Selinexor suppresses Downstream Effectors of B-cell activation, proliferation and migration in chronic lymphocytic leukemia cells

Letter to the Editor

Chronic lymphocytic leukemia (CLL) is characterized by the progressive accumulation of clonal mature B-cells in the blood, bone marrow, and secondary lymphoid organs. While peripheral CLL is quiescent, up to 1% of clonal cells are generated daily within proliferative centers¹. CLL cells isolated from these structures display activation of survival and proliferation pathways ((for example, phosphatidylinositol 3-Kinase (PI3K), Bruton's tyrosine kinase (BTK) and nuclear factor (NF-kB)¹ due to interactions with accessory cells. In CLL, B-Cell Receptor (BCR) signaling represents one of the central pathways known to enhance survival and proliferation. The importance of BCR signaling in CLL pathogenesis is underscored by the prognostic significance of somatic mutations in the immunoglobulin heavy-chain variable regions (IGHV) of the BCR. In general, CLL patients with unmutated IGHV have shorter progression-free survival and lower response rates². CLL cells isolated from patients with unmutated IGHV rely heavily on survival signals and respond preferentially to BCR and TLR9 stimulation³, suggesting that therapies which block these signals may be particularly effective in thisaggressive subset of CLL.

The recently unveiled mechanisms which control CLL cell survival and expansion have prompted the rapid development of therapeutics which disrupt CLL-microenvironment interactions and block BCR-driven activation (i.e fostamatinib, idelalisib, ibrutinib)^4,5-7 and have demonstrated profound clinical activity^{4, 5}. Unfortunately, some patients do not respond or develop resistance, emphasizing the importance of alternative therapeutic strategies⁵.

Exportin 1 (XPO1/CRM1) is a nuclear export protein overexpressed in CLL⁸. Nuclear export is emerging as an exciting target as increasing evidence is being generated that nuclear-cytoplasmic shuttling proteins have a direct role in the pathophysiology of various hematologic malignancies^{9, 10}. We were the first to demonstrate that inhibition of XPO1 by selective inhibitors of nuclear export (SINEs) killed CLL cells in vitro and increased survival in a CLL mouse model⁸. Although nuclear export inhibitors are believed to mediate their effect by forcing mainly nuclear retention and activation of tumor suppressor proteins (i.e p53, FoxO3a, IκB), recent reports also indicate a possible role of SINEs in induction of autophagy and inhibition of ribosomal biogenesis and translational flux^{11, 12}. Selinexor is a new clinically viable SINE that is currently in phase I clinical trials for the treatment of both liquid (NCT01607892) and solid tumors (NCT01607905 and NCT01896505). Preliminary data from a cohort of 18 heavily pretreated/refractory NHL and CLL patients with progressive disease on study entry indicates that Selinexor is well tolerated with favorable pharmacokinetic, pharmacodynamics and antitumor properties inducing tumor shrinkage or disease stabilization in 80% of the patients including one ibrutinib/refractrory CLL patient with Richter's transformation¹³. To improve our understanding of Selinexor in the setting of CLL therapy, we evaluated survival and tissue homing circuits in-vitro and in-vivo using the validated mouse models of CLL.

Selinexor maintained robust in-vitro cytotoxicity in primary CLL cells comparable to its pre-clinical predecessor KPT-251 (Figure 1A) even in stromal or monocyte-derived nurse-like cells (NLCs) cu-culture conditions (Figure 1B and C) with modest cytotoxicity against normal B cells (Figure 1D). Similar to KPT-251, Selinexor exhibited enhanced killing of unmutated IGHV CLL cells (Figure 1E), suggesting that Selinexor may be especially active against a traditionally drug-resistant and highly aggressive subset of CLL.

Figure 1. Selinexor induces selective cytotoxicity in CLL cells.

(A) Selinexor and KPT-251 induce comparable level of cytotoxicity of CLL cells at 72 hr time point as measured by MTS assay (n=8 each). (B) CD19+ cells from CLL patients were isolated from peripheral blood and incubated with or without Selinexor (2.5μM) for 12 hr. Drug was then washed out and cells were incubated in suspension or on an HS5 cell layer for additional 48 hr. Viability was determined by annexin-V/PI flow cytometry, and is shown relative to time-matched DMSO controls for each group. Red filled circles represent averages. (C) CLL cells were co-cultured with NLC in medium (control) or medium containing DMSO or 2.5 μM Selinexor. The bar diagram represents the mean relative viabilities of CLL cells co-cultured with NLC (control) compared with CLL cells co-cultured with NLC and Selinexor or vehicle control. Viabilities of treated samples were normalized to the viabilities of control samples (means ± s.d., n=7). CLL cell survival in the presence of NLC was significantly inhibited by Selinexor (P<0.0001). (D) Selinexor is not cytotoxic to normal B cells as measured by annexinV/PI flow staining (n=8 each). (E) IGHV mutational status and del(17p) status was examined for differences in response to 0.5 μM Selinexor. Viability was measured by Annexin/PI flow staining at 72h time point. Horizontal bars represent averages; p<0.01. (F) CD19+ cells from CLL patients (n=8) were incubated with or without 0.5μM Selinexor and 3.2μM CpG685. Proliferation was assessed by tritiated thymidine incorporation 5 days later. IGHV mutational status was examined for differences in response to CpG. Selinexor prevents CpG induced proliferation of CLL cells (p<0.001). (G) CLL cells were stimulated with 3.2μM CpG685 in presence or absence of 0.5 μM Selinexor (SEL) or vehicle control (C) for 1h or 24h. Activation of AKT, ERK, and cMyc was analyzed by immunoblot. A representative experiment is shown; n=5. (H) CLL cells were stimulated with 3.2μM CpG685 in presence or absence of 0.5 μM Selinexor (S) or vehicle control (C) for 48h. Expression of cyclin A2 was analyzed by immunoblot. A representative experiment is shown; n=5. (I) CLL cells were stimulated with immobilized anti-IgM, or isotype in presence or absence of 0.5 μM Selinexor or 1 μM ibrutinib for 24h. Cytotoxicity was measured by annexin/PI. (J) CLL cells were stimulated with immobilized anti-IgM or control in presence or absence of Selinexor (SEL), vehicle control (C) or 1 μM ibrutinib (Ibr) for 15’ or 4h. Activation of AKT and ERK was analyzed by immunoblot. A representative experiment is shown; n=5. (K) CLL cells were treated with or without 0.5 μM Selinexor (SEL) for 1 h, prior to stimulation with 100ng/ml CXCL12 for 30 minutes or 4 h. Activation of ERK was analyzed by immunoblot. A representative experiment is shown; n=3. (L) CLL cells were pre-incubated in media containing DMSO or 0.5 μM Selinexor (SEL) and then allowed to migrate toward 200 ng/ml of CXCL12, or 1 μg/ml of CXCL13, or medium alone. Bars represent the percentage of migration determined as the number of migrated CLL cells in the lower chamber compared to the input. Selinexor significantly decreased CLL cell chemotaxis toward CXCL12 and CXCL13 (p<0.05, n=7).

CLL cells are categorically unresponsive to in-vitro stimuli and it is notoriously difficult to assay proliferation ex-vivo. However, stimulation via Toll-like receptor 9 using synthetic CpG-oligodeoxynucleotides, which mimic bacterial DNA motifs, can occasionally induce limited ex-vivo proliferation³. This proliferative response is clone-dependent and has been correlated with IGHV-unmutated disease for which the magnitude of ERK and AKT signaling is amplified leading to upregulation of cyclin A2. To determine if XPO1 inhibition could abate CpG-induced ex-vivo proliferation in unmutated IGHV disease, we treated CpG-stimulated primary CLL cells with Selinexor or vehicle control for 5 days and measured proliferation by ³H thymidine incorporation. Selinexor significantly blocked CpG-induced proliferation in the IGHV-unmutated CLL cells (Figure 1F). As previously reported, IGHV mutated CLL cells did not proliferate in response to CpG (data not shown)³. To delve into the complex molecular mechanism by which XPO1 blockade prevents CpG-induced CLL proliferation, we investigated the effect of Selinexor on TLR9 mediated ERK and AKT survival pathways. CLL cells were stimulated with CpG for 1 hour and then treated with Selinexor or vehicle control for additional 1 h or 24h. Proliferation was confirmed by ³H thymidine incorporation after 48 hours of treatment (data not shown). Our data reveal XPO1 inhibition prevents CpG-induced phosphorylation of ERK and AKT, thereby preventing expression of cMyc and cyclin A2 and downstream proliferation of IGHV unmutated CLL cells (Figure 1F-H).

Ex-vivo BCR crosslinking with immobilized anti-IgM elicits prolonged AKT and ERK phosphorylation, activates CLL cells, and protects them from undergoing spontaneous apoptosis³. We previously showed that KPT-185 enforces nuclear retention of IκB, leading to inhibition of NF-κB in CLL cells⁸. Among its many functions, NF-κB has been shown to up-regulate BTK¹⁴. Interestingly, SINEs induce BTK downmodulation in CLL cells (Supplementary Figure 1A). Since BTK is essential to B lymphocyte development, survival, and signaling we sought to determine the ability of Selinexor to block downstream BCR signaling. After crosslinking the BCR on CLL cells we analyzed phosphorylation of spleen tyrosine kinase (SYK), AKT, and ERK by immunoblot and discovered that Selinexor abrogates phosphorylation of AKT and ERK, but not SYK (Figure 1I and J), possibly through the targeted inhibition of the AKT or ERK pathways in CLL (e.g. down-modulation of BTK), as opposed to non-specific down-modulation of multiple kinase-mediated pathways.

CXCL12 is a powerful microenvironment recall chemokine that plays key physiologic roles in controlling mature B and CLL cell trafficking to germinal centers¹⁵. In CLL, signaling via the CXCL12/CXCR4 axis drives prolonged activation of the ERK-MAPK pathway, reduces spontaneous and drug-induced apoptosis, and facilitates the migration of CLL cells beneath stromal cells ex-vivo^{6, 15}. Our transwell migration assay demonstrates that Selinexor significantly inhibits CXCL12-mediated signaling preventing recall migration of CLL cells (Figure 1K and L). Notably, this was not due to downregulation of surface CXCR4 (data not shown).

To validate Selinexor therapeutic efficacy in-vivo we utilized the Eμ-TCL1-SCID transplant model of CLL⁸. Selinexor or vehicle treatment was initiated after leukemia establishment at 2 weeks post-engraftment and we examined two different dosing schedules: a) BIW (Twice per week, Monday and Tuesday) or b) QoDX3 (Three times per week, Monday, Wednesday and Friday). Mice were treated BIW with 5mg/kg, or 15mg/kg or QoDX3 with 3mg/kg, 10mg/kg or 15mg/kg of Selinexor for up to 36 weeks with overall survival (OS) as the primary endpoint. Surviving animals were followed up for an additional 4 weeks and we observed no indication of leukemic relapse by analysis of the peripheral blood lymphocyte count (data not shown).

Selinexor at doses of >10 mg/kg BIW and QoDX3 significantly prevented the rapid increase of peripheral leukemic cells associated with progressive disease in this model (Supplementary Figure 1B and 1C). Notably, we often do not witness the peripheral lymphocytosis observed with conventional BCR-targeted therapies in clinical trials using the Eμ-TCL1 model conceivably because the Eμ-TCL1 model is driven by an aggressive oncogene and not the endogenous BCR.

Within the BIW group, Selinexor at 15 mg/kg showed a significant improvement over vehicle with an impressive 100% survival rate (Figure 2A). Within the QoDX3 group, Selinexor at 10 and 15 mg/kg QoDX3 both showed a significant improvement over vehicle (Figure 2B). Selinexor was also effective in mice with leukemic phase (i.e., very high tumor burdens) where treatment was initiated 10 weeks post engraftment (Figure 2C and supplementary Figure 1D). Volume of spleen is an important metric of disease progression and response to therapy in the Eμ-TCL1 mouse model of CLL. Selinexor significantly prevented spleen enlargements at doses of >10 mg/kg BIW and QoDX3 (Figure 2D and 2E) and in mice with leukemic phase where treatment was initiated 10 weeks post engraftment (Figure 2F). Overall, Selinexor was also well tolerated and drug-induced weight loss seen at the beginning of the study (Supplementary Fig 1E and 1F) was fully reversed by the end of the study and did not adversely affect the animals.

Figure 2. Selinexor prolongs survival in a mouse model of CLL.

(A) Overall survival (OS) curves for TCL1-SCID mice (n=9 each group) treated orally with Selinexor 5 mg/kg, Selinexor 15 mg/kg (median OS = undefined p=0.0002) or vehicle control (median OS = 98 days) twice per week (BIW; Monday and Tuesday). Treatment was initiated 14 days post engraftment; (B) OS curves for TCL1-SCID mice treated orally with: Selinexor 15 mg/kg (median OS = 213 days; p<0.0001), Selinexor 3 mg/kg (median OS = 95 days), Selinexor 10 mg/kg (median OS = undefined, p<0.0001), and vehicle control (median OS = 98.5 days) three times per week (QODX3; Monday, Wednesday and Friday). n=9 each group. (C) OS curve for TCL1-SCID mice treated 70 days post engraftment with 15 mg/kg (median OS = undefined; p=0.0008). Selinexor or vehicle control (median OS = 99 days) QODX3; n=9 each group. (D-E-F) Caliper-derived spleen volumes of TCL1-SCID mice treated with Selinexor BIW (D), early QoD3X (E), or Late QoD3X (F). Spleen volumes were measured at time of death using a manual caliper (with 1mm error). Splenic volume was then calculated using the standard clinical ellipsoid equation of length × width × thickness × 0.523. (G-H) In vivo EdU labeling was performed in 6 Eμ-TCL1 transgenic animals treated for 2 days with vehicle control or Selinexor 15 mg/kg (n=3 for each group). EdU was injected on day 3 and spleen and Bone Marrow samples were collected 24 hours later. Spleen (G) and Bone Marrow (H) were analyzed by flow cytometry for percentage of Edu positive cells within the leukemic population (CD45+/CD19+/CD5+ cells). (I) Ki67 staining of paraffin-embedded spleen sections from Eμ-TCL1 transgenic mice treated for 2 days either with Selinexor 15 mg/kg (upper left and right, using 4X or 10X magnification respectively) or vehicle control (bottom left and right, using 4X or 10X magnification respectively). A representative mouse for each group is shown (n=3) and these were selected based on comparable percentage of leukemic cells as measured by flow cytometry. Ki67-positive cells are dark brown.

To validate our in-vitro results we evaluated the effects of Selinexor on leukemic cell proliferation in-vivo using Eμ-TCL1 transgenic animals. Mice with established leukemia were treated for 2 consecutive days with vehicle control or Selinexor at 15 mg/kg. 5-ethynyl-2′-deoxyuridine (EdU) was injected on day 3 allowing for incorporation into the genetic material of proliferating cells for a full 24 hours. Afterwards, spleen and bone marrow samples were collected for FACS analysis of proliferating CD5⁺CD19⁺ leukemic cells as well as confirmatory Ki67 immunohistochemical staining. Selinexor effectively inhibited the percentage of proliferating leukemic cells (Figure 2G-I) confirming our invitro findings that Selinexor can abrogate proliferation of the malignant B cells in vivo. In conclusion, Selinexor was strikingly active in prolonging survival of Eμ-TCL1-SCID mice. Collectively, our data indicate that XPO1 inhibition prevents BCR-induced survival as well as trafficking of CLL cells to the protective stromal microenvironment. Although these mechanisms warrant further investigation, our data suggest the potential of this clinically viable class of inhibitors to increase sensitivity of CLL cells to current therapies and reduce the likelihood of leukemic relapse.

METHODS

Cell isolation

Human CLL and normal B cells were isolated and cultured as previously described¹. Nurse-like cell (NLC) co-cultures were established as previously described². HS5 stromal cells were obtained from ATCC.

Assessment of cell death

Cell death was assessed using either Annexin-V/PI staining or MTS assay as previously described¹.

Migration studies

CLL cells were stimulated for 1h with or without 10 μg/ml soluble anti-IgM in presence or absence of Selinexor (1 μM), washed and transferred to a transwell culture insert (1 × 10⁶ in 100 ul) located on the top of wells containing recombinant CXCL12 (100 ng), CXCL13 (500 ng) or media alone, and incubated at 37° C for 3h. Percent migration was determined as the number of migrated CLL cells in the lower chamber compared to the input as measured by flow cytometry.

Proliferation assay

CLL cells were stimulated with or without 3.2 μM CpG685 for 120 hours in presence or absence of Selinexor. Cells were pulsed with [³H]-thymidine (1.0 μCi/well) overnight and [³H]-thymidine incorporation was measured as described³.

Immunoblot analysis

Immunoblot analysis was performed as described by our group⁴ using the following antibodies: anti-p44/p42 MAPK, anti-phospho p44/p42 MAPK (Thr202/Tyr204), and anti-phospho Akt (Ser473) from Cell Signaling Technologies (Danvers, MA); anti-cyclin A (Santa Cruz Biotechnologies, Santa Cruz CA); and anti-GAPDH (Millipore, Billerica MA).

Quantitative Real-Time PCR

Real-time PCR was performed using cDNA prepared as described¹ and the following TaqMan gene expression assays: CCNA2 (ID: Hs00996788_m1) and 18S (ID:HS03003631_g1) (Life Technologies, Carlsbad CA).

Animal studies

Experiments were approved by the Ohio State University Institutional Animal Care and Use Committee. Eμ-TCL1-SCID mice were generated as described⁵ and treated with vehicle (30% PEG400 and 28% HPBCD) or Selinexor as indicated. Mice were sacrificed on development of peripheral blood leukocyte count (PBL) >200,000/μL and presence of other disease criteria causing discomfort as previously described¹. Spleen volumes were measured at time of sacrifice using a manual caliper (with 1mm error). Splenic volume was then calculated using the standard clinical ellipsoid equation of length × width × thickness × 0.523⁶.

Ki67 staining

Spleens and femurs were fixed in 10% neutral buffered formalin, embedded in paraffin blocks, and sectioned at 4um onto glass slides. Femurs were decalcified in EDTA prior to embedding. Sections were deparaffinized and hydrated to distilled water. Sections were then pre-treated with citrate buffer pH 6.0 in a decloaking chamber at 125 degrees Celsius for 15 minutes. Endogenous peroxidase was blocked with 3% hydrogen peroxide in methanol for 15 minutes. Serum-free protein block was applied for 10 minutes and rinsed with buffer. Sections were then incubated with primary anti-Ki67 antibody (Thermo-Shandon) diluted 1:100 at room temperature. Sections were rinsed in buffer and then incubated with biotinylated goat anti-rabbit secondary antibody (Vector Labs, Burlingame, CA) for 30 minutes at room temperature. Slides were treated with ABC Link (Vector Labs, Burlingame, CA) for 30 minutes at room temperature followed by DAB chromagen (Dako, Carpinteria, CA) for 5 minutes. Samples were rinsed in distilled water and then counterstained in hematoxylin for 40 seconds. Sections were finally dehydrated and coverslipped with a resinous mounting medium.

EdU (5-ethynyl-2’-deoxyuridine) Study

Eμ-TCL1 mice with active leukemia were injected intraperitoneally with 100 μg EdU. Spleen were collected and divided in half for flow cytometry and immunohistochemistry studies. Single-cell suspensions were prepared from spleen and bone marrow and EdU incorporation was detected by flow cytometry according to manufacturer protocol (Life Technologies). Leukemic cells were identified by anti-mouse CD45, CD19, and CD5 antibodies (BD Biosciences).

Statistical analysis

All analyses were performed by the OSU Center for Biostatistics; using previously described models¹.