Human STAT5b mutation causes dysregulated human natural killer cell maturation and impaired lytic function.

BACKGROUND
Patients with STAT5b deficiency have autoimmunity, recurrent infections and combined immune deficiency, which affects T-cell homeostasis and leads to natural killer (NK) cell impairment.


OBJECTIVE
In this study, we characterized the NK cell defect in STAT5b-deficient human NK cells as well as Stat5b-/- mice.


METHODS
We used multiparametric flow cytometry, functional NK cells assays, microscopy as well as a Stat5b-/- mouse model to elucidate the effect impaired and/or absent STAT5b has on NK cell development and function.


RESULTS
This alteration generated a non-functional CD56bright NK cell subset, characterized by low cytokine production. The CD56dim NK cell subset had decreased expression of perforin and CD16 and a higher frequency of cells expressing markers of immature NK cells. We observed low NK cell numbers and impaired NK cell maturation suggesting that STAT5b is involved in terminal NK cell maturation in Stat5b-/- mice. Furthermore, human STAT5b-deficient NK cells had low cytolytic capacity, and fixed-cell microscopy showed poor convergence of lytic granules. This was accompanied by decreased expression of co-stimulatory and activating receptors. Interestingly, granule convergence and cytolytic function was restored after IL-2 stimulation.


CONCLUSIONS
Our results show that, in addition to the impaired terminal maturation of NK cells, human STAT5b mutation leads to impairments in early activation events in NK cell lytic synapse formation. Our data gives further insight into NK cell defects caused by STAT5b deficiency.


Introduction
Natural killer (NK) cells are major effectors of the innate immune system and early effector cells against viral infections, pathogens, and malignant cells. They represent 10-15% of lymphocytes in the peripheral blood. 1-3 NK cells mature through distinguishable stages (stages 1, 2a, 2b, 3, 4a, 4b-CD56 bright , 5-Early CD56 dim , and 6-late CD56 dim ), each of which is characterized by a specific pattern of surface expression, proliferative, and functional capacities, as well as in vivo trafficking. [4][5][6][7][8] The ex vivo characterization of NK cell developmental intermediates in human secondary lymphoid tissues revealed a new schema for the stages of NK cell development. 5 This classification was proposed according to CD34, CD117, IL-1R, CD94, NKp80, CD16 and CD57 expression. 5,9 The differential NKp80 expression defines stages 4a and 4b. Both stages lack the hallmark phenotype of mature NK cells, but have the capacity to acquire properties of mature NK cells. 9 Finally, the up-regulation of CD57 expression defines two functional CD56 dim mature (stage 5&6) NK cell subsets; the "early" CD57 − CD56 dim and the "late" CD57 + CD56 dim . The CD57 + CD56 dim subset has been characterized with a more mature phenotype and a potent cytolytic capacity. 6,10 IL-15 can induce NK cell development from human bone marrow-derived hematopoietic progenitor cells and is required for the terminal maturation of fully functional NK cells. [11][12][13][14] Activation through the intracellular domain of the IL-15 receptor leads to the recruitment and phosphorylation of STAT5 proteins, as a principal downstream effector of this signaling pathway. 15,16 STAT5 proteins are phosphorylated by Janus kinase (JAK) proteins (JAK1 and JAK3). This modification leads to the formation of homo-or heterodimers that translocate to nucleus and bind to consensus sequences in the promoters of genes such as CD25, Bcl-2, and FOXP3. [15][16][17] Most notably, it is known that Stat5b −/− mice have significantly lower levels of perforin expression at baseline and greatly decreased cytolytic function in NK cells. 18 The first steps of the interaction between NK cells with their target cells are mediated by a group of receptors which include CD226 (DNAM-1), CD314 (NKG2D), CD244, and lymphocyte function-associated antigen-1 (LFA-1). [19][20][21] The transmission of activation signals through these receptors results in firm adhesion to the target cell, and progression to NK cell cytotoxicity. 20 Moreover, human NK cell subsets have different expressions of adhesion receptors that correlates with their distinct trafficking pattern and development. 4,22 In humans, STAT5b deficiency is associated with severe growth hormone (GH)-resistant growth failure and combined primary immunodeficiency. 15,[23][24][25] Affected individuals develop recurrent bacterial and viral infections, and their immunophenotype reveals lymphopenia, decreased accumulation of regulatory T cells, and decreased number of NK cells. 17,24,26 Here, we focused on the functional defects identified in NK cells of two siblings with a homozygous deletion in STAT5B. These patients have increased susceptibility to viral infections and a significantly low number of NK cells in peripheral blood. We describe a functionally impaired CD56 bright subset and an immature CD56 dim subset characterized by low levels of perforin. Additionally, we identified NK cell subsets in our patients with low levels of expression of co-stimulatory and activating receptors, which affects granule convergence to the microtubule-organizing center (MTOC). We further extended our study to Stat5b −/− mice and found similarly impaired NK cell terminal maturation. In both mouse and human, STAT5b-deficient NK cells have been characterized with increased levels of VEGF that may trigger their pro-tumorigenic potential. Together, these observations demonstrate that STAT5b plays a critical role in the terminal maturation and function of NK cells in humans and mice.

Study approval
Blood samples were collected from STAT5b-deficient patients and healthy donors under approved protocols from the Baylor College of Medicine Institutional Review Board. Consent was obtained from all patients and donors and samples were obtained under guidelines established by the Declaration of Helsinki. All peripheral blood samples were collected in sodium heparin collection tubes. PBMCs from healthy donors and patients were purified by Ficoll-Paque Plus density gradient centrifugation (GE Healthcare).

Imaging flow cytometry
BLCLs (10 6 cells) from healthy donor and STAT5b-deficient patients were stimulated with 50 ng/ml of IL-21 for 30 minutes. The cells were then fixed and permeabilized with Perm buffer III (BD), followed by an incubation with anti-spSTAT5 (pY694) Alexa Fluor 647 conjugated antibody (BD Phosflow) and DAPI (Sigma Aldrich). Samples were acquired on an ImageStream MkII (Luminex) with a 60X objective. Images were processed using IDEAS software v6.1 (Luminex).

Cytotoxicity assays
Antibody-dependent cell-mediated cytotoxicity (ADCC) and NK cell cytotoxicity were measured using the Cr 51 release assay. 29 ADCC was evaluated with the Raji cell line, which was incubated in the presence or absence of anti-CD20 (Rituximab, 20 μg/mL) and cocultured with fresh PBMCs for 4 hours at 37°C in 5% CO 2 . For natural cytotoxicity, PBMCs from patients and healthy donors were incubated for 4 hours with or without IL-2 (1000 U/mL; Proleukin-aldesleukin, Prometheus) or IL-15 (25 ng/ml Peprotech) and the K562 target cell line. Lytic units were calculated as the number of effectors required to lyse 10% of target cells by using the curves generated with the range of effector to target cell ratios.
Flow cytometry-based NK cell phenotype assay and antibodies NK cells from STAT5b-deficient patients and healthy donors were studied phenotypically by flow cytometry. To ensure inclusion of age-and gender-matched controls, data from 4 healthy female age-matched donors that were previously reported and published by our group 30 were compared to the adult healthy donor control PBMCs evaluated on the same day with the patient samples. Median fluorescence intensity (MFI) and percentage from age/ gender matched controls were found to be within the same range for MFI and percentage as HD controls acquired on the same day; these data were therefore pooled for analysis. PBMCs were stained as described previously with the same antibodies and protocol used for the age/gender matched donors. 30,31 Intracellular cytokines were evaluated in cells stimulated with PMA and Ionomycin (Sigma Aldrich) for 6 hours. Brefeldin A (final concentration 10 ug/mL -Sigma Aldrich) was added 4 hours before antibody staining. The cells were fixed and permeabilized with Cytofix/Cytoperm (BD Biosciences). For IFN-γ production by NK cells, PBMCs from healthy donor and patient 1 were stimulated with IL-12 (10 ng/ml), IL-15 (1 ng/ml), and IL-18 (10 ng/ml) by 12 hours.

VEGF-A expression in human and murine NK cells:
Human NK cells from patient 1 and a healthy donor were isolated from PBMCs using NK cell isolation kit (MACS, Miltenyi Biotec) and the total RNA was isolated using the RNeasy Mini kit (Qiagen). The expression of VEGF-A and the house-keeping gene GAPDH was analysed by qPCR. Following primers were used: hVEGFA-for: AAGGAGGAGGGCAGAATCAT; hVEGFA-rev: ATCTGCATGGTGATGTTGGA; GAPDH-for: TCTCCTCTGACTTCAACAGCG; GAPDH-rev: ACCACCCTGTTGCTGTAGCC. For in vivo tumor model Stat5b +/+ (WT) and Stat5b +/− mice were subcutaneously injected with 10 6 RMA cells and the tumor growth was monitored. 10 days post injection the mice were sacrificed and the tumors were weighed and snap-frozen. The VEGF-A protein expression was analysed in tumor lysates by ELISA (R&D Systems) according to manufacturer's protocol and normalized to the total protein amount.

Fixed-cell microscopy assay
PBMCs (10 6 cells) were co-cultured with 10 5 target cells for 15 minutes in presence or absence of IL-2 (1000 UI/ml) or IL-15 (25 ng/ml), and then adhered to silane-coated slides for 20 minutes at 37°C. Fix ation, permeabilization, and staining were performed as described 33 . The antibodies were used in the following sequence: 1) biotinylated monoclonal mouse anti-tubulin (Invitrogen); 2) streptavidin Alexa Fluor 647 (Invitrogen), Alexa Fluor 488-conjugated mouse anti-perforin clone δG9, and Alexa Fluor 568-conjugated phalloidin (Invitrogen). The slides were mounted with Prolong Gold Antifade reagent (Life Technologies). Cellular conjugates were imaged on a SP8 laser scanning confocal microscope (Leica Microsystems) equipped with 100X 1.4NA objective, a white light laser illumination source and adjustable HyD detectors tuned to the excitation/emission properties of each fluorescent dye. For granule convergence analysis, the lytic granules were detected using the perforin fluorescent immunostaining channel and the MTOC was detected based on the brightest object detected in alpha tubulin immunostaining fluorescent channel. Measurement of the average distance between granules to MTOC as a marker of convergence was performed using Volocity as described previously 34 , directly on the unprocessed raw data. Measurement of MTOC polarization was performed using Fiji. 35 Briefly, the center of the synapse was identified as the centroid of the F-actin positive structure at the interface between the NK cell and the target for each conjugate and the location of the MTOC as the centroid of the brightest dot-shapped element in the alphatubulin channel. The Euclidean distance between the two points was measured, plotted and analyzed in Prism version 8.1.1 (GraphPad Software Inc). Measurement of the accumulation of F-actin at the synapse was performed using Fiji on the raw 3D volumes. Identically sized region of interest were then selected: one at the immune synapse encompassing only the NK cell side and the other one on the opposite side of the NK cell, away from any synaptic structure. The ratio was calculated, plotted and analyzed in Prism. Measurement of total perforin content by fluorescence microscopy was also performed using Fiji. An area including all the cytoplasm was drawn and the total fluorescence intensity was measured. The same area was then captured in the target cell to account for background due to the nonspecific staining of the perforin antibody. Following background subtraction, data was plotted and analyzed in Prism. Raw images were subjected to signal re-scaling exclusively using linear transformation for display purpose in the figures.

Human STAT5b mutation results in a dramatic reduction of NK cell number in peripheral blood.
In this study, we evaluated two patients with STAT5b deficiency. These patients were originally reported in 2007 by Hwa and co-workers 36 (Table I, and see supplementary data  and supplementary table E1 for their clinical details and immunological profiles). STAT5 has two isoforms, STAT5b and STAT5a. The STAT5b protein is encoded by the STAT5B gene located approximately 12 kb apart from the STAT5A gene on chromosome 17. 37 To asses the effect of the variant in STAT5B (c.1680delG), first, we evaluated the protein expression by Western blot in peripheral blood mononuclear cells (PBMCs) from both patients. This showed a residual expression of STAT5b in primary cells (Fig 1, A) from both STAT5bdeficient patients; we went on to assess the level of STAT5a expression. Interestingly, we observed STAT5a expression in healthy donors and PBMCs from Pt 1 and Pt 2 (Fig 1, A).
To determine the effect of c.1680delG variant on STAT5a and STAT5b expression, we generated B lymphocyte cell lines (BLCL) from both patients. STAT5b expression was absent in BLCLs with comparable levels of STAT5a to those found in BLCLs from healthy donors (see Fig E1, A).
Since we observed similar STAT5a protein levels in BLCLs from patients and healthy donors (see Fig E1, A), we evaluated the patient's B cell lines by imaging flow cytometry using a total phopho-STAT5 antibody to determine the activation and nuclear recruitment of STAT5a dimers. In the absence of stimulation, STAT5a activation was not detected in BLCL from patients and healthy control. Stimulation with IL-21 increased localization of STAT5a in nucleus (see Fig E1, B). This was also observed for healthy donor and patient BLCLs, the colocalization between DAPI and STAT5a as calculated by similarity score of 66.89% for healthy donor, 54.18% for Pt 1, and 62.85% for Pt 2 (median values) (see Fig E1, C).
STAT5b is a principal effector molecule downstream of IL-15, a critical cytokine for NK cell maturation, survival, and proliferation. 11,15 To determine the frequency of NK cells within peripheral blood, we performed flow cytometry analysis of lymphocytes from STAT5bdeficient patients and from six healthy donors (HD) pooled together; two adult HD and 4 matched by age and gender reported previously by our group given MFI and percentage from age/gender matched controls were found to be within the same range for MFI and percentage as same day HD controls. 30 NK cells were identified in the lymphocyte gate as CD56 + CD3 − and further characterized as CD56 bright CD3 − and CD56 dim CD3 − NK cell subsets (Fig 1, B, and see Fig E2 for our gating strategy). Gating on CD56 + CD3 − showed a significant decrease in total NK cell numbers in both siblings (Pt 1, 4.680 ± 1.684%; Pt 2, 3.490 ± 0.7930%; mean ± standard deviation) compared to healthy donors (10.44 ± 0.5138%) (Fig 1, C). Together, these analyses demonstrate that STAT5b expression is absent in STAT5B (c.1680delG) deficient NK cells, and led to reduced NK cell frequency in the peripheral blood. Furthermore, STAT5a expression is retained in both patients and was detectable at similar levels compared to healthy donors, thus showing that STAT5a and STAT5b are not fully redundant.
Human STAT5b mutations result in decreased perforin expression and impaired terminal NK cell maturation.
The CD56 dim subset represents the majority of circulating NK cells which have cytotoxic capacity enabled by expression of CD16 (FcγRIII), perforin, and granzyme B. 2,4,22 To determine if the decreased NK cell frequency seen in peripheral blood from STAT5bdeficient patients is associated with an abnormal NK cell phenotype, we performed extensive flow cytometric analysis. We identified a CD56 dim subset with significantly decreased median fluorescence intensity (MFI) for perforin expression in both patients (Fig  1, D, and see Fig E3, A). Interestingly, this difference persists when we analyzed the percentage of CD56 dim perforin + NK cells between healthy donors and the patients (see Fig  E3, B). Therefore, human STAT5b mutations affect the frequency of perforin-expressing NK cells in peripheral blood and the median intensity of perforin expression in CD56 dim NK cells.
Changes in the expression of cell surface receptors mark the transition from CD56 bright to CD56 dim NK cells. 2 Our analyses identified a seemingly immature CD56 dim NK cell subset with low expression of CD8, CD16, CD57, and granzyme B (Fig 1, D). Furthermore, this subset exhibited higher levels of markers classically found on immature CD56 bright cells, such as CD94, CD117, and CD215 (IL-15Rα) (Fig 1, D). The same differences were observed between healthy donors and STAT5b-deficient patients when the percentage of positive cells for each receptor was evaluated (Fig E3, B). The impairment in NK cell maturation observed in these patients was accompanied by high expression levels of CD215 (IL-15Rα) (Fig 1, D), which is necessary for human NK cell development. 11

Stat5b −/− mice have a block in terminal NK cell maturation.
In murine bone marrow, natural killer precursor cells (NKps, NK1.1 − and NKp46 − ) are characterized as Lin − CD122 + cells. 38 The immature NK cells (iNKs) are characterized by acquisition of the NK1.1 receptor, followed by the mature NK cells (mNKs) characterized by an upregulation of NKp46 and CD49b. 39 As NK cells mature further, they acquire CD27 expression early and lose CD27 expression as the cells mature. 40,41 Ultimately, cells migrate to the periphery, upregulate CD11b, and acquire their functional capacity. All of these incremental changes represent the critical landmarks of the last stages of mouse NK cell development. 39 To correlate the impairment in NK cell maturation observed in humans with the phenotype of Stat5b −/− mice, we analyzed the NK cell development progression from Stat5b −/− mice by flow cytometry. For NK cell development in early stages, we investigated the frequency of precursor and immature NK cells in bone marrow. Interestingly, we observed that Stat5b −/− mice, but not Stat5a −/− mice, had a significantly decreased percentage of NK cell precursors (0.01 ± 0.004%) and immature NK cell (0.02 ± 0.007%) compared to wild-type mice (NKps: 0.03 ± 0.01% and iNKs 0.05 ± 0.01%) (Fig 2, A).
In the spleen from Stat5b −/− mice, the percentage of mature NK cells was lower (1.86 ± 0.9371%; mean ± SD) compared to Stat5a −/− mice (3.631 ± 1.15%) and wild-type controls (4.072 ± 0.9889%) (Fig 2, B). Spleens from Stat5b −/− mice, but not Stat5a −/− mice, had a significantly decreased frequency of mature CD27 − CD11b + NK cells (11.85 ± 4.791%) compared with wild-type controls (47.34 ± 13.32%) (Fig 2, C). Moreover, spleens from Stat5b −/− mice had a larger percentage of immature CD27 + CD11b − NK cells (27.7 ± 7.789%) compared to wild-type controls (12.81 ± 4.04%) (Fig 2, C). These findings support our results seen in the peripheral blood NK cells of STAT5b-deficient patients with less circulating NK cells compared to healthy donor. These results show that STAT5b is important in early NK cell development. The low number of mature NK cells in spleen is a consequence of the lower numbers of mature NK cells seen in bone marrow (Fig 2, A) and suggests that the paucity of precursor cells in the first stage of NK cell development hinders the transition to mature NK cells.
Cytokine production is largely affected in CD56 bright NK cell subset by human STAT5b mutations.
CD56 bright NK cells provide early IFN-γ and other cytokines to antigen-presenting cells, thereby promoting control of infections. 22,42 In response to PMA and ionomycin, the CD56 bright NK cell subset produces significantly higher levels of cytokines compared with the CD56 dim NK cell subset. 4,22,42 However, in response to these stimulus we observed significantly lower percentages of CD56 bright NK cells positive for interferon (IFN)-γ, tumor necrosis factor (TNF)-α and GM-CSF (see Fig E4, A-B), with low median fluorescence intensities (MFI) for each cytokine (Fig 3, A). NK cells are innate lymphocytes with a higher capacity to produce immunoregulatory cytokines, specially IFN-γ in response to IL-12 and IL-18 in the presence of low-dose of IL-15 as a survival factor. 43,44 Further, stimulation with IL-12/15/18 did not show an increase in the percentage of IFN-γ producing CD56 bright NK cells in STAT5b-deficient NK cells (see Fig. E4, C). The low levels of IFN-γ produced by this subset may in part explain the persistent viral infections seen in these patients.
Besides pro-inflammatory cytokines, human CD56 + CD16 − NK cells have been shown to produce pro-angiogenic cytokines (e.g. VEGF-A). In murine NK cells, VEGF-A expression is repressed by STAT5b and deletion of total STAT5 changes the NK cell phenotype to protumorigenic. 45 We analyzed the expression of VEGFA transcripts in resting NK cells from STAT5b-deficient patient 1 and found a 6 fold-enhanced VEGFA levels (p.<0.006) compared to healthy donor (Fig 3, B). To confirm that STAT5b-regulated VEGF-A expression results in a pro-angiogenic phenotype, we challenged Stat5b +/− mice with the RMA lymphoma cell line. The use of mice heterozygous for Stat5b allowed us to retain sufficient numbers of NK cells (see Fig E4, D), which showed enhanced expression of VEGF-A compared to wild-type NK cells (see Fig E4, E). Indeed tumors, which grew in mice lacking one allele of Stat5b, showed an enhanced angiogenesis (Fig 3, C), as indicated by increased levels of VEGF-A in tumor homogenates (Fig 3, D). In addition, tumors displayed enhanced tumor weight (see Fig E4, F). These data suggest the absence of STAT5b might not only impair the CD56 bright NK cell functionality but also render them pro-angiogenic, which may lead to enhanced tumor growth.

The expression of adhesion and co-stimulatory receptors is dramatically affected in both NK cell subsets by STAT5b mutations.
Human NK cell subsets have different expression patterns of adhesion receptors that correlate with their trafficking pattern. 4,22 The interaction between NK cells and their target cells are mediated by a combination of activating receptors, as DNAM-1, NKG2D, and integrin heterodimer such as LFA-1, 19-21 resulting in a firm adhesion between NK cells and their target cells. 20 Flow cytometric analysis shows that STAT5b-deficient patients had decreased expression of CD62L (L-selectin) in the CD56 bright subset (Fig 4, A). This NK cell subset expresses elevated levels of CD62L, which mediates the interaction with the vascular endothelium. 46 The down-regulation of CD62L in CD56 bright NK cells suggests an abnormal migration through secondary lymphoid organs, which could affect the maturation and transition to the next stage of development. We also observed a significantly decreased expression of CD2, CD11a, CD11b, and CD18 in peripheral blood NK cell subsets (Fig 4,  A). Moreover, we observed lower expression of NKp44, NKp46, and DNAM-1 in both NK cell subsets (Fig 4, A) along with significantly low percentage of NK cells positive for DNAM-1, NKp44, and NKp46 compared with healthy donors (see Fig E5).

Lytic granule convergence to the MTOC is defective in human STAT5b-deficient NK cells.
The integrin LFA-1 (CD11a/CD18) is a key adhesion receptor that participates in the first steps of the formation of the immune synapse in NK cell and is required for optimal cell adhesion and granule convergence. [47][48][49][50] Using an antibody (clone m24) that binds selectively to the extended conformation of LFA1, we observed significantly decreased levels of activated LFA-1 in both NK cell subsets from STAT5b-deficient patients compared to those from healthy donors after PMA stimulation (Fig 4, B).
Deficiency of the CD18 subunit in leukocyte adhesion disorder type 1 (LAD-1) is characterized by impaired NK cell cytotoxicity with diminished conjugate formation and reduced convergence of lytic granules to the microtubule organizing center (MTOC). 47,51 When measured by flow cytometry, we found that the formation of conjugates by STAT5bdeficient patient cells was significantly delayed at 30 minutes but approached normal levels after 60 minutes (Fig 4, C). Based on this observation, we sought to find out if the crucial features of the immune synapse established by NK cells were also impaired in absence of STAT5b. Formation of the immune synapse is typically sanctioned by 3 majors events easily quantifiable using fluorescence microscopy: 52 (1) accumulation of filamentous actin at the interface with the target cell, (2) convergence of the lytic granules around the MTOC and (3), polarization of the granule-loaded MTOC towards the immune synapse.
To assess whether decreased activation of LFA-1 leads to impaired granule convergence in NK cells from patients with STAT5b deficiency, we evaluated the convergence of lytic granules using confocal microscopy images of NK cells conjugated to susceptible target cells. STAT5b-deficient NK cells were incubated with K562 target cells, and then the conjugates were fixed and stained for perforin to highlight lytic granules, and α-tubulin to localize the MTOC.
Filamentous actin was revealed using fluorescently conjugated phalloidin. Measurement of the positioning of lytic granules in STAT5b-deficient NK cells demonstrated that they failed to converge to the MTOC after conjugation with target cells (Fig 5, A and C). However, we observed normal F-actin accumulation and MTOC polarization at the immunological synapse (Fig 5, D). Additionally, normal levels of CD107a in NK cells from both patients indicates normal NK cell degranulation (Fig 5, E). Therefore, low levels of activated LFA-1 observed in STAT5b-deficient NK cells lead to decreased lytic granule convergence to the MTOC but do not affect polarisation of the MTOC to the actin enriched immunological synapse or the capacity for the NK cell to degranulate.

Lytic granule convergence is restored and NK cell cytotoxicity is partially restored after IL-2 stimulation
The canonical IL-2 signaling via STAT5b as the principal downstream effector is a potent NK cell activator and initiator of granule convergence in the first stages of target cell engagement. 15,16 STAT5b-deficient NK cells formed conjugates with K562 target cells after 30 minutes of stimulation with IL-2 (1000 U/mL) and we observed restored lytic granule convergence in response to IL-2 stimulation (Fig 5, B and C). These results confirm previous reports that canonical IL-2 signaling is not necessary for lytic granule convergence, 47 and that the non-canonical IL-2 signaling which promotes the activation of MAPK and PI3K pathways to enhance NK cell activation, 53, 54 could drive increased killing capacity observed after IL-2 stimulation.
We looked to see if another critical STAT5b dependent cytokine, IL-15, could also restore granule convergence. In support of the hypothesis that IL-2 signaled through an alternative pathway other than STAT5b to restore granule convergence, we show that lytic granule convergence was absent after IL-15 stimulation in NK cells from patient 1 because IL-15 signaling is most dependent on STAT5b (see Fig E6, A and B).
We tested the cytotoxic activity of NK cells in isolated PBMCs. In resting cells, the capacity of NK cells to kill target cells was reduced in both patients compared with a pool of three healthy donors (Fig 5, F). Interestingly, we observed partially restored cytotoxic capacity in presence of IL-2 in both patients (9-and 8-fold, respectively) (Fig 5, F). The cytolytic capacity of STAT5b-deficient patients was evaluated by calculating lytic units required to lyse 10% of K562 target cells. We observed that NK cells from STAT5b-deficient patients had reduced lytic capacity compared with those of the healthy donor after adjustment for the frequency of NK cells within the PBMCs population (Fig 5, F). In addition, we observed partially restored cytotoxic capacity with IL-15 (see Fig E6, C). The lack of lytic granule convergence suggests this cytotoxicity is a result of non-directed degranulation affecting bystander cells 34 as the NK cells do have the ability to degranulate with activation as evidenced by normal CD107a expression (Fig 5, E). We additionally measured ADCC using rituximab opsonized Raji target cells for recognition by CD16, and found that ADCC was impaired in both patients (Fig 5, G). The decreased ADCC NK cell cytotoxicity observed further supports a profound functional NK cell defect in immature NK cells which do not express terminal maturation markers like CD16.

Discussion
Autosomal recessive STAT5b deficiency is characterized by severe growth failure, recurrent infections, accompanied by significantly decreased levels of circulating regulatory T cells and NK cells. 17,23 It is important to distinguish from the somatic mutations in STAT5B associated with lymphocytic leukemia 55 as well as the recently characterized novel variant N642H gain-of-function mutation shown to be associated with eosinophilia, urticaria, dermatitis, and diarrhea. 55,56 The homozygous mutation identified in our patients was a deletion of a single G at the junction of the exon13-intron13 36 which affects the expression of STAT5b in PBMCs and BLCL but not STAT5a. The STAT5b-deficient siblings included in this study had significantly reduced NK cells in the periphery. This likely occurs since STAT5 is the principal downstream effector for the IL-15 signaling pathway that induces development, activation, and proliferation of NK cells. 11,15 The low numbers of NK cells leave these patients susceptible to recurrent viral infections, including rhinovirus, adenovirus, and EBV viremia. The absence of STAT5b also alters the expression of FOXP3 and CD25 17 accounting for the low circulating population of regulatory T cells. The impaired homeostasis of Tregs results in autoimmunity and inflammatory disease observed in these patients including chronic lymphocytic lung disease, autoimmune hypothyroidism, juvenile idiopathic arthritis (JIA), and chronic anemia. In addition, the interaction between GH circulating with their receptor leads to STAT5b activation. The absence of STAT5b also explains the severe growth failure observed in these patients. 15,23 During our study we analyzed NK cell numbers in peripheral blood and by using comprehensive flow cytometric analysis, we evaluated the effects of STAT5b deficiency on NK cell development and phenotype. We characterized a CD56 dim NK cell subset with decreased levels of perforin which correlates with results from Stat5b −/− mice. 18 The expression of CD16 in the CD56 dim subset was also significantly decreased, thereby leading to impaired ADCC. In addition, this NK cell subset had high levels of markers associated with immature NK cells, such as CD94, CD117, and low expression of CD57 and CD8 receptors normally found on mature cytotoxic cells. These results suggest an abnormal developmental transition between CD56 bright to CD56 dim leading to the low cytotoxicity capacity of these cells.
Mice with conditional ablation of total STAT5 in NK cells display many similar characteristics seen in the described human patients. The Stat5 fl/fl Ncr1iCre mice exhibit impaired NK cell development in bone marrow leading to diminished numbers of NK cells in periphery. 57  and Gata-3 which is associated with impaired development, homeostasis, and function of innate lymphoid cells (ILCs). 58 Interestingly, the impact was higher with deletion of Stat5b. 58 They also describe depressed cytokine production similar to our results in the human CD56 bright cells. CD56 bright cells provide cytokines promoting control of viral infections. 22,42,46 Our patient's CD56 bright NK cell subset was characterized by significantly reduced production of cytokines (IFN-γ, TNF-α, IL-13, and GM-CSF) and an abnormal expression of CD62L (L-selectin). CD62L is a part of the NK cell homing mechanism. It has been shown to play a role in cell migration as well. 46 We observe a down-regulation of CD62L which suggests there may be a possible effect on NK cell migration. Along with the defective cytokine production including low levels of IFN-γ, these immature poorly differentiated NK cells heightened the susceptibility to severe viral infections observed in these patients. Furthermore, the patient's NK cells were characterized by enhanced expression of the pro-angiogenic factor VEGF-A, which may indicate that these poorly cytotoxic NK cells display a pro-tumorigenic phenotype. 45 Therefore, tumor surveillance by NK cells might be impaired in STAT5b-deficient patients by more than one mechanism.
The first interaction between NK cells and their target cells is mediated by activating and adhesion receptors that induce commitment to cytotoxicity. 19,59 Our findings show that NK cells from STAT5b-deficient patients had significantly decreased expression of these important receptors: CD2, DNAM-1, NKG2D, NKp46, and CD16. Synergy between NKp46 with DNAM-1, NKG2D or CD2 induces degranulation. 60 In NK cells, NKG2D signaling contributes to adhesion, granule convergence, and degranulation through PI3K and CrkL activation. 61 Our results suggest a defect in the first steps of NK cell activation leads to impaired cytotoxicity in STAT5b-deficient patients.
The engagement of LFA-1 stimulates convergence of lytic granules to the MTOC 47 and MTOC polarization to the immune synapse. 62 IL-2 also induces lytic granule convergence independently of JAK3 activation through Src-dependent protein signaling 47 , suggesting that the canonical IL-2 pathway is not necessary for lytic granule convergence. 47 Low levels of activated LFA-1 and decreased lytic granule convergence to the MTOC observed in NK cells from STAT5b-deficient patients, along with low surface expression of CD18 and CD11a, suggests that the absence of STAT5b impairs the early steps of NK cell activation and cytotoxicity. Interestingly, we observed a partial rescue of lytic capacity after IL-2 stimulation. This, combined with our data and others showing that non-canonical IL-2 signaling pathway promotes lytic granule convergence, independently of STAT5b, suggests that restoration of granule convergence by IL-2 is responsible for the improved lytic function in STAT5b-deficient NK cells. Alternatively, STAT5a, which is not affected in our patients, may overcome the STAT5b deficiency under high concentration of IL-2.
In addition, this is supported by our results using IL-15 stimulation. We showed partial restoration of cytolytic capacity but not lytic granule convergence with IL-15 stimulation. Hsu et al. showed NK cells that converge their lytic granules prior to degranulation have more precise and directed killing with less "collateral damage". 34 NK cell cytotoxicity in the absence of granule convergence occurs through degranulation in a non-directed manner affecting bystander cells 34 . Our data demonstrated that IL-2 induces convergence of the lytic granules around the MTOC and polarization of the granule-loaded MTOC towards the immune synapse in our patients' NK cells. This is important for directed killing and partial restoration of function seen in our patients. Further, it suggests there is a non-canonical IL-2 signaling pathway in play to restore appropriate convergence of lytic granules for directed killing. However, our data also showed that IL-15 does not induce lytic granule convergence in absence of STAT5b. As disussed, degranulation in a non-directed manner without proper convergence of lytic granules to the MTOC can occur resulting in bystander killing. The concern for bystander killing is morbidity and destruction of surrounding tissues.
Additional experiments were performed to create a model human NK cell lines depleted of STAT5b. However, any efficient reduction of endogenous STAT5b expression led to premature cell death (data not shown), emphasizing the critical role for this regulator of NK cell growth and maturation. In summary, we have defined the NK cell deficiency that occurs as a result of a deleterious STAT5b mutation. This alteration significantly impairs NK cell function and decreases the number of NK cells circulating in peripheral blood, therefore mimicking the abrogated NK cell terminal maturation observed in Stat5b −/− mice. This work demonstrates that STAT5b plays a critical role in NK cell development, maturation, and activation.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material. A, Western blot analysis of STAT5b on total extracts from peripheral blood mononuclear cells (PBMCs) from healthy donors and STAT5b-deficient patients. Polyclonal STAT5a and STAT5b antibodies were used. B, NK cells were identified as CD56 bright CD3 − and CD56 dim CD3 − NK cells gated on lymphocytes. C, Frequency of total NK cells in peripheral blood. D, CD56 dim NK cells from healthy donors and STAT5b-deficient patients were studied by flow cytometry. The Median fluorescence intensity (MFI) data from age-and sexmatched healthy donor are included (light blue circles) 30 in addition to data from unmatched controls acquired at the time of the experiment (dark blue circles). Horizontal bars represent the median, and the vertical bars indicate the standard deviation. Statistical significance was analyzed using the unpaired Student's t-test. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. HD, healthy donors (dark blue circle); previously reported healthy female age and gender matched controls (light blue circle); Pt, STAT5b-deficient patients, Pt 1 (red box); Pt 2 (green box).  A, Median fluorescence intensity (MFI) for each cytokine and receptor in CD56 bright NK cell after stimulation with PMA and ionomycin. Median fluorescence intensity (MFI) data from age-and sex-matched healthy donor are included (light blue circles) 30 in addition to data from unmatched controls acquired at the time of the experiment (dark blue circles).Horizontal bars represent the median, and the vertical bars indicate the standard deviation. Statistical significance was analyzed using the unpaired Student's t-stest. *p<0.05, **p<0.01, ****p<0.0001. HD, healthy donors (dark blue circle); previously reported healthy female age and gender matched controls (light blue circle); 30 Pt, STAT5b-deficient patients, Pt 1 (red box); Pt 2 (green box). B, Expression of VEGF-A gene validated by qPCR of total