Microenvironment‐induced PIM kinases promote CXCR4‐triggered mTOR pathway required for chronic lymphocytic leukaemia cell migration

Abstract Lymph node microenvironment provides chronic lymphocytic leukaemia (CLL) cells with signals promoting their survival and granting resistance to chemotherapeutics. CLL cells overexpress PIM kinases, which regulate apoptosis, cell cycle and migration. We demonstrate that BCR crosslinking, CD40 stimulation, and coculture with stromal cells increases PIMs expression in CLL cells, indicating microenvironment‐dependent PIMs regulation. PIM1 and PIM2 expression at diagnosis was higher in patients with advanced disease (Binet C vs. Binet A/B) and in those, who progressed after first‐line treatment. In primary CLL cells, inhibition of PIM kinases with a pan‐PIM inhibitor, SEL24‐B489, decreased PIM‐specific substrate phosphorylation and induced dose‐dependent apoptosis in leukaemic, but not in normal B cells. Cytotoxicity of SEL24‐B489 was similar in TP53‐mutant and TP53 wild‐type cells. Finally, inhibition of PIM kinases decreased CXCR4‐mediated cell chemotaxis in two related mechanisms‐by decreasing CXCR4 phosphorylation and surface expression, and by limiting CXCR4‐triggered mTOR pathway activity. Importantly, PIM and mTOR inhibitors similarly impaired migration, indicating that CXCL12‐triggered mTOR is required for CLL cell chemotaxis. Given the microenvironment‐modulated PIM expression, their pro‐survival function and a role of PIMs in CXCR4‐induced migration, inhibition of these kinases might override microenvironmental protection and be an attractive therapeutic strategy in this disease.


| INTRODUCTION
B-cell chronic lymphocytic leukaemia (CLL) is characterized by progressive accumulation of mature monoclonal B cells in the peripheral blood, bone marrow and secondary lymphoid tissues. 1 Several prognostic markers, such as the Rai and Binet staging systems, TP53 and other cytogenetic abnormalities, ZAP70 expression and immunoglobulin heavy variable (IGHV) gene mutational status can be used to predict the survival outcome of patients with CLL. 2 IGHV mutations distinguish two main biologically distinct subtypes of the disease, with different underlying genetic lesions, degree of clonal evolution, epigenetic changes and activated signalling pathways. The mutated IGHV subtype is associated with a good prognosis and the unmutated IGHV subtype with a poor prognosis. 1 While majority of circulating CLL cells are arrested in the G0 phase of the cell cycle, replenishment of the leukaemic population is dependent on a proliferating fraction in the bone marrow and lymphoid tissues. 3 In these compartments CLL cells interact with multiple bystander cell types, including bone marrow stromal cells (BMSCs), nurse-like cells (NLCs), follicular dendritic cells (FDCs), endothelial cells and T cells. 4 These microenvironment components create niches that communicate with CLL cells via direct contact and paracrine signals, protecting them from spontaneous and druginduced apoptosis, and fostering proliferation. Consistent with this, primary CLL cells isolated from lymph nodes exhibit gene expression signatures characterized by activation of the B-cell receptor (BCR) pathway, NFjB pathway and increased expression of E2F target genes. 5 Trafficking of neoplastic B cells to these proliferation-conducive compartments is controlled by chemokines. 6,7 One of the key chemokines involved in CLL cells homing is CXCL12 (formerly stromal-cell derived factor 1, SDF1). Activation of CXCR4 induces CLL cells chemotaxis, transendothelial migration and exhibits direct antiapoptotic effects. [8][9][10][11] Given the role of CXCR4 in CLL cell motility and viability, mechanisms regulating CXCR4 activity and CXCR4-triggered signal transduction are particularly interesting as potential therapeutic targets. Accordingly, highly active B-cell receptor signalling inhibitors, such as ibrutinib, lead to egress of CLL cells from the lymphoid compartments to a periphery in a mechanism that involves decrease of surface CXCR4 expression. 8 CXCR4 surface expression and recycling are regulated by PIM (provirus integration site for Moloney murine leukaemia virus) kinases, which phosphorylate CXCR4 on serine 339. 9 PIMs have been postulated as a key mechanism downstream of BCR, responsible for modulation of CXCR4 in CLL. 8,10 The family of PIM proteins involves three conserved oncogenic serine/threonine kinases, PIM1, PIM2 and PIM3. PIMs phosphorylate a broad range of substrates, which are engaged in cell growth, metabolism, proliferation, migration and drug resistance. [12][13][14] Increased activity of PIM kinases con- and MYC stabilization. 15 Moreover, PIM kinases phosphorylate 4E-binding protein 1 (4EBP1) and thus promote protein translation and tumour growth. [16][17][18] Given these pleiotropic effects, inhibition of PIM kinases appeared a highly promising therapeutic strategy in multiple human malignancies, including lymphoma. In this study, we investigated the expression of PIM kinases in CLL patients and further characterized the consequences of their inhibition. We demonstrate that PIMs expression is induced by the microenvironment-derived signals. Blocking PIMs activity with a newly developed small molecule inhibitor SEL24-B489 overrides protective microenvironment signals and induces CLL cell death. PIM inhibition blocks CLL cells migration in the CXCL12 chemokine gradient by affecting CXCR4 surface expression and CXCR4-dependent mTOR activation.
Consistent with these pathogenetic findings, we demonstrate that expression of individual PIM isoforms is higher in patients with more aggressive and advanced disease. Thus, PIM kinases directly support CLL cell survival and participate in the cross-talk between leukaemic cells and their microenvironment.

| CLL patient samples and cell culture
The study enrolled 141 newly diagnosed and 9 relapsed CLL patients, and was conducted after local bioethics committee approval and according to Declaration of Helsinki. Patient baseline characteristics are given in Table 1. Peripheral blood mononuclear cells were separated by Ficoll gradient centrifugation. B cells were isolated with the B cell isolation kit II (Miltenyi Biotec). After isolation CLL cells were maintained in RPMI-1640 medium supplemented with 10% autologous plasma, 10% FBS, 1% penicillin-streptomycin and 25 mmol/L HEPES buffer, at a density of 1 9 10 7 cells/mL. For coculture experiments CLL cells were layered over the 30%-confluent HS5 stromal cells and treated as indicated. After 48 hours CLL cells were harvested by gentle agitation and further processed as described.

| Quantitative PCR
RNA was isolated using Universal RNA Purification Kit (EURx, Gdansk, Poland) and transcribed to cDNA with Transcriptor First Strand cDNA Synthesis Kit (Roche). Transcript abundance was measured using SYBR Green PCR Master Mix (Applied Biosystems) and LightCycler 480 as described. 19 In brief, obtained CT values for individual PIM isoforms and housekeeping control (glyceraldehyde-3phosphate dehydrogenase; GAPDH) were used to calculate relative BIALOPIOTROWICZ ET AL. | 3549 transcript abundance, using the 2 ÀDDCT method. 19 Primer sequences are given in Table S1.

| Chemicals, apoptosis and migration assays
Pan-PIM kinase inhibitors SEL24-B489 and AZD1208 were kindly provided by Selvita S.A. 20 The mTOR inhibitor OSI-027 was purchased from SelleckChem (Houston, TX, USA). After incubation with the compounds, CLL cells were stained with AnnexinV-PE/7AAD kit (BD Biosciences) and analysed using the FACS Canto II. Annexin V-positive and double Annexin V/7AAD-positive cells were considered apoptotic. Migration assays were performed using transwell system (Corning, NY, USA). Briefly, CLL cells were incubated for 10 hours with 10 lmol/L SEL24-B489 or vehicle (dimethyl sulfoxide, DMSO); thereafter cells were counted and 5 9 10 6 cells were placed in the top chamber of the transwell dish (24-well plate format). The lower chamber contained medium with 500 ng/mL CXCL12 (R&D Systems). After 6 hours incubation (37°C, 5% CO 2 ) cells migrated into the lower chamber were counted using trypan blue exclusion assay.

| Immunoblotting
Protein extracts were prepared using RIPA buffer as previously described. 21 Protein extracts were PAGE-separated, electrotransferred to PVDF membranes (Millipore) and immunoblotted with primary and appropriate secondary antibodies (Table S2) Densitometric quantifications of band intensities were performed using Image Studio Lite software (https://www.licor.com/bio/produc ts/software/image_studio_lite/). PIM1/PIM2 levels were quantified relative to a pooled sample from all investigated patients, mixed at equal amounts and assigned as an arbitrary value 1. Quantified GAPDH was used as an internal control to normalize protein loading between samples.

| CXCR4 surface expression
Chronic lymphocytic leukaemia cells were incubated for 2-10 hours with 10 lmol/L SEL24-B489 or vehicle (DMSO), washed and stained with APC-conjugated CXCR4 antibody or APC-conjugated mouse IgG2a, j as an isotype control (Table S2), and analysed using the FACS Canto II.

| Statistical analysis
Comparisons between variables were performed with GraphPad Prism 6 software (GraphPad, La Jolla, CA, USA), using indicated tests; P < .05 was considered statistically significant.

| PIM1 and PIM2 are associated with unfavourable CLL prognosis
Given the established oncogenic function of PIM kinases in CLL cells, 9,22  disease (Binet C) exhibited significantly higher PIM2 transcript and protein levels, and higher PIM1 protein expression than CLL patients at earlier stages (Binet A/B; Figure 1A, Figure S2A). Significantly higher PIM2 transcript/protein level and PIM1 protein expression were also observed at the time of diagnosis in patients, who progressed after first line treatment during follow-up (median observation time = 27 months, range 2 to 230 months), ( Figure 1B, Figure S2B). Moreover, patients with unmutated IGHV loci showed significantly higher PIM1 transcript level than patients with mutated IGHV genes ( Figure 1C). In contrast, PIM3 transcript/protein abundance was not associated with any of clinical characteristics in CLL patients. These data indicate that PIM1 and PIM2 transcript and/or protein expression are increased in more aggressive CLL, prompting further questions about the mechanisms of their induction and consequences of their activity for the disease biology.

| Microenvironment signals induce PIM expression
Chronic lymphocytic leukaemia cell fate depends on microenvironment signals, which promote anti-apoptotic and proliferative circuitry involving STAT and NFjB transcription factors. 6,7 Since these transcription factors are known PIM inducers, 23

Contacts of CLL cells with T cells in the microenvironment
engage pro-survival, NFjB-inducing CD40-CD40L pathway. 24 Thus, we assessed expression of PIM kinase isoforms after incubation of peripheral CLL cells with CD40 ligand (CD40L). Activation of the CD40 receptor in CLL cells significantly increased PIM1-3 mRNA expression and protein abundance already after 1 hours of incubation ( Figure 2B, Figure S3). Finally, we co-cultured CLL cells with HS5 stromal cells. This interaction highly increased PIM3, but not PIM1 and PIM2 expression, when compared to CLL cells cultured without stromal support ( Figure 2C). Taken together, these data indicate that PIM kinases are under the control of tumour microenvironment, but the pattern of response to external stimuli differs between PIM isoforms.

| PIM inhibitor SEL24-B489 induces apoptosis in CLL cells
We next investigated the consequences of PIM inhibition in CLL cells using newly developed pan-PIM inhibitor, SEL24-B489. 20,25,26 Incubation of CLL cells with 1-10 lmol/L SEL24-B489 for 24 hours caused a significant, dose-dependent decrease in phosphorylation of PIM substrates: threonine 24/threonine 32 (T24/T32) of FOXO1/3a, serine 65 (S65) of 4EBP1 and serine 112 (S112) of BAD ( Figure 3A-C). SEL24-B489 inhibitor decreased phosphorylation of these substrates irrespective of IGHV gene mutation status ( Figure 3A, right panel). Similar effects were also observed with a referential pan-PIM inhibitor, AZD1208, indicating that SEL24-B489 induces expected, specific biochemical effects ( Figure 3A, left panel). 27 In contrast, SEL24-B489 did not decrease the phosphorylation of FOXO1/3 and 4EBP1 in normal B lymphocytes ( Figure S4A). Having confirmed inhibitor's on-target activity, we next assessed the effect of SEL24-B489 (1-10 lmol/L, 48 hours) on viability of CD19+ CLL cells obtained from peripheral blood of 23 treatment-na€ ıve patients and 5 healthy individuals (Table S3). Incubation with SEL24-B489 for 48 hours did not perturb normal B-cells viability; in marked contrast, SEL24-B489 triggered a dose-dependent increase in apoptosis of CLL cells (Figure 3D, Figure S4B). Cells obtained from IGHV-unmutated and IGHV-mutated CLL patients were equally susceptible to SEL24-B489-induced apoptosis ( Figure 3D, Table S3). CLL cells obtained from 5 patients who progressed after initial treatment also responded to the inhibitor, reaching 39%-83% of apoptotic cells for a 10 lmol/L SEL24-B489 dose ( Figure 3D, Table S3). Importantly, CLL cells carrying del17p13/TP53 point mutations were similarly sensitive to SEL24-B489 as p53-wild-type cells ( Figure 3D and E, Tables S1 and S3). Since microenvironment-derived signals typically protect CLL cells from spontaneous and drug-induced apoptosis, we examined sensitivity of CLL cells grown on HS5 monolayers to SEL24-B489. In 6 out of 7 cases, the compound at least partially overrode the protective signals from HS5 cells and markedly triggered apoptosis ( Figure 3F). It was previously shown that stromal cells induce CLL cells to express anti-apoptotic MCL1 protein, which could account for HS5 chemoprotective effects. 28  | 3551 F I G U R E 1 PIM expression is associated with CLL clinical parameters. PIM1/2 transcript levels were assessed by qPCR in 88 newly diagnosed CLL patients. Relative abundance of PIM1/2 transcripts was determined using 2 ÀDDCT method, with GAPDH used as a reference gene. PIM1/2 protein expression was determined by densitometric quantification of Western blots. GAPDH protein was used as a loading control. (A) PIM1 protein and PIM2 transcript/protein levels are significantly higher in patients with advanced CLL (Binet C), compared to subjects in earlier disease stages (Binet A/B). Please see Supplemental Figure 2A for example source Western blots. (B) PIM1 protein and PIM2 transcript/protein levels (at diagnosis) are higher in patients who eventually progressed after first-line treatment. Please see Supplemental Figure 2B for example source Western blots. (C) PIM1 transcript abundance is significantly elevated in patients with unmutated IGHV status (U-CLL) comparing to subjects with mutated IGHV configuration (M-CLL). Statistics were calculated using one-way ANOVA followed by Tukey's post-hoc test for three-group comparison and Mann-Whitney test for comparison between two groups. *** for P < .001, ** for P < .01 and * for P < .05; "n" refers to the number of patients. GAPDH was used as a housekeeping reference for qPCR analyses.

| PIMs modulate mTOR activity downstream of CXCR4
Given the marked decrease in CXCR4-driven migration caused by PIM inhibition and its relatively moderate effect on CXCR4 surface expression, we hypothesized that PIM kinases influence CXCR4driven migration by an additional mechanism. We found that incubation of leukaemia cells with CXCL12 led to increased phosphorylation of mTOR (S2448) and AKT (S473), revealing CXCL12-mediated activation of this pathway in CLL ( Figure 5A). To determine whether PIM inhibition interferes with mTOR pathway, we first incubated CLL cells with SEL24-B489 and found decreased phosphorylation of mTOR pathway components, including p-mTOR (S2448), p-TSC2 (S1798) and p-AKT (S473; Figure 5B). Consistent with this, we also found decreased phosphorylation of direct mTOR substrates, 4EBP1 serine 37 and threonine 46 (S37/T46), indicating that PIM inhibition blocks signalling through mTOR pathway (Figure 5B). 29 We next determined whether PIM inhibition could block CXCL12-mediated mTOR activation. As expected, SEL24-B489 or OSI-027 (an mTORC1/2 inhibitor) markedly decreased CXCL12induced phosphorylation of mTOR pathway components, demonstrating that PIM inhibitors block CXCR4-dependent mTOR activation and signalling ( Figure 5C). We next investigated the impact of mTOR inhibition on CXCR4-dependent migration. For these F I G U R E 2 Microenvironment signals induce the expression of PIM kinases. CLL cells from 7 donors were incubated with 10 lg/mL anti-IgM (a-IgM) for 8 h and 24 h (A) or CD40L (50 ng/mL, 1 h) (B), or co-cultured with HS5 cells for 48 h (C), and then collected for qPCR analyses. PIM1/2/3 transcript abundance was quantified using 2 ÀDDCT method, where GAPDH was used as a reference gene. The results are expressed relative to the value of untreated sample, assigned to an arbitrary value 1. * for P < .05; Wilcoxon matched pairs test. F I G U R E 5 CXCR4/CXCL12 signal is transduced through mTOR pathway in a PIM-dependent manner. (A) CXCL12 activates mTOR signalling pathway. After incubation with 500 ng/mL CXCL12 for 0-60 min, primary CLL cells were lysed and assessed for p-mTOR and p-AKT levels by WB. Numbers below the blots indicate relative changes in phospho-protein abundance, determined by densitometric quantification using Image Studio Lite programme. (B) SEL24-B489 blocks the baseline activity of mTOR pathway. Primary CLL cells were incubated with 10 lmol/L SEL24-B489 for 1 h. Changes in phospho-protein abundance were determined by Western blotting, and quantified using Image Studio Lite programme. (C) Pan-PIM inhibitor SEL24-B489 and mTOR inhibitor OSI-027 inhibit CXCL12-activated mTOR pathway. CLL cells were pre-incubated with 10 lmol/L SEL24-B489 or OSI-027 for 1 h and then stimulated with CXCL12 (500 ng/mL, 15 min). (D) Inhibition of PIM and mTOR kinases impairs CLL cells migration in the CXCL12 gradient. Primary CLL cells were pretreated with SEL24-B489 or OSI-027 (both at 10 lmol/L) and placed in a transwell chamber in CXCL12 gradient. Numbers of migrated cells were determined after 6 h using trypan blue exclusion assay. Bars represent mean AE SD from triplicates, ***P < .001, **P < .01 and *P < .05 calculated with Mann-Whitney test.
signalling or ligation of TNF family receptors, such as CD40. 34,35 Of note, both these receptors lead also to a delayed induction of STATs gatekeeper site, restoring its pro-apoptotic activity. 40 In addition, inhibition of PIM-dependent protein translation decreases abundance of an antiapoptotic BCL2 family protein, MCL1. 30 We show here that PIM kinase inhibition also markedly reduced the expression of MCL1 protein induced by stromal cell contact. Thus, PIM inhibition triggers proapoptotic mechanisms that are not blocked, or only partially blocked, by microenvironmental support, resulting in p53-independent cell death. 41 The homing of CLL cells to a proliferation-conducive and protective lymphoid compartments is predominantly regulated by CXCR4 chemokine receptor. 6 Thus, interference with the activation of CXCR4 receptor in CLL cells facilitates their egress from the lymph node niche and/or prevents their homing to lymphoid organs. 9  Taken together, our results suggest that SEL24-B489 pan-PIM