SUDHL4 (ATCC, CRL-2957), JeKo-1 (ATCC, CRL-3006) and RAMOS (ATCC, CRL-1596) cells were cultured in R10 medium (RPMI 1640 [Gibco] with 1% penicillin-streptomycin [Life Technologies] and 10% heat inactivated fetal bovine serum [FBS; Sigma]). Cells were treated with 50, 500 or 2000nM selinexor (KPT-330, provided by Karyopharm Therapeutics), 50nM leptomycin B (Sigma) or DMSO control (for the indicated 0nM negative controls) for 16 hours at 37°C before use. To prevent drug-induced apoptosis, cell lines were incubated with Q-VD-OPh (QVD) (10-20µM) (Sigma) for 30 minutes prior to addition of XPO1 inhibitors or DMSO control.

Healthy donor peripheral blood mononuclear cells (PBMC) were obtained with full ethical approval from the National Research Ethics Committee (reference 06/Q1701/120). PBMC from patients with chronic lymphocytic leukemia (CLL) were collected from patients attending clinic at Southampton General Hospital. All patients provided written informed consent and the study was approved by the Institutional Review Boards at the University of Southampton (REC: H228/02/t). PBMCs were cryopreserved and stored in liquid nitrogen. CD56+CD3- NK cells were isolated from cryopreserved healthy PBMCs using the Miltenyi human NK cell isolation kit and cultured in R10 medium at a density of 1.5x10⁶ cells/mL and incubated with 1 ng/mL IL-15 (R&D Systems) overnight before use in functional assays.

B cell lymphoma cell lines were stained with Cell Trace™ Violet Cell Proliferation Kit (Invitrogen™) following the manufacturer’s instructions then incubated with QVD and selinexor or DMSO control for 16 hours. Isolated NK cells were then co-cultured with lymphoma cells at an effector: target (E:T) ratio of 5:1 for 4 hours at 37°C. After co-culture, cells were stained with 1.6 µg/mL propidium iodide (Invitrogen™) and NK cell specific lysis of Violet stained target cells assessed by flow cytometry. Cells were acquired on a BD FACS Aria II (BD Biosciences) machine using FACSDiva software (BD Biosciences) and analysed with FlowJo v10.7.1 (BD Biosciences). Lysis was defined as uptake of propidium iodide by the target cells.

PBMCs were incubated with 1 ng/mL IL-15 overnight at a cell density of 2x10⁶ cells/mL and then co-cultured with selinexor- or leptomycin B-treated lymphoma cell lines at an E:T ratio of 5:1 for 4 hours at 37°C. Immediately before co-culture, 0.17 µg/mL α-CD107a (LAMP)-eFluor660 (clone eBioH4A3, Invitrogen) was added to PBMCs. After 1-hour of co-culture, GolgiStop (per manufacturer recommendations, BD Biosciences) was added. Following 4-hours of incubation, cells were incubated with 10% human serum at 4°C for 15 minutes before surface staining with antibodies against CD3-PerCP (clone UCHT1, Biolegend), CD56-PE/Cy7 (HCD56, Biolegend) and NKG2A-FITC (REA110, Miltenyi Biotech) in FACS buffer (PBS, BSA 1%, Sodium Azide 0.05%) at 4°C for 30 minutes. Cells were then permeabilized and fixed with BD Cytofix/Cytoperm (BD Biosciences) per manufacturer recommendations and stained with anti-IFNγ-BV421 (BD Biosciences) at 4°C for 30 minutes. Cells were then washed twice with 1X Perm/Wash buffer and immediately assessed by flow cytometry using a BD FACS Aria II (BD Biosciences) and FACSDiva software (BD Biosciences) as above.

Ramos, SU-DHL-4 and JeKo-1 cells were incubated with selinexor (50-2000nM), leptomycin B (50nM) or DMSO control for 16 hours in the presence of QVD. Cells were then surface stained with antibodies against activating NK cell ligands Vimentin-A488 (clone 280618, R&D Systems), ULBP-1-PE (170818, R&D Systems), ULBP-2/5/6-PerCP (165903, R&D Systems), CD54-PB (HCD54, Biolegend), B7H6-APC (875001, R&D Systems) and MICA/B-PE/Cy7 (6D4, Biolegend); the inhibitory NK cell ligand HLA-E-PE/Cy7 (3D12, Biolegend) and pan-HLA class-I molecules (W6/32, Biolegend) for 30 min at 4°C. CLL cells were incubated with selinexor (50-2000nM) or DMSO control for 40 hours in the presence of QVD then incubated with 10% human serum at 4°C for 15 minutes before CD5+CD19+ CLL cells were surface stained with anti-HLA-E-PE/Cy7 (3D12, Biolegend) for 30 min at 4°C. Normal PBMC were incubated with selinexor (500-2000nM) or DMSO control for 16 hours then surface stained with anti-HLA-E-APC (3D12, Biolegend), anti-CD3-PerCP (UCHT1, Biolegend), anti-CD19-PE (HIB19, Biolegend) and anti-CD56-PE/Cy7 (HCD56, Biolegend) for 30 min at 4°C. Cells were then acquired on a BD FACS Aria II (BD Biosciences) using FACSDiva software (BD Biosciences) as above.

Ramos, SU-DHL-4 and JeKo-1 cells were incubated with selinexor (50-2000nM) or leptomycin B (50nM) for 16 hours in the presence of QVD then lysed in NP40 Cell Lysis Buffer (Fisher Scientific UK) supplemented with PMSF (Sigma, 174 µg/mL) and protease inhibitor (Sigma, 1:100 final dilution). Proteins were separated on 10% polyacrylamide gels (Thermo Fisher Scientific), transferred to nitrocellulose membranes (Amersham) and blocked in 5% BSA (Sigma) before being probed with antibodies against XPO1 (D6V7N, Cell Signalling Technology), PARP (4C10-5, BD Pharmingen), HLA-E (Sigma), p53 (1C12, Cell Signalling Technology) or β-actin (8H10D10, Cell Signalling Technology). Protein bands were detected following incubation with HRP-linked secondary antibodies (Dako) and chemiluminescence reagents (Thermo Scientific) and were visualized using the ChemiDoc-It imaging system (UVP). Primary and secondary antibodies were used at concentrations recommended by the manufacturer.

SU-DHL-4 cells were incubated with selinexor (50-2000nM) for 16 hours in the presence of QVD then incubated with the anti-CD20 antibodies rituximab or obinutuzumab or isotype control (1 µg/mL) for a further 20 minutes at 37°C prior to co-culture with PBMC (degranulation assay) or isolated NK cells (cytotoxicity assay) as described above.

Statistical analyses were performed using GraphPad Prism V.9.0 software. Paired t-test was used to compare differences between the means of two groups and paired or unpaired one-way ANOVA followed by Dunnett’s post-hoc test analysis was used to compare differences in means between multiple groups. Significance values are defined as: *P < 0.05, **P < 0.01, ***P < 0.005 and ****P < 0.001.

To assess whether selinexor modulated NK cell activation against malignant B cells, we incubated selinexor (50-2000 nM) for 16 hours with SUDHL4 (DLBCL), JeKo-1 (mantle cell lymphoma) or RAMOS (Burkitt lymphoma) cells and then assessed cytotoxicity following co-culture with NK cells isolated from healthy human donors. Selinexor increased NK cell cytotoxicity in a concentration-dependent manner against SUDHL4 (p<0.005), JeKo-1 (p<0.005) and RAMOS (p<0.01) cells ( Figures 1A, B ). This lysis was specific to NK cell activity because in these experiments apoptosis induced by selinexor was prevented by addition of the caspase inhibitor QVD, as shown by no increase in propidium iodide staining ( Figures 1A, B ) in the absence of NK cells. These results indicate that blockade of intrinsic apoptosis in lymphoma cells, and hence resistance to direct selinexor toxicity, does not prevent NK mediated lysis stimulated by selinexor. This is in agreement with previous reports of caspase independent cell death induced by effector lymphocytes and purified granzyme B (29, 30). Selinexor activity was confirmed by clear concentration-dependent XPO1 degradation ( Figure 1C ), as previously reported (31). In accordance with our observations that selinexor enhanced NK cell activity against lymphoma cells, we measured degranulation (CD107a) of CD56^dim and CD56^bright NK cell subsets when co-cultured with these target cells. This revealed that selinexor significantly increased the degranulation of the more mature and cytotoxic CD56^dim NK cell population against all cell lines tested compared to untreated controls ( Figures 1D, E ). Degranulation of the more regulatory and proliferative CD56^bright subgroup of NK cells was significantly enhanced by selinexor against SUDHL4 cells ( Figure 1E ). Together these data demonstrate that the XPO1 inhibitor selinexor increases NK cell cytotoxicity against B lymphoma cell lines.

NK cell activity is tightly controlled by a plethora of activating and inhibitory receptors which recognize and engage ligands expressed on the surface membrane of infected, transformed or stressed target cells (7, 10). Therefore, to investigate the mechanism for enhanced NK cell activation and killing following selinexor incubation with lymphoma cells, we screened SUDHL4, JeKo-1 and RAMOS cells by flow cytometry for changes in a panel of ligands for activating and inhibitory NK cell receptors. Following selinexor incubation for 16 hours (50-2000nM), no significant change was evident in expression of the activating ligands Vimentin and ULBP-2/5/6, a trend to down-regulation for ULBP-1 and a significant reduction in expression of MICA/B, B7-H6 and ICAM1 ( Figure 2A ). Signals from these activating receptors do not account for the increased NK cell activation noted following selinexor pre-treatment of lymphoma cells ( Figure 1 ). Therefore we assessed expression of ligands for inhibitory NK cell receptors. The surface expression of HLA-E as determined by the HLA-E specific antibody 3D12 was downregulated by selinexor on SUDHL4 (57% reduction, p<0.001), JeKo-1 (19% reduction, p<0.01) and RAMOS (63% reduction, p<0.001) cells in a concentration-dependent manner ( Figures 2B, C ). This was not due to a decrease in cell size as there was no change in forward scatter following selinexor treatment ( Figures 2B, C ). In addition, selinexor caused a much lower, but statistically significant, downregulation of total HLA molecules as measured by the antibody clone W6/32 ( Figures 2B, C ). This decrease likely corresponds to the downregulation of HLA-E as the W6/32 antibody clone recognizes the HLA proteins HLA-A, -B, -C in addition to HLA-E (32, 33). Consistent with this model, HLA-E represents a relatively small fraction of total HLA on the surface of cell. We then assessed whether selinexor modulated the expression of HLA-E on primary tumor cells using samples derived from patients with CLL. Selinexor reduced surface HLA-E expression on CD5+CD19+ CLL cells in all four patient samples tested, with a mean reduction of 49% at 2000nM after 40 hours in vitro incubation ( Figure 2D , p<0.005). We then addressed whether surface HLA-E downregulation by selinexor was specific to malignant B cells or whether selinexor also modulates expression of HLA-E on lymphocytes from healthy donors. Incubation of selinexor (500-2000nM) for 16 hours with healthy donor PBMC caused a significant downregulation of surface HLA-E expression on normal B cells compared to both normal T cells (p<0.0001) and NK cells p<0.0001) ( Figure 2E ). This data indicates that XPO1 inhibition induces loss of surface HLA-E expression on malignant B cells as well as on normal B cells relative to other lymphocyte populations. Selinexor did not reduce total protein levels of HLA-E in SUDHL4 lymphoma cells ( Figures 2F, G ) and therefore XPO1 mediated downregulation of surface HLA-E was not due to targeted inhibition of HLA-E protein production, but more likely was a result of a reduction in supply of other HLA-E binding substrates that lead to HLA-E upregulation. Indeed, surface expression of HLA-E compared to other HLA molecules is highly sensitive to blockade of newly synthesized protein transport (34, 35) and selinexor potently inhibits protein translation (36).

Taken together these data show that selinexor downregulates HLA-E surface expression on lymphoma cell lines (SUDHL4, JeKo-1 and RAMOS) and primary CLL cells. HLA-E is the ligand for the inhibitory receptor NKG2A and this suggests a potential mechanism by which it may augment NK cell activity (37, 38). We thus hypothesized that selinexor would selectively activate NKG2A+ NK cells against lymphoma cells. Indeed, selinexor induced activation of NKG2A+ and not NKG2A- NK cells against SUDHL4 cells as measured by CD107a expression and IFNγ production ( Figure 3 ). As previously described (39), NKG2A+ NK cells compared to NKG2A- NK cells showed enhanced CD107a and IFNγ expression against target cells in the absence of selinexor, however this activation was further enhanced in NKG2A+ cells by the addition of selinexor ( Figure 3 ). We noted substantial baseline variability between donors in CD107a and IFNγ expression in the absence of selinexor as previously described (15), however selinexor treatment increased CD107a in all donors tested, and overall IFNγ in three of the six donors tested ( Figure 3B ). These data confirm that selinexor increases NK cell activation against lymphoma cells through downregulation of surface HLA-E, reducing the engagement of this inhibitory receptor and promoting activation of NKG2A+ NK cells.

To confirm XPO1 as the target for selinexor induced HLA-E downregulation we utilized an alternative XPO1 inhibitor, leptomycin B. This is a metabolite from Streptomyces which potently inhibits XPO1 function but does not induce XPO1 degradation (40, 41). SUDHL4 cells incubated with leptomycin B (50nM) showed a significant reduction in surface HLA-E expression, mirroring that mediated by selinexor ( Figure 4A ). Decreased HLA-E expression was not caused by changes in cell size as revealed by measurement of forward scatter ( Figure 4B ). In accordance with this, SUDHL4 cells incubated with leptomycin B significantly increased activation of NKG2A+ but not NKG2A- NK cells as measured by CD107a expression ( Figures 4C, D ). We then confirmed using immunoblotting that leptomycin B did not reduce XPO1 protein levels, in contrast to selinexor ( Figure 4E ) as previously reported (40). Leptomycin B did however increase p53 expression, consistent with previous reports (42). In addition, blockade of leptomycin B-induced apoptosis by QVD was confirmed by the absence of cleaved PARP (cPARP) ( Figure 4E ). This data demonstrates that inhibition of XPO1, in the absence of its degradation, is sufficient for HLA-E downregulation on tumor cells and resultant NKG2A+ NK cell activation.

Through their expression of the Fc gamma receptor CD16 (FcgRIIIA), NK cells can elicit anti-tumor functions during anti-CD20 monoclonal antibody (mAbs) treatments in lymphoma (43). As selinexor is currently being clinically evaluated in combination with anti-CD20 antibodies for patients with advanced B cell non-Hodgkin lymphoma (NCT03147885), we addressed whether selinexor could potentiate activity of the anti-CD20 mAbs rituximab and obinutuzumab. SUDHL4 cells were incubated with selinexor (50-2000nM) for 16 hours then cultured with rituximab, obinutuzumab or isotype control for 20 minutes prior to the addition of healthy donor PBMC, containing NK cells. NK cell degranulation in both the NKG2A+ and NKG2A- NK cell populations was increased by the addition of rituximab or obinutuzumab ( Figure 5A ) and in accordance with our previous data, this activation was further increased by selinexor in the NKG2A+ but not NKG2A- NK cell populations ( Figure 5A ). This data indicates that selinexor enhances NK cell activation in the presence of two separate clinically relevant anti-CD20 antibodies. To confirm that this led to increased lysis of lymphoma cells, we performed cytotoxicity assays with isolated NK cells. NK cell specific cytotoxicity against SUDHL4 cells was significantly increased by selinexor (500nM) in the presence of isotype control, rituximab and obinutuzumab ( Figure 5B ). In summary, selinexor enhances ADCC against lymphoma cells in combination with anti-CD20 antibodies.