Targeting STAT5 or STAT5-Regulated Pathways Suppresses Leukemogenesis of Ph+ Acute Lymphoblastic Leukemia

Philadelphia chromosome-positive (Ph+) acute lymphoblastic leukemia (ALL) is characterized by the t(9;22) translocation that generates the p190- and, less frequently, the p210-BCR-ABL1 fusion protein, both of which have constitutive tyrosine kinase activity (1, 2). The Philadelphia chromosome is the most common cytogenetic abnormality in adult patient with ALL, occurring in about 20% to 30% of the cases (3, 4).

The incidence of Ph+ ALL increases with age, with up to 50% being diagnosed in patients older than 60 years (5). Allogeneic hematopoietic stem cell transplantation (HSCT) is an effective consolidation therapy for patients with Ph+ ALL who have achieved a complete response (CR) after induction of remission chemotherapy (6). However, 10% to 20% of patients fail to achieve CR and allogeneic HSCT is only available for patients with suitable matched donors. In addition, the odds of treatment-related mortality as well as relapse are high (7). The outcome of patients with Ph+ ALL has improved significantly with the introduction of imatinib and second-generation TKI as a first-line therapy, especially when used in combination with chemotherapy (8). However, resistance to TKI develops rapidly in most patients with Ph+ ALL and the 5-year overall survival is <50% (9–11).

Thus, inhibiting the BCR-ABL1 kinase with TKI fails to eradicate most Ph+ ALL clones due to BCR-ABL1-dependent and -independent mechanisms of resistance (12). Collectively, these findings indicate that additional pathways need to be identified and targeted for an effective treatment of Ph+ ALL.

STAT5 has a pivotal role in promoting cell survival and the subsequent B-cells' expansion during early B-cell development. CD19-Cre-mediated deletion of STAT5A/B in the B-cell compartment impairs IL7-activated survival pathways, blocking B-cell differentiation at the pre-pro-B stage (13–15). Of interest, the defective B-cell development induced by genetic deletion of STAT5 was rescued by restoring expression of STAT5-regulated BCL2 (16).

In malignant precursor B cells, deregulated JAK-STAT5 activity may allow survival of leukemic cells independently of stroma-derived cytokine signals (17). The STAT5 pathway is constitutively active in Ph+ ALL and in a subset of B-ALL that contains activating mutations in JAK1 or JAK2 (18–20). Importantly, STAT5 can be activated in Ph+ leukemia cells either indirectly through JAK2 phosphorylation or directly by BCR-ABL1 because STAT5 is a known substrate of BCR-ABL1 (21), and an intact STAT5 signaling is required for maintenance of BCR-ABL1-driven leukemias (19). Furthermore, STAT5 is a marker of disease progression in Ph+ chronic myeloid leukemia (CML), based on correlation of high STAT5 mRNA levels with TKI resistance and advanced disease stages, irrespective of the presence of tyrosine kinase domain (TKD) BCR-ABL1 mutations (22, 23).

Together, these data suggest that STAT5 itself or STAT5-regulated pathways could serve as rational targets not only to circumvent the BCR-ABL1-dependent TKI resistance of Ph+ ALL but also to suppress growth-promoting STAT5 signals activated through BCR-ABL1-independent mechanisms. In this study, using genetic and pharmacologic approaches, we show that STAT5 is critical for the growth of Ph+ ALL cell lines and of newly diagnosed and relapsed/TKI-resistant patient-derived Ph+ ALL cells ex vivo and in mice. Moreover, we found that the growth-promoting effects of STAT5 depend on changes in the expression/activity of PIM-1, BIM, and BCL2 and that these proteins can serve as therapeutic targets ex vivo and in xenografts of patient-derived Ph+ ALL cells.

The SUP-B15 cell line (Ph+ ALL) was purchased from ATCC; the Z181 cell line (Ph+ ALL) was kindly provided by Dr. Z. Estrov (M.D. Anderson Cancer Center, Houston, TX); the BV173 cell line (Ph+ CML-lymphoid blast crisis; ref. 24) was kindly provided by Dr. N. Donato (NIH). EBV-immortalized B cells GM03798 and GM12878 were purchased from the Coriell Institute (Camden, NJ). All these cell lines and the TKI-resistant T315I-BV173 derivative line (25) were cultured in Iscove's Modified Dulbecco's Medium (Corning) supplemented with 10% heat-inactivated FBS (Biowest), 1% penicillin–streptomycin (Thermo Fisher Scientific), and 1% l-glutamine (Thermo Fisher Scientific) at 37°C, 5% CO₂. The Ph-like ALL MUTZ-5 and MHH-CALL-4 cell lines were kindly provided by Dr. M. Carroll (University of Pennsylvania, Philadelphia, PA). These lines were cultured in RPMI supplemented with 20% FBS, 1% penicillin–streptomycin, and 1% l-glutamine at 37°C, 5% CO₂.

Cell lines were tested for Mycoplasma every 3 months as described (26). Ph+ ALL cell lines were routinely authenticated by monitoring B-cell markers and BCR-ABL1 isoform expression. Ph-like cell lines were authenticated by B-cell immunophenotyping (CD19 and CD10), and by monitoring CRLF2 and phospho-STAT5 expression.

Primary adult human Ph+ ALL cells were kindly provided by: Dr. Alessandro Rambaldi (Hematology and Bone Marrow Transplant Unit, Bergamo, Italy), Dr. Luke F. Peterson (University of Michigan), Dr. Michael Caligiuri (City of Hope Cancer Center, Duarte, CA), Dr. Pierluigi Porcu (Thomas Jefferson University, Philadelphia, PA), and Dr. Martin Carroll (University of Pennsylvania). Main characteristics of the samples are described in Supplementary Table S1. G-CSF–mobilized peripheral blood CD34⁺ primary cells (J48 and J50) from healthy donors were obtained from the Bone Marrow Transplantation Unit, Thomas Jefferson University. Primary adult human Ph-like cells were provided by Dr. Pierluigi Porcu. Primary cells were maintained in SFEM (Stem Cell Technologies) supplemented with IL3 (10 ng/mL), IL6 (10 ng/mL), IL7 (10 ng/mL), Flt3-L (20 ng/mL), and stem cell factor (30 ng/mL; ProSpec).

Relapsed/TKI-resistant Ph+ ALL samples were assessed for the presence of ABL1 TKD mutations by Sanger sequencing; briefly, RNA was extracted with RNAeasy Plus Mini Kit (Qiagen), cDNA was reverse transcribed by using Superscript III (Thermo Fisher Scientific). The ABL1 TKD was PCR-amplified with a forward primer located on BCR exon 1 (p190 FW: 5′-CTCGCAACAGTCCTTCGACA-3′) or with a forward primer located on exon 12 (P210 FW: 5′-CTGCAGATGCTGACCAACTC-3′) and a common reverse primer (TKD REV: 5′-CCTGCAGCAAGGTACTCACA-3′). Gel purified PCR products were sequenced in both directions using TKD REV primer or TKD FW primer (5′-CCCACTGTCTATGGTGTGTCC-3′).

Cell viability was assessed by MTT assay in 96-multiwell plates. Cells were seeded at a concentration of 150,000 or 300,000 cells/mL (depending on length of the treatment and cell line growth rate) and then treated with DMSO (Ctrl) or the specific drug. At the time point of interest, 100 μL of cell cultures were transferred in 96-multiwell plates and incubated with 10 μL of 0.5 mg/mL 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT; Sigma Aldrich) at 37°C for 3 hours. Then, formazan crystals were dissolved with 0.1 M HCl in 2-propanol and absorbance was measured at 570 nm using a μQUANT microplate scanning spectrophotometer (BioTek Instruments) and KC4 V.3.4 software.

Cells were incubated at 100,000/mL with 1 μL of the CellEvent Caspase 3/7 Green Detection Reagent (Thermo Fisher Scientific) for 25 minutes at 37 °C and analyzed by flow cytometry using a 488 nm excitation laser.

For GFP-positive cell lines, apoptosis was measured by Annexin V staining: 100,000 cells were washed in Annexin V Binding Buffer (10 mmol/L HEPES, 140 mmol/L NaCl, and 2.5 mmol/L CaCl2, pH 7.4), spun and resuspended in 50 μL of Annexin V Binding Buffer, then incubated with 1.5 μL of Cy 5.5-Annexin V (BD Biosciences) for 15 minutes at room temperature. Samples were analyzed by flow cytometry using a 640 nm laser.

shRNA lentivirus-transduced or drug-treated SUP-B15, BV173, or Z181 cells were plated in methylcellulose (Stem Cell Technologies) supplemented with Iscove's Modified Dulbecco's Medium, 10% heat-inactivated FBS, 1% penicillin–streptomycin, and 1% L-glutamine. Colonies were counted 10 days later.

shRNA lentivirus-transduced or drug-treated blast cells from Ph+ ALL, Ph-like patients, or CD34⁺ cells from healthy donors, were plated in methylcellulose supplemented with SFEM, IL3 (10 ng/mL), IL6 (10 ng/mL), IL7 (10 ng/mL), Flt3-L (20 ng/mL), and stem cell factor (30 ng/mL). Colonies were counted 7 to 10 days later.

STAT5 short hairpin RNA (shSTAT5) designed against the STAT5 mRNA sequence ATCTGGCTTGTTAATGAGTAG (TRCN0000019356) and SCR short hairpin RNA (shSCR) CCTAAGGTTAAGTCGCCCTCG were obtained in the pLKO.1-puro vector (Dharmacon) and subsequently cloned in the Tet-pLKO-puro vector (Addgene) for inducible shRNA expression, using AgeI and EcoRI restriction enzyme sites and following Addgene's protocols.

BIM short hairpin RNA (shBIM) designed against the BIM mRNA sequence CCAGACCACTACTGAATATAA (TRCN0000020155) and cloned into the Tet-pLKO-Neo lentiviral-vector were kindly provided by Dr. Z. Jagani (Novartis; ref. 27).

The PIM1-s cDNA (NCBI: NM_0026483) was generated from total RNA purified from BV173 cells and inserted in the XbaI-SalI restriction enzyme sites of the pUltra, GFP-expressing, lentiviral vector developed by Dr. Malcolm Moore (Addgene; plasmid #24129). The BCL2 cDNA (isoform α, NM_000633.2) was PCR-amplified from reverse transcribed total RNA of BV173 cells with a forward primer including the XbaI restriction sequence and with a reverse primer including a HA-tag coding sequence (TACCCATACGATGTTCCAGATTACGCT) and an EcoRI restriction site sequence. The product was then inserted into the XbaI-EcoRI sites of the pUltra-hot, mCherry expressing, lentiviral vector (Addgene; plasmid #24130).

The FG12 MCL1/IRES-GFP lentivirus was kindly provided by Dr. Maria S. Soengas (Spanish National Cancer Centre, Madrid, Spain).

For lentiviral production, 293T cells were transiently transfected with the indicated plasmids, the envelope plasmid pMD2.G and the second-generation lentiviral packaging plasmid psPAX2 (Addgene). Infectious supernatant was collected at 24 hours, concentrated by ultracentrifugation and used to infect Ph+ ALL cell lines or primary patient's cells by spinoculation. SUP-B15, BV173, and Z181 cell lines were transduced with the inducible shSTAT5 or shBIM lentivirus and selected in the presence of puromycin (3 μg/mL) or G418 (400 μg/mL) for 5 or 14 days, respectively. shRNAs (shSTAT5 or shBIM) were induced with doxycycline (2.5 μg/mL). Because of the limited viability of primary Ph+ ALL cells in liquid culture, these cells were transduced with the stable shSTAT5 or shSCR pLKO.1 lentivirus 16 hours after thawing, and selected in the presence of puromycin. Western blot lysates were obtained after 60 hours of puromycin selection.

For studies of PIM1-s overexpression, shSTAT5–BV173 cells were transduced with the PIM1-s lentivirus and selected by EGFP sorting, using a BD FACSAria II cell sorter.

For studies of MCL1 or BCL2 overexpression, shSTAT5–BV173 or shSTAT5–shBIM–BV173 cells were transduced with the MCL1 or BCL2 lentivirus and selected by EGFP or mCherry sorting respectively, using a BD FACSAria II cell sorter.

Cells were counted and lysed at a density of 10,000/μL in Laemmli Buffer supplemented with 5% β-Mercaptoethanol. Lysates where run on a 4% to 20% gradient polyacrylamide gel (Biorad) and transferred onto a nitrocellulose membrane (Fisher Scientific) using a semi-dry trans-blot transfer cell (Bio-Rad). Membranes were then blocked in 5% nonfat dry milk/TBS-T and incubated with the following primary antibodies: STAT5 (C-17; 1:4,000; Santa Cruz Biotechnology; #sc-835), p-STAT5 (Y649; 1:1,000; BD #611964), STAT3 (1:1,000; Transduction lab #610190), β-actin (8H10D10; 1:4,000; Cell Signaling Technology; #3700S) or β-actin (D6A8; 1:4,000; Cell Signaling Technology; #8457P), BCL2 (1:1,000; BD #610538), MCL1 (S-19; 1:1,000; Santa Cruz #sc-819), BCL-XL (1:1,000; Cell Signaling Technolgy; #2762S), PIM-1 (1:250; Cell Signaling Technology #2907S), p-BAD (S112) (40A9; 1:500 Cell Signaling Technology; #5284T), BAD (1:1,000; Santa Cruz; #sc-8044), p-4EBP1 (T37/46; 1:500; Cell Signaling Technology; #236B4), 4EBP1 1:1,000; Cell Signaling Technology; #9644P), p-MYC (phospho-S62; 1:1000; Abcam; #ab185656), c-MYC (N-262; 1:1000, Santa Cruz; #sc-768), BIM (1:1,000; Cell Signaling Technology; #2818S), NOXA (1:500; Novus Biological; #NB600-1159), PUMA (1:1,000; AbCam; #ab9643).

After incubation with HRP-conjugated secondary antibodies (Thermo Fisher Scientific), signals were visualized by chemiluminescent reaction using Supersignal West Pico or Dura Chemiluminescent Substrates (Thermo Fisher Scientific). When different antibodies were probed to the same nitrocellulose membrane, previous signals were removed by incubation with 0.5% sodium azide for 10 minutes at room temperature or by stripping in 62 mmol/L Tris-HCL pH 6.8, 2% SDS, 0.7% β-mercaptoethanol for 20 minutes at 50°C.

RNA was isolated from shSCR–BV173 and shSTAT5–BV173 cells, untreated or treated with doxycycline at 2.5 μg/mL, using the RNeasy Plus Mini Kit (Qiagen) and reverse-transcribed (2 μg) with the High-Capacity cDNA Reverse Transcription Kit (ThermoFisher Scientific). Quantitative PCR was performed with the QuantStudio 12k Flex (Life Technologies) instrument and QuantStudio 12K Flex software, using the following primers: MYC FW 5′-GTCACACCCTTCTCCCTTCG-3′; MYC RV 5′-ATGTCTCCTCCCAGCAGCTC-3′; TRIB3 FW 5′-TGCGTGATCTCCAAGCTGTGT-3′; TRIB3 RV 5′-GCTTGTCCCACAGGGAATCA-3′; ITGα5 FW 5′-GGCTTCAACTTAGACGCGGA-3′; ITGα5 RV 5′-GGCTGGCTGGTATTAGCCTT-3′; ITGβ1 FW 5′-CCGCGCGGAAAAGATGAATTT-3′; ITGβ1 RV 5′-CCACAATTTGGCCCTGCTTG-3′; PIM1 FW 5′-ACACGGACTTCGATGGGACC-3′; PIM1 RV 5′-GATGGTCTCAGGGCCAAGCA-3′; IDH2 FW 5′-GCCGGCACTTTCAAAATGG-3′T; IDH2 RV 5′-TCCTTGACACCACTGCCATC-3′; UBC FW 5′-GTCGCAGTTCTTGTTTGTGGATC-3′; UBC RV 5′-GTCTTACCAGTCAGAGTCTTCACGAAG-3′; GAPDH FW 5′-CCCATCACCATCTTCCAGGAG-3′; GAPDH RV 5′-CTTCTCCATGGTGGTGAAGACG-3′.

Animal experiments were approved by Thomas Jefferson University Institutional Animal Care and Use Committee under protocol 00012.

For leukemogenesis assays, 10⁶ leukemia cells (shSTAT5-cell lines or primary cells from patients with Ph+ ALL) were injected intravenously in 7- to 9-week-old NOD/SCID/IL-2Rγ^null mice (The Jackson Laboratory).

To induce STAT5 downregulation in vivo, mice were continuously treated with doxycycline (2 g/L) in D(+)-sucrose-supplemented (30 g/L) drinking water starting 72 hours post-cell injection.

IST5-002 was dissolved in 0.2% hydroxypropyl methylcellulose and administered intraperitoneally (50 or 100 mg/kg) daily for a total of 14 days.

Sabutoclax (5 mg/kg; ApexBio) was dissolved in 10% Kolliphor-EL (Sigma-Aldrich)-10% ethanol-80% PBS and administered intraperitoneally every other day for a total of seven doses (14 days).

AZD1208 (100 mg/kg; SelleckChem) was dissolved in 0.5% hydroxypropyl methylcellulose-0.1% Tween 80 and administrated by oral gavage every day for 14 days.

Leukemia formation was monitored by flow cytometry detection of the human CD19 antigen (by antibody #555415 from BD Biosciences) in peripheral blood obtained by retro orbital bleeding.

Data, expressed as mean ± SD of three experiments, were analyzed for statistical significance by unpaired, two-tailed Student t test. P < 0.05 was considered statistically significant. For drugs combination studies, combination indexes (CI) were calculated by the Chou–Talalay method (28) and synergism was defined as CI = 0.1–0.9.

Kaplan–Meier plots for mice survival experiments were generated using the GraphPad Prism 6.0 software. Differences in survival were assessed by log-rank test.

To investigate the role of STAT5 in Ph+ ALL, we transduced human Ph+ leukemia cell lines BV173, SUP-B15, and Z181 with a lentiviral plasmid carrying a doxycycline-inducible shSTAT5 (shSTAT5-Tet-pLKO) and assessed cell growth and survival upon doxycycline treatment.

Doxycycline treatment induced a decrease in STAT5 expression in each cell line (Fig. 1A). However, the effect was more pronounced in BV173 cells and became detectable after a 72 hours treatment due to the long half-life of the STAT5 protein (29). As control, levels of STAT3 were not affected by the doxycycline-induced shSTAT5 (Fig. 1A).

STAT5 silencing markedly reduced cell growth of all three lines, but the effect was more evident in BV173 cells (Fig. 1B), probably reflecting the near complete inhibition of STAT5 expression in these cells (Fig. 1A).

STAT5 silencing caused a marked increase in the apoptosis of BV173 and SUP-B15 cell lines measured by flow cytometry detection of caspase 3/7 activation (Fig. 1C). Apoptosis was detected at 72 hours and increased after a 96 and 120 h of doxycycline treatment, consistent with the kinetics of doxycycline-induced STAT5 downregulation. In BV173 cells, apoptosis induced by STAT5 silencing for 72 to 120 hours ranged from 46% to 95%, whereas in SUP-B15 cells, at the same time points, it ranged from 20% to 26%. Compared with control cells, induction of apoptosis in STAT5-silenced Z181 cells was not statistically significant.

The dependence of Ph+ ALL cell lines on STAT5 expression was also assessed by methylcellulose colony formation assays. These assays revealed that doxycycline-treated, STAT5-silenced BV173, SUP-B15, and Z181 cells were markedly less clonogenic than the untreated counterparts (Fig. 1D). Doxycycline treatment had no effect on STAT5 expression, apoptosis, or colony formation of shScramble-transduced Ph+ ALL cells (Supplementary Fig. S1A–S1C).

Next, we assessed STAT5 requirement for leukemia development in NOD/SCID-IL-2Rγnull (NSG) mice injected with Ph+ BV173, SUP-B15, or Z181 cells transduced with the doxycycline inducible shSTAT5 lentivirus. Mice were injected in the tail vein with 10⁶ cells and, starting at day 3, doxycycline was added to the drinking water while control mice were left untreated. Control mice injected with shSTAT5–BV173 or SUP-B15 cells died of leukemia (bone marrow heavily infiltrated by leukemic cells and splenomegaly) within 60 days of leukemic cell injection (median survival = 58 days). Mice injected with shSTAT5–BV173 cells and given doxycycline to induce STAT5 silencing survived up to 188 days with a median survival of 162 days; such increase in overall survival was statistically significant (P < 0.001; Fig. 1E). Mice injected with shSTAT5–SUP-B15 cells and treated with doxycycline survived up to 105 days with a median survival of 82 days, which is significantly longer of that of the untreated mice (median survival = 55 days; P < 0.001; Fig. 1E). Doxycycline treatment also induced a statistically significant increase in the survival of NSG mice injected with shSTAT5–Z181 cells (Fig. 1E). Untreated mice injected with shSTAT5–Z181 cells survived significantly longer (median survival = 102 days) than those injected with shSTAT5–BV173 or SUP-B15 cells; however, doxycycline-treated mice injected with shSTAT5–Z181 cells survived more than 180 days (Fig. 1E).

To further evaluate the requirement for STAT5 expression for growth of primary Ph+ ALL cells we utilized colony formation assays of patient-derived primary Ph+ ALL cells. For these assays, peripheral blood blast cells from four patients with Ph+ ALL, two newly diagnosed (ALL #004 and ALL #011) and two from relapsed ALL post-therapy with TKIs (ALL #5775 and ALL #3961) were transduced with a lentivirus constitutively expressing a STAT5 shRNA used to silence STAT5 expression in K562 cells (30), or a scramble shRNA and plated (10⁵ cells/plate) in methylcellulose in the presence of puromycin. As shown in Fig. 2A, STAT5 expression was downregulated in shSTAT5 lentivirus-transduced cells from each patient (left). Expression of STAT5-regulated BCL2 and MCL1 was also downregulated in STAT5-silenced samples, except for BCL2 in sample ALL #004. Importantly, STAT5-silenced cells from patient's samples were markedly less clonogenic than the scramble-transduced counterparts (Fig. 2A, right).

Of interest, suppression of colony formation and a marked decrease of BCL2 and MCL1 levels compared with the scramble-transduced counterpart was also observed in STAT5-silenced blast cells from a patient with Ph-like ALL (#005, Supplementary Table S2; Fig. 2B).

In silico screening of chemical structure databases has recently identified IST5-002 as a candidate small molecule inhibitor of STAT5 (31). Such activity was confirmed based on its ability to suppress JAK2 and BCR-ABL1-mediated STAT5 phosphorylation, STAT5A/B dimerization, and to inhibit the transcriptional activity of STAT5A and STAT5B (31).

Thus, we assessed if IST5-002 can mimic the ex vivo and in vivo effects of STAT5 silencing in Ph+ ALL cells. Treatment with IST5-002 markedly suppressed the growth of Ph+ cell lines, including the TKI-resistant T315I BV173 derivative (Supplementary Fig. S2). IST5-002-induced growth suppression was associated with markedly decreased STAT5 phosphorylation (Supplementary Fig. S2).

More importantly, treatment with IST5-002 also suppressed STAT5 phosphorylation and colony formation of leukemic blasts from eight patients with Ph+ ALL (Fig. 3A and B), including five samples (ALL #2534, ALL #5775, ALL #1577, ALL#27574, and ALL #3961) from patients with relapsed leukemia post-therapy with TKIs.

The Ph-like ALL cell lines MUTZ-5 and MHH-CALL-4 exhibit expression of phospho-STAT5 (Supplementary Fig. S3A) driven by the IGH/CRLF2 translocation and JAK2 mutation (32). Treatment with IST5-002 suppressed STAT5 phosphorylation and growth of these cell lines (Supplementary Fig. S3B) and STAT5 phosphorylation and colony formation of a primary Ph-like ALL sample (#005) also carrying the IGH/CRLF2 translocation (Supplementary Table S2; Supplementary Fig. S3C).

To assess whether IST5-002 treatment has detrimental effects on normal hematopoietic cells, we performed colony formation assays of cytokine-treated stem cell-enriched G-CSF–mobilized CD34⁺ cells, samples J48 and J50, from two healthy donors. Treatment with IST5-002 (5 or 10 μmol/L) suppressed cytokine-dependent STAT5 phosphorylation in both samples (Supplementary Fig. S4); however, IST5-002 had no effect on colony formation at the 5 μmol/L concentration and partially reduced colony formation in sample J50 only at the 10 μmol/L concentration (Supplementary Fig. S4).

Then, we tested whether treatment with IST5-002 suppressed leukemogenesis in NSG mice injected with Ph+ ALL cells. For these experiments, we used the Ph+ ALL SUP-B15 cell line and primary cells from a patient with a TKI-resistant Ph+ ALL with the T315I ABL1 kinase domain mutation.

In the SUP-B15 model, mice were treated with vehicle (10/group) only or with 50 mg/kg IST5-002 (6/group) for 14 consecutive days starting 7 days after cell injection. Vehicle-treated mice had a median survival of 56 days; by contrast, mice treated with IST5-002 survived up to 89 days, with a median survival of 68 days (P < 0.05; Fig. 3C).

In the TKI-resistant Ph+ ALL primary cells model, mice (four/group) were injected with 10⁶ cells/mouse and when peripheral blood CD19⁺ cells were 5% to 9% (6 weeks post cell injection) mice were treated intraperitoneally for 14 days with a daily dose of 100 mg/kg IST5-002 or with diluent only. Peripheral blood leukemia burden was assessed at the end of the treatment by flow cytometry analysis of CD19⁺ cells. Although vehicle-treated mice showed an increase in the percentage of CD19⁺ leukemia cells from 5% to 8% to about 20%, leukemia growth was essentially blocked in the IST5-002–treated mice (Fig. 3D).

These findings support the concept that STAT5 or STAT5-regulated pathways may serve as targets for the treatment of Ph+ ALL.

The growth suppression of Ph+ ALL cells induced by STAT5 silencing is largely caused by enhanced apoptosis (Fig. 1C). Thus, we assessed the expression of members of the BCL2 family, some of which were previously reported to be regulated by STAT5 (33–36); such analysis is a necessary first step to further investigate if there is a functional link between changes in the expression of BCL2 family proteins and apoptosis induced by STAT5 silencing. Compared with the untreated counterparts, doxycycline-treated shSTAT5-BV173, shSTAT5-SUP-B15, and shSTAT5-Z181 cells exhibited a decrease in the expression of anti-apoptotic BCL2 and MCL1 proteins (Fig. 4A). Decreased expression of anti-apoptotic BCL2 and MCL1 proteins was also observed in STAT5-silenced Ph+ primary ALL cells and the Ph-like primary sample #005 (Fig. 2).

Doxycycline-treated shSTAT5-BV173, shSTAT5-SUP-B15, and shSTAT5-Z181 cells also exhibited an increase in the expression of the BH3-only proapoptotic BIM protein, although such an increase was detected at 72 but not at 96 hours in Z181 cells (Fig. 4A). Expression of anti-apoptotic BCL-XL protein was unchanged (Fig. 4A).

To investigate whether increased BIM expression can explain the enhanced apoptosis of STAT5-silenced Ph+ ALL cells, we generated the shSTAT5-BV173 derivative cell line transduced with a doxycycline-regulated BIM shRNA lentivirus (shSTAT5–shBIM–BV173 line; Fig. 4B) and assessed apoptosis induced by STAT5 silencing. As shown in Fig. 4C and D and Supplementary Fig. S5, downregulation of BIM expression suppressed apoptosis induced by STAT5 silencing; however, a fraction of doxycycline-treated shSTAT5–shBIM BV173 cells were apoptotic at 96 hours (Fig. 4D), in spite of apparently complete inhibition of BIM expression (Fig. 4B). Next, we generated the shSTAT5–BCL2–BV173 or the shSTAT5–MCL1–BV173 derivative line (Fig. 4E, left) and assessed whether restoring BCL2 or MCL1 expression would also rescue the apoptosis of STAT5-silenced BV173 cells. In these lines, we assessed apoptosis only by Annexin V staining because GFP expression does not allow to measure active caspase 3/7. As shown in Figure 4C and D, expression of BCL2 was as effective as BIM silencing in inhibiting apoptosis induced by STAT5 silencing while the effect of MCL1 expression was negligible. Based on these findings, we generated the shSTAT5-shBIM-BCL2-BV173 and the shSTAT5–shBIM–MCL1–BV173 derivative lines (Fig. 4E, right) and asked whether combining BIM silencing with BCL2 or MCL1 expression would be more effective than BIM silencing or BCL2/MCL1 expression alone in blocking apoptosis induced by STAT5 silencing. As shown in Fig. 4C and D, combining BIM silencing with BCL2 or MCL1 expression rescued almost completely the apoptosis of STAT5-silenced BV173 cells.

In a previous study (31), we performed oligonucleotide microarray hybridization to generate the gene expression profile of STAT5-silenced Ph+ K562 cells. Thus, we assessed whether the expression of selected STAT5-regulated genes identified in K562 cells was also modulated in STAT5-silenced BV173 cells. As shown in Supplementary Fig. S6, expression of MYC, TRIB3, ITGB1, IDH2, ITGA5, PIM1 was markedly reduced in STAT5-silenced BV173 cells, compared with control cells. Then, we focused on STAT5-regulated PIM-1 because of: (i) its established oncogenic effects (35); (ii) commercially available compounds that inhibit PIM-1 and PIM-2 (37–39); (iii) the efficacy of the selective PIM kinase inhibitor AZD1208 in preclinical models of acute myeloid leukemia (40).

First, we assessed if PIM-1 protein levels were reduced in STAT5-silenced Ph+ ALL cell lines. doxycycline-treated shSTAT5 Ph+ cell lines all showed decreased expression of the short isoform of PIM-1 (PIM1-s; Supplementary Fig. S7A).

Expression of PIM1-s in blast cells from patients with Ph+ ALL showed sample-to-sample variation; however, it correlated well with levels of total and tyrosine phosphorylated STAT5 (Supplementary Fig. S7B).

Of interest, levels of PIM1 and STAT5A or STAT5B mRNAs were also correlated in a microarray dataset from patients with acute B-cell leukemia; in the Ph+ subset, only the STAT5A–PIM1 correlation was statistically significant (Supplementary Fig. S7C–S7F). However, assessing levels of phospho-STAT5 is likely to be more informative than evaluating mRNA levels in correlative studies (41).

Then, we asked whether restoring PIM-1s expression in STAT5 silenced BV173 cells would rescue the growth inhibition of these cells. PIM-1s overexpression did not enhance the overall growth of STAT5-silenced BV173 cells because apoptosis and BIM expression induced by STAT5 silencing was not rescued (Supplementary Fig. S8A and S8B).

These findings indicate that STAT5 activation downstream of BCR-ABL1 promotes growth and survival of Ph+ ALL cells through different effectors.

The role of the PIM-1 kinase in Ph+ ALL cells was assessed by pharmacologic inhibition with the pan-PIM inhibitor AZD1208 (40).

BV173, SUP-B15, and Z181 cell lines were treated with 3 μmol/L AZD1208 and PIM kinase-regulated pathways were analyzed by Western blotting (Fig. 5A).

Previous studies showed that PIM-1 stabilizes c-MYC expression through phosphorylation of Serine 62 (42, 43). Consistent with these findings, AZD1208-treated BV173, SUP-B15, or Z181 cells exhibited reduced levels of phospho (S62) MYC, but total levels of c-MYC were little changed (Fig. 5A).

Phosphorylation of BAD by PIM-1 at serine 112 is a mechanism through which BAD is sequestered in the cytoplasm in complex with 14-3-3 allowing mitochondrial BCL2 to suppress BAX-dependent apoptosis (44).

As expected, treatment of Ph+ ALL cell lines with AZD1208 led to a decrease in S112 phosphorylation with no changes in total levels of BAD (Fig. 5A).

The translation regulator eukaryotic elongation factor 4E-BP1 is also regulated by PIM kinases, directly at the Threonines 37/46 priming sites and indirectly via mTOR-dependent mechanisms, causing the dissociation from eukaryotic initiation factor 4, and promoting the activation of cap-dependent translation (45–47).

Multiple bands, indicative of hyperphosphorylated 4E-BP1, were detected in untreated BV173, SUP-B15, and Z181 cells (Fig. 5A). Treatment with AZD1208 induced a marked decrease of phospho (T37/46)-4E-BP1 expression in SUP-B15, BV173, and Z181 cells (Fig. 5A) but total 4E-BP1 levels did not change (Fig. 5A).

Consistent with the effect of AZD1208 in Ph+ ALL cell lines, AZD1208-treated blast cells from three patients with Ph+ ALL also showed decreased phosphorylation of BAD (S112), c-MYC (S62), and 4E-BP1 (T37/46; Fig. 5B). Levels of total BAD and 4E-BP1 remained unchanged (Fig. 5B)

Of interest, STAT5 silencing in BV173, SUP-B15, or Z181 cells had effects similar to those of AZD1208 treatment on the phosphorylation of c-MYC (S62), 4E-BP1 (T37/46), and BAD (S112; Fig. 5C), likely reflecting downregulation of PIM-1 levels. However, total levels of c-MYC were markedly reduced only after STAT5 silencing, suggesting that STAT5 regulates c-MYC expression through multiple pathways, including enhanced transcription (see Supplementary Fig. S6). Expression of PUMA and NOXA was not affected by STAT5 silencing in Ph+ ALL cell lines (Fig. 5C).

Together, these data indicate that PIM-1 might regulate cell growth of Ph+ ALL cells via activation of multiple pathways and suggest that pharmacologic inhibition of PIM-1 could suppress Ph+ ALL cell growth.

Changes in the expression of BCL2 and BIM appear to be essential for the apoptosis of STAT5-silenced Ph+ ALL cells (Fig. 4) and several growth-promoting pathways are suppressed in these cells by pharmacologic inhibition of STAT5-regulated PIM kinase (Fig. 5). Thus, we tested the effects of the PIM kinase inhibitor AZD1208 and the pan-BCL2 family inhibitor Sabutoclax in Ph+ ALL cells ex vivo and in NSG mice.

Sabutoclax functions as a BH3 mimetic that, like BIM, binds anti-apoptotic members of the BCL2 family (BCL2, BCL-XL, MCL1) blocking their anti-apoptotic activity (48, 49).

MTT assays revealed that treatment with AZD1208 or Sabutoclax suppressed the growth of BV173, SUP-B15, and Z181 cells, including the TKI resistant BV173 (T315I) cell line and that the combined treatment was more effective than either drug alone (Supplementary Fig. S9A–S9D), suggesting that these drugs target nonoverlapping pathways.

Although treatment with Sabutoclax alone strongly increased apoptosis in each Ph+ ALL cell line, treatment with AZD1208 alone did not. However, the AZD1208/Sabutoclax combination further increased apoptosis in BV173, BV173 (T315I), SUP-B15, and Z181 cells (Supplementary Fig. S9E–S9H). These findings suggest that suppressing PIM kinase signaling (e.g., phosphorylation of BAD) is insufficient to provide a strong apoptotic signal, but it enhanced apoptosis induced by the BCL2 family antagonist Sabutoclax.

These findings also suggest that the AZD1208/Sabutoclax combination may inhibit synergistically the growth of Ph+ ALL cells.

Thus, Ph+ ALL cell lines were treated with three doses of AZD1208 and/or Sabutoclax, and the effects on cell growth were analyzed by MTT assay (Supplementary Fig. S10A–S10D). Combination indexes (CI) were calculated using CompuSyn software and plotted in Supplementary Fig. S10E–S10H.

The combined treatment had synergistic effects at all drug concentrations except one in parental and TKI-resistant BV173 cells and in Z181 cells while the effect was predominantly additive in SUP-B15 cells (Supplementary Fig. S10E–S10H).

Interestingly, cotreatment with the two drugs also enhanced apoptosis of Ph-like MUTZ-5 and MHH-CALL-4 cell lines compared with treatment with AZD1208 or Sabutoclax alone (Supplementary Fig. S11).

To further analyze the efficacy of the combined AZD1208/Sabutoclax treatment on Ph+ ALL cells, we performed methylcellulose colony assays of drug-treated BV173, BV173 (T315I), SUP-B15 cell lines, and nine primary samples from patients with Ph+ ALL. These samples are from six newly diagnosed and three patients with relapsed/TKI-resistant ALL (sample #557 expresses the p210 isoform and carries the T315I ABL1 mutation; sample #3961 expresses the p190 isoform and carries the T315I mutation; and sample #11463 expresses the p190 isoform with no mutations detected in the TKD). Treatment with AZD1208 or Sabutoclax alone suppressed, with varying degrees, colony formation in each case, with the exception of Sabutoclax treatment of ALL samples #1539 and #3934, but the most marked effect was observed when the two drugs were used in combination (Fig. 6A and B).

Next, we assessed the effect of the AZD1208/Sabutoclax combination in leukemia progression in vivo. NSG mice were injected with Ph+ ALL #004, #536, or #557 (T315I) cells and, when peripheral blood CD19⁺ cells were 5% to 20%, mice were treated for 14 days with AZD1208 alone, Sabutoclax alone, or with both drugs. Then, peripheral blood leukemia burden was assessed; the AZD1208/Sabutoclax combination was more effective than either drug alone in suppressing Ph+ ALL, resulting in leukemia burdens lower than those observed before the treatment, in each Ph+ ALL sample (Fig. 7A and B).

In this study, we assessed the dependence of Ph+ ALL cells on the expression/activity of STAT5 and its downstream effectors and used such knowledge for targeted therapy of Ph+ ALL in patient-derived xenografts.

The role of STAT5 for in vivo maintenance of oncogenic Abl-driven ALL was previously investigated in p210-BCR-ABL1 or v-Abl-induced leukemia upon conditional deletion of STAT5A/B (18, 19, 50). STAT5 expression was found to be dispensable for p210-BCR-ABL1-driven B-cell leukemia in Balb/c mice whereas STAT5A/B deletion markedly suppressed v-Abl- and p190-BCR-ABL1-driven B-cell leukemia in C57BL/6 mice (18, 19, 50). Whether such outcomes reflect differences in genetic background of recipient mice and/or biological properties of p210-BCR-ABL1 versus v-Abl-transformed B-cell precursors is, at the moment, unclear.

Despite the potential relevance of STAT5 in BCR-ABL1-driven leukemia, the role of STAT5 expression has not been previously assessed in human Ph+ ALL cells. Using Ph+ ALL cell lines expressing a doxycycline-regulated shSTAT5 that inhibits the expression of both STAT5 isoforms, we show here that downregulation of STAT5 expression leads to markedly decreased proliferation and colony formation, induction of apoptosis, and significantly prolonged survival of NSG mice injected with Ph+ ALL cells. Of interest, FACS-sorted CD19+ BV173 cells isolated from the bone marrow of a mouse that developed leukemia, in spite of being continuously treated with doxycycline, expressed STAT5, suggesting that re-expression of STAT5 caused the re-growth of Ph+ ALL.

STAT5-silenced primary Ph+ ALL cells from patients with newly diagnosed or relapsed/TKI-resistant disease were markedly less clonogenic than the empty vector-transduced counterpart.

Of interest, STAT5 silencing also suppressed colony formation from blast cells of a patient with IGH/CRLF2-positive Ph-like ALL.

Based on these findings, we assessed the reliance of Ph+ ALL cells on STAT5 activity by ex vivo and in vivo assays using IST5-002, a small molecule inhibitor of STAT5A/B tyrosine phosphorylation and dimerization, recently identified through structure-based in silico screening of compounds binding to the STAT5 SH2 domain (31).

Such STAT5 inhibitor was highly effective in suppressing colony formation and proliferation of Ph+ and Ph-like ALL cells while sparing a large fraction of normal stem cell-enriched CD34⁺ hematopoietic cells, in spite of suppressing STAT5 phosphorylation very effectively also in these cells. The in vivo effects of IST5-002 were more modest of those induced by STAT5 silencing; since it is unclear whether any of the commercially available STAT5 inhibitors, including IST5-002, will be ever used in clinical trials, we investigated if Ph+ ALL cells are dependent on STAT5-regulated pathways that may provide alternative targets for the therapy of Ph+ ALL. Targeting such pathways should phenocopy, in part, the effects induced by STAT5 silencing or pharmacologic inhibition.

We found that the marked apoptosis induced by silencing STAT5 expression in Ph+ BV173 cells was largely dependent on downregulation of BCL2 or upregulation of proapoptotic BIM while modulating MCL1 expression had negligible effects. Although these findings suggest that levels of BCL2 and BIM are critical for STAT5-regulated survival of BV173 cells, other Ph+ ALL cell lines and patients'-derived primary ALL cells may have different requirements for STAT5-regulated BCL2 family members. In particular, this would not be surprising based on our findings that levels of BCL2 family members exhibit wide variations in Ph+ ALL cells (51).

The growth and colony formation of Ph+ ALL cells, including those derived from blast cells of newly diagnosed or relapsed/TKI-resistant Ph+ ALL patients, was also markedly suppressed by pharmacologic inhibition of the STAT5-regulated PIM1 kinase. However, restoring PIM1 expression was insufficient to rescue the impaired growth of STAT5-silenced BV173 cells. These findings suggest that the anti-apoptotic signals that may be mediated by the PIM1 kinase through phosphorylation of the BH3-only protein BAD cannot compensate for the increased expression of BIM protein induced by STAT5 silencing.

However, PIM kinase inhibition suppressed growth-promoting pathways dependent on c-MYC expression and 4EBP-1 phosphorylation (42, 45–47), likely explaining the ex vivo growth inhibitory effects of AZD1208 in Ph+ ALL cells.

Given that that Ph+ ALL cells rely for their growth on the expression/activity of STAT5-regulated effectors that can be targeted pharmacologically, we assessed the growth of Ph+ ALL cells ex vivo and in NSG mice, upon treatment with the PIM kinase inhibitor AZD1208 and the pan-BCL2 inhibitor Sabutoclax.

In most Ph+ ALL samples, including three samples from patients with relapsed/TKI-resistant disease, cotreatment with Sabutoclax and AZD1208 had synergistic or additive growth-suppressive effects compared with the treatment with single agents; these findings are not surprising since AZD1208 and Sabutoclax inhibit non-overlapping pathways regulated, at least in part, by STAT5 in Ph+ ALL cells.

The additive growth-suppressive effect of the AZD1208/Sabutoclax combination was also detected in vivo by leukemogenesis assays in NSG mice injected with three Ph+ ALL primary samples, including sample #557, which carries the TKI-resistant T315I mutation.

Although these assays are not directly comparable, it should be noted that in vivo growth suppression by treatment with AZD1208 or Sabutoclax alone did not correlate exactly with data from colony formation assays. In particular, leukemia load was markedly suppressed by treatment with Sabutoclax alone whereas treatment with AZD1208 alone was effective only in one of three Ph+ ALL samples, in contrast to the statistically significant inhibition of colony formation induced by AZD1208 in eight of nine samples.

Although these data point to the need of performing additional studies to assess the extent the which treatment with AZD1208 suppresses Ph+ ALL growth ex vivo, they support the main finding of our in vivo studies, which is the marked suppression of leukemia load induced by the AZD1208/Sabutoclax combination.

In summary, our data indicate that targeting STAT5 or STAT5-regulated pathways by pharmacologic approaches might provide an alternative treatment for patients with Ph+ ALL, especially those with relapsed or TKI-resistant disease.

No potential conflicts of interest were disclosed.

Conception and design: V. Minieri, B. Calabretta

Development of methodology: V. Minieri, M. De Dominici, S.A. Mariani, M.T. Nevalainen

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): V. Minieri, M. De Dominici, P. Porazzi, O. Spinelli, A. Rambaldi

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): V. Minieri, M. De Dominici, P. Porazzi, O. Spinelli, P. Porcu, B. Calabretta

Writing, review, and/or revision of the manuscript: V. Minieri, P. Porazzi, A. Rambaldi, P. Porcu, M.T. Nevalainen, B. Calabretta

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L.F. Peterson

Study supervision: V. Minieri, B. Calabretta

This work was supported, in part, by NCI grant RO1-CA167169 to B. Calabretta, RO1-CA113580 to M.T. Nevalainen, and R21CA178755 to M.T. Nevalainen. We thank Dr. C.M. Eischen for critically reviewing the article. We thank Dr. Z. Jagani (Novartis) for kindly providing the Tet-pLKO-Neo shBIM lentiviral vector, Dr. M. Carroll from the Stem Cell and Xenograft Core of the University of Pennsylvania for providing primary Ph+ ALL samples, and Dr. N. Flomenberg and the Bone Marrow Transplantation unit at Thomas Jefferson University for providing CD34⁺ human hematopoietic progenitor cells.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.