LCK over-expression drives STAT5 oncogenic signaling in PAX5 translocated BCP-ALL patients.

The PAX5 gene is altered in 30% of BCP-ALL patients and PAX5 chromosomal translocations account for 2-3% of cases. Although PAX5 fusion genes significantly affect the transcription of PAX5 target genes, their role in sustaining leukemia cell survival is poorly understood. In an in vitro model of PAX5/ETV6 leukemia, we demonstrated that Lck hyper-activation, and down-regulation of its negative regulator Csk, lead to STAT5 hyper-activation and consequently to the up-regulation of the downstream effectors, cMyc and Ccnd2. More important, cells from PAX5 translocated patients show LCK up-regulation and over-activation, as well as STAT5 hyper-phosphorylation, compared to PAX5 wt and PAX5 deleted cases. As in BCR/ABL1 positive ALL, the hyper-activation of STAT5 pathway can represent a survival signal in PAX5 translocated cells, alternative to the pre-BCR, which is down-regulated. The LCK inhibitor BIBF1120 selectively reverts this phenomenon both in the murine model and in leukemic primary cells. LCK inhibitor could therefore represent a suitable candidate drug to target this subgroup of ALL patients.


INTRODUCTION
The PAX5 gene belongs to the PAX gene family of transcription factors and is essential for B cell commitment [1]. It functions both as a transcriptional activator and a repressor of different target genes involved in lineage development. Furthermore, PAX5 has been recently reported to be target of aberrancies, including mutations, deletions and translocations, in about 30% of pediatric patients affected by BCP-ALL, [2] the most frequent leukemia subset in children [3]. PAX5 translocations occur in approximately 2-3% of patients, [4] with a variety of partner genes, such as transcription factors, kinases, structural proteins and others [2,4,5]. The t(9;12) is the most recurrent PAX5 translocation and it encodes for the PAX5/ETV6 fusion gene, [2,[5][6][7][8] which results in the juxtaposition of two transcription factors, fundamental in hematopoiesis and in B cell development [9]. The PAX5/ETV6 fusion protein retains the DNA binding domain of PAX5, while it substitutes its regulatory domains with the DNA binding, dimerization and transcription regulation domains of ETV6 [7].
We previously reported that PAX5/ETV6 is an aberrant transcription factor that localizes in the nucleus, [10] and alters the transcription profile of pre-BI cells, mainly deregulating genes involved in pre-BCR assembly and signaling [11]. Moreover, among the differentially expressed genes, a significant number has been described to be transcriptional direct targets of Pax5. One of the top-ranking up-regulated genes was the Src kinase Lck, a Pax5 repressed target gene, which has a potential role in a wide range of hematological malignancies [12][13][14][15][16] and interestingly, acts as an inducer of Stat5 hyper-phosphorylation in Ba/F3 pro-B cells [17].
We and others previously contributed in describing the ability of PAX5/ETV6 and other PAX5 fusion proteins in impairing the rearrangement of the μ heavy chain and down-regulating the expression of genes involved in the signaling pathway downstream of the pre-BCR, [11] a mechanism known to be responsible not only for pre-BI cells differentiation but also for their proliferation and survival.
Therefore, the aim of the present study was to test the hypothesis that the activation of STAT5 through LCK up-regulation could represent a pro-survival signaling pathway activated by PAX5 fusion proteins and alternative to the pre-BCR, switched-off in this context. Primary cultures of wild type (wt) pre-BI cells stably transduced with the retroviral PAX5/ETV6 vector and patients primary cells carrying PAX5 fusion genes were used to model this hypothesis.

Lck kinase activity is enhanced in PAX5/ETV6 positive cells
Lck kinase activity is regulated via the reversible phosphorylation of the negative regulatory Y505 residue in the C-terminal segment, thus inducing the clamp of the C-tail on its own SH2 domain, locking the kinase in an inactive, closed conformation [19].
The total Lck protein level was up-regulated (mean FC = 1.79, range 1.42-2.48, p < 0.01; mean FC = 1.97, range 1.10-3.02, p < 0.05; mean FC = 1.12, range 1.06-1.40, n.s., in LY5.1FL, B6BAFL and FLB6-67 cells, respectively), and PAX5/ETV6 induced a statistically significant de-phosphorylation of the Lck inhibitory domain (Lck Y505 ), causing Lck over-activation ( Figure 1C Lck phosphorylation in the Y505 residue is regulated by the C-terminal Src kinase Csk, which induces a close conformation to the Lck protein, thus preventing the accessibility to its catalytic domain [19]. Indeed, we observed a significant down-regulation of Csk expression, which could be responsible for the de-phosphorylation at the Lck inhibitory domain ( Figure 1E, Supplementary Figure S1D and S2D). Moreover the Lck target Zap70 was found up-regulated in all the pre-BI cell populations ( Figure 1F, Supplementary Figure S1E and S2E).
To better characterize the patients cohort considered for this study, we performed MLPA analysis of BM diagnostic samples (Supplementary Table S1). All the PAX5 deleted cases presented CDKN2A/CDKN2B deletions (2 of them homozigous), possibly reflecting a larger deletion on chromosome 9. Patient number 9 showed IKZF1 and BTG1 deletions as well, and patient 7 showed IKZF1 deletion. Therefore, at best of this knowledge, there is no evident correlation between gene deletions and different LCK expression levels.
Out of the 5 cases carrying PAX5 translocations, cells were available from only 3 patients (namely, 2 PAX5/AUTS2 and PAX5/SOX5). Therefore, due to the small quantity of the samples, we expanded the cells by intravenous injection into NOD/SCID mice. When the mice became ill due to overt leukemia, they were sacrificed, and leukemia cells were harvested from the bone marrow. Leukemic infiltration was confirmed by flow cytometry, as shown in Supplementary Figure S3. Quantitative PCR confirmed that LCK was up-regulated in PAX5 translocated patient-derived xenograft BM cells (mean FC = 21.58, p < 0.01) (Figure 2A), demonstrating that are representative of diagnostic samples.

PAX5/ETV6 induces Stat5 over-phosphorylation
We previously demonstrated in the murine model that PAX5/ETV6 cells fail to successfully rearrange the μ heavy chain and thus to express a functional pre-BCR on the cellular surface [11]. In pre-BI cells, the pre-BCR signaling is fundamental not only for their differentiation, but also for their survival and proliferation together with the IL7R pathway, [21] which leads to Stat5 phosphorylation and the consequential activation of cell cycle progression inducers. Since PAX5/ETV6 provided a survival advantage upon IL7 withdrawal, [10] we wondered if it could be due to constitutive, ligand-independent over-activation of the www.impactjournals.com/oncotarget IL7R-Stat5 pathway. This could be exerted through Lck over-expression, which had been previously described to induce Stat5 hyper-phosphorylation in Ba/F3 pro-B cells [17].  [21].
Downstream in the pathway, we observed the up-regulation of cMyc and Ccnd2, at early and late time points after IL7 stimulation, thus indicating that Stat5 hyper-phosphorylation led to the activation of the transcription of its target genes ( Figure 3B-C, Supplementary Figure S4D-E, S5C-D, S6C-D) [22].

The Lck inhibitor BIBF1120 can revert the advantage of PAX5/ETV6 cells
The treatment with the Lck inhibitor BIBF1120 [14] caused a significant reduction of Stat5 Y694 phosphorylation in PAX5/ETV6 cells, as indicated by the decreased number of positive cells and the Mean Fluorescence Intensity (MFI), while it had no effect on MIGR-GFP control, where Lck is expressed at lower levels ( Figure 4A-B). Overall, this supports the specificity of Lck activity on cell survival. By tracking of the replication cycles, we demonstrated that BIBF1120 reduces the replicative rate of PAX5/ETV6 cells, thus strikingly abrogating their proliferative advantage ( Figure 4C). Indeed, cell cycle analysis showed that BIBF1120 reduced the increase in the replicative S phase exclusively in PAX5/ETV6 cells (Supplementary Figure S7A-B).
Furthermore, overnight administration of BIBF1120 was able to induce cell death in a proportional manner to LCK mRNA level. In particular, LCK levels were inversely correlated to cell viability (10 out of 12 patients with a minimal cellular viability of 15% were considered), ( Figure 5C and Supplementary Figure S10). This makes BIBF1120 a candidate drug to target PAX5 translocated ALL.

DISCUSSION
The role of PAX5 fusion genes in leukemogenesis and transformation events is not fully understood yet. It has been demonstrated that PAX5/ETV6, as well as other PAX5 fusion proteins, impair the expression of the μ heavy chain and down-regulates pre-BCR associated molecules at the surface level [23]. In the present study,  we wanted to test the hypothesis that the activation of Stat5 through Lck up-regulation could represent a prosurvival signaling pathway activated by PAX5 fusion proteins and alternative to the non-functional pre-BCR.
Indeed, we initially showed Lck over-expression in an in vitro model of murine wt pre-BI cell populations co-coltured on OP9 stroma and transduced with PAX5/ ETV6. We demonstrated that, despite a moderate overexpression of the total form of Lck protein, Lck was significantly de-phosphorylated in the inhibitory residue Lck Y505 , thus causing Lck over-activation. Further confirming the de-phosphorylation at the Lck inhibitory domain, we demonstrated the down-regulation of the C terminal Src kinase (Csk) in PAX5/ETV6 cells.
Indeed, Csk regulates Lck phosphorylation at the residue Y505, thus inducing the clamp of Lck C-tail on its own SH2 domain, locking the kinase in an inactive, closed conformation. As a confirmation of Lck hyperactivation, we demonstrated the up-regulation of its target Zap70 [19].
Thereby, in PAX5/ETV6 cells Lck could be overactivated through two main mechanisms: a) the PAX5/ETV6 mediated up-regulation of Lck transcription; b) the downregulation of Csk, thus resulting in the de-phosphorylation of its inhibitory domain. The same mechanism can potentially be hypothesized for other PAX5 fusion proteins.
Importantly, for the first time we showed LCK overexpression in bone marrow primary cells from leukemic activation in frozen blast cells at basal level and after stimulation with hIL7 (50 ng/ml for 30 minutes). Despite the low basal level with no biological differences in the patients' groups, stimulation with hIL7 induces significantly higher STAT5 activation in PAX5 translocated cells. (B) Overnight in vitro co-culture of blast cells on OP9 stroma confirmed that PAX5 translocated cells show a higher STAT5 activation profile compared to PAX5 wt and PAX5 deleted patients. Stimulation with hIL7 increased the level of STAT5 Y694 positive cells and PAX5 translocated cells maintained the highest activated profile. Overnight treatment with BIBF1120 is able to reduce STAT5 activation in PAX5 fusion gene positive cells, but it doesn't exert any significant effect on PAX5 wt and PAX5 deleted cells. Treatment with BIBF1120 abolishes the effect of stimulation with hIL7 in PAX5 translocated cases, keeping the percentage of STAT5 activated cells similar to the levels of the other groups. (C) LCK levels are inversely correlated to leukemia cells viability. Test t, *p < 0.05; **p< 0.01; ***p < 0.001. patients carrying at diagnosis different PAX5 fusion genes, [20] thus demonstrating that this is a common feature of PAX5 translocations.
At protein level, not only PAX5 translocated, but also PAX5 deleted patients showed higher LCK expression compared to PAX5 wt cases, as a result of the PAX5 haploinsufficiency on its physiological repressed targets. However, exclusively PAX5 fusion genes induced de-phosphorylation at the LCK Y505 residue, thus resulting in LCK over-activation, while PAX5 deleted patients didn't display strikingly over-activation of LCK. Taken together, these data suggest that LCK over-expression, together with its over-activation, is specifically associated to the dominant activity of PAX5 fusion proteins and contributes to the biological distinction of the PAX5 translocated from PAX5 deleted patients.
In murine pre-BI cells, the survival and proliferation role of the pre-BCR signaling is exerted together with the IL7R pathway, [21] which leads to Stat5 phosphorylation, and consequentially activation of cell cycle progression inducers. However, we previously demonstrated that PAX5/ETV6 cells do not express a functional pre-BCR [11] and they show a short term survival advantage upon IL7 withdrawal, which could be due to the constitutive, ligand-independent over-activation of the IL7R-Stat5 pathway. Interestingly, Lck over-expression had been previously described to induce Stat5 hyperphosphorylation in Ba/F3 pro-B cells and in T lymphoma cells as well [17,24]. This feature resembles what reported in Ph+ BCP-ALL cases, where blasts do not express a functional pre-BCR, but they show STAT5 over-activation [25][26][27].
Indeed, we demonstrated by phosphoflow analysis that PAX5/ETV6 transduced cells showed increased Stat5 phosphorylation, and the up-regulation of two downstream targets cMyc and Ccnd2, known to have a role in cell cycle progression and malignant transformation [22,28,29]. We also demonstrated that IL7Rα (CD127) was expressed at the same level both in PAX5/ETV6 and MIGR-GFP cells, thus suggesting that increased Stat5 phosphorylation was independent from classical IL7R signaling cascade [21].
Moreover, PAX5/ETV6 pre-BI cells display additional typical features of oncogenic transformation, such as an increased replicative S phase population and a faster replication rate. This phenomenon may be caused by Stat5 hyper-phosphorylation, together with the downregulation of tumor suppressor genes (Gadd45b and Lats2) [11,30,31].
In the human setting, we confirmed that LCK overexpression leads to the over-activation of STAT5 signaling pathway both at the basal level and after stimulation with hIL7 in PAX5 translocated cases, compared to PAX5 wt and PAX5 deleted blast cells.
Interestingly, STAT5 over-activation and PAX5 haplo-insufficiency have been described as synergistic players to initiate ALL [28,32]. In the present study we demonstrated that both these aberrancies are not only necessary to drive leukemia, but they are a direct consequence of the expression of the PAX5 fusion protein, which can a) repress the activity of endogenous PAX5 [11] and b) activate STAT5 through LCK ( Figure 6).
In this scenario, PAX5/ETV6 (and PAX5 fusion proteins in general) share some features of BCR-ABL1 leukemic transformation: a) both the fusion genes impair the μ heavy chain rearrangement and the pre-BCR exposure on the cellular surface; [10,11,26] and b) both induce STAT5 hyper-phosphorylation, via LCK in PAX5 translocated cases and via BCR/ABL1 itself in Ph+ cells, respectively [25,26,32]. This similarity might at least partly explain the relatively frequent occurrence of PAX5 abnormalities in Ph-like ALL [33].
The LCK inhibitor BIBF1120 specifically abrogated the survival advantages of both murine PAX5/ETV6 pre-BI cells and PAX5 translocated blast cells, down-tuning STAT5 phosphorylation and preventing the effect of stimulation with IL7, indicating the specificity of the LCK direct mechanism ( Figure 6).
LCK has recently been reported as a key molecule in multiple subsets of leukemia. Indeed, LCK over-activation has been found to correlate with poor response to treatment in pediatric BCP-ALL patients, [12] MLL-rearranged ALL [13] and in chronic lymphoblastic leukemia (CLL), [14,15] thus proposing LCK activation as a putative marker for glucocorticoid resistance. Moreover, LCK has been also shown to be essential for the proliferation and survival in a subset of T-ALL characterized by NUP214/ ABL1 activity (6% of T-ALL cases) [16].
Besides BIBF1120, LCK has been found as a target of several tyrosine kynase inhibitors, such as Imatinib and Dasatinib, and it has been demonstrated in multiple studies that TKI treatment specifically down-regulates LCK expression and diminishes LCK activation [14,34,35]. Therefore, the LCK driven oncogenic signaling may candidate tyrosine kinase inhibitors as a possible treatment for PAX5 translocated patients.

Ethic statement
Investigation has been conducted in accordance with the ethical standards and according to the Declaration of Helsinki and according to national and international guidelines and has been approved by the authors' institutional review board.

Patients' cohort
PAX5 translocated (n = 5), PAX5 deleted (n = 4) and PAX5 wt (n = 5) pediatric BCP-ALL cases enrolled in the AIEOP-BFM ALL2000 protocol and selected based on cytogenetics and FISH were considered for this study [20]. Patients' clinical and cytogenetics features are reported in Table 1.

Multiplex Ligation dependent Probe Amplification (MLPA)
The selected cohort was screened for aberrations in B cell genes (PAX5, ETV6, RB1, BTG1, EBF1, CDKN2A, CDKN2B, and P2RY8-CRLF2) by the MLPA SALSA p335 kit (MRC Holland, Amsterdam, The Netherlands) using 125 ng of genomic DNA. The assays were performed according to the manufacturers' protocol. Electrophoresis and quantification of fluorescein amidite-labeled amp icons were performed on an ABI-3130 genetic analyzer (Applied Biosystems, Carlsbad, CA). The resulting peak intensities were normalized to the manufacturers' control probes and to normal DNA as a reference. An intensity ratio between 0.75 and 1.3 was considered to represent a normal copy number, a ratio between 0.25 and 0.75 a monoallelic deletion, and a ratio < 0.25 a biallelic deletion. The results of MLPA analysis are reported in Supplementary Table S1.

Primary leukemia cell xenograft
BM cells (1 × 10 6 ) collected at the diagnosis of PAX5 translocated BCP-ALL patients were inoculated via intravenous injection into NOD/SCID mice (Charles River Laboratories, Wilmington, MA, USA) after sublethal irradiation (125 Rad) with Radgil (Gilardoni, Mandello (LC), Italy). When the mice became ill due to overt leukemia, they were sacrificed, and leukemia cells were harvested from bone marrow and spleen. Leukemic infiltration was confirmed by flow cytometry, as shown in Supplementary Figure 3.

Retroviral transduction
Wt pre-BI cell populations were transduced with either MIGR-PAX5/ETV6 or MIGR-GFP vector. The retroviral supernatant was obtained using MIRUS transfection reagent (Mirus Bio LLC, Madison, Wisconsin, USA) and Phoenix packaging cell line according to the manufacturer's instruction. On day +3 post transduction, the GFP positive cells were sorted using FACS Aria sorter (Becton Dickinson Biosciences, San Jose, California, USA).

Reverse transcription-PCR and real-time quantitative-PCR assays
RNA extraction was performed using TRIZOL reagent (Invitrogen, Life Technologies, Carlsbad, California, USA), following the manufacturer's protocol.
Superscript II enzyme (Invitrogen, Life Technologies, Carlsbad, California, USA) was used for cDNA synthesis. Primers and probes for real time quantitative PCR were selected according to the Software Probe Finder (Roche Diagnostics, Basel, Switzerland). Data were expressed using the comparative ΔΔCt method. In the murine setting, the Hprt transcript was considered as a reference and MIGR-GFP cells as standardization control, whereas in human cell samples the ABL1 gene was selected as a housekeeping gene, and PAX5 wt patients as standardization controls [36]. Both t test and SD values refer to triplicates of a single experiment and 3 independent experiments were performed for each gene. Primers and probes are reported in Supplementary  Table S2.

BIBF1120 treatment
Murine pre-BI cells were treated with either 50nM LCK inhibitor BIBF1120 (Selleckchem, Houston, Texas, USA) or vehicle (DMSO) after synchronization by overnight IL7 withdrawal. Human blast cells were cultured overnight with 50nM LCK inhibitor BIBF1120 or vehicle.

Viability assay
The viability of human leukemia cells was evaluated by staining with BD Fixable Viability Stain V450, according to manufacturer's protocol, and analyzed by BD FacsCanto II (Becton Dickinson Biosciences, San Jose, California, USA).