre-labelled with CFSE and NK cells with anti-NK1.1. anti-CD3 and the Aqua dead cell dye were used to exclude contaminating T cells and dead cells. YAC-1 -NK cell conjugates were fixed and permeablized (Cytofix/Cytoperm kit; BD Biosciences) and stained with Phalloidin (Life Technologies) and anti-Granzyme B (GB12, Invitrogen). The conjugates were acquired on an ImageStream X Mark II (Amnis) and gated as NK1.1 + CFSE + doublets. Analysis was performed using the IDEAS v.6 software (Amnis).

Stimulated Emission Depletion (STED) Microscopy. NK cells from WT and WASp KO C57Bl/6 mice
were stimulated for 4 h on glass coverslips, coated with 20 μg/mL anti-NKp46 antibodies or PBS. The cells were then fixed, permeabilized, and labeled with Phalloidin-Alexa Fluor 488 to detect F-actin and NK1.1-APC. The imaging was performed on a Leica TCS SP8 microscope with a time-gated stimulated emission depletion (STED) module. Image analysis was performed on the Image J software.
Statistics. All statistical analyses were performed using the Prism software (version 6, GraphPad). Results are expressed as means ± SEM. All data was analyzed by ROUT (Q = 1.0%) and outliers were excluded as indicated in the Fig. legends. Differences between individual groups were analyzed for statistical significance using the non-parametrical Mann-Whitney test. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant.