BCR-ABL1 promotes leukemia by converting p27 into a cytoplasmic oncoprotein

Abstract

Recent studies have revealed that p27, a nuclear cyclin-dependent kinase (Cdk) inhibitor and tumor suppressor, can acquire oncogenic activities upon mislocalization to the cytoplasm. To understand how these antagonistic activities influence oncogenesis, we dissected the nuclear and cytoplasmic functions of p27 in chronic myeloid leukemia (CML), a well-characterized malignancy caused by the BCR-ABL1 tyrosine kinase. p27 is predominantly cytoplasmic in CML and nuclear in normal cells. BCR-ABL1 regulates nuclear and cytoplasmic p27 abundance by kinase-dependent and -independent mechanisms, respectively. p27 knockdown in CML cell lines with predominantly cytoplasmic p27 induces apoptosis, consistent with a leukemogenic role of cytoplasmic p27. Accordingly, a p27 mutant (p27(CK-)) devoid of Cdk inhibitory nuclear functions enhances leukemogenesis in a murine CML model compared with complete absence of p27. In contrast, p27 mutations that enhance its stability (p27(T187A)) or nuclear retention (p27(S10A)) attenuate leukemogenesis over wild-type p27, validating the tumor-suppressor function of nuclear p27 in CML. We conclude that BCR-ABL1 kinase-dependent and -independent mechanisms convert p27 from a nuclear tumor suppressor to a cytoplasmic oncogene. These findings suggest that cytoplasmic mislocalization of p27 despite BCR-ABL1 inhibition by tyrosine kinase inhibitors may contribute to drug resistance, and effective therapeutic strategies to stabilize nuclear p27 must also prevent cytoplasmic mislocalization.

Figure 1

Effect of BCR-ABL1 kinase inhibition and kinase-inactive BCR-ABL1 on subcellular localization of p27 in CML cells. (A) CML cell lines (K562, KYO-1, and KCL-22) and CD34⁺ cells from CML patients (N = 11) in CP (N = 8) or BP (N = 3) were treated with imatinib for 16 hours. Cell lines were serum starved in 1% serum and primary samples were cultured in cytokine-free 20% BIT media before performing the experiments. Cytoplasmic (C) and nuclear (N) proteins were fractionated and subjected to immunoblot analysis, using Sp1 and α-tubulin distribution to monitor the purity of the nuclear and cytoplasmic fractions, respectively. Representative immunoblots are shown. Densitometry was performed by normalizing nuclear and cytoplasmic p27 protein levels to levels of Sp1 and α-tubulin, respectively. Expression of untreated cytoplasmic p27 levels was set to 100%. Values represent mean ± SD from 3 independent experiments. *P < .050. (B) Subcellular localization of p27 was analyzed by immunofluorescence (IF) microscopy in CD34⁺ cells from BM of normal donors or CML patients. Left panel, cells were freshly isolated; right panel, cells were treated with imatinib in cytokine-free 20% BIT media. Representative experiments are shown. Scale bars represent 10 μm. (C) p27^−/− MEFs were stably cotransduced with RFP-p27 (wild type) and GFP retroviral vectors containing native BCR-ABL1, the kinase-inactive BCR-ABL1^K271R mutant, or empty vector controls. Cells engineered to express native BCR-ABL1 were also treated with 2.5 μM imatinib for 16 hours. Double-positive (GFP⁺/RFP⁺) cells were sorted by FACS and the subcellular localization of p27 was visualized by IF using an Alexa Fluor-647–conjugated secondary antibody. Coexpression of GFP and RFP was verified under the microscope. DAPI was used for nuclear staining. Scale bars represent 20 μm. Quantification of nuclear/cytoplasmic p27 signal intensity is shown in supplemental Figure 3. BIT, bovine serum albumin, insulin, and transferrin; DAPI, 4,6 diamidino-2-phenylindole; GFP, green fluorescent protein; RFP, red fluorescent protein.

Figure 2

Cytoplasmic p27 promotes survival of CML cells. (A) p27 immunoblotting was performed on the nuclear and cytoplasmic fractions of exponentially growing CML cell lines. K562 and KYO-1 cells show predominantly cytoplasmic p27, whereas Mo7ep210 cells show predominantly nuclear p27. (B) K562, KYO-1, and Mo7ep210 cells were infected with p27 or scrambled shRNA lentivirus and total p27 expression was determined by immunoblot analysis. (C) Cells infected with p27 or scrambled shRNA were analyzed for apoptosis using Annexin V staining. *P < .050; **P < .010. (D) Effect of cytoplasmic-only p27 (p27^ΔNLS) on cell growth of CML cells. p27^−/− MEFs were transduced with Flag-p27^ΔNLS and localization was validated by IF analysis using anti-Flag antibody (M2; Sigma-Aldrich). DAPI was used for nuclear staining. K562 and Mo7ep210 cells were transduced with p27^ΔNLS or empty vector control and the effect on cell growth was measured over time.

Figure 3

The net function of p27 in BCR-ABL1–mediated leukemogenesis is tumor-suppressive. (A) BM harvested from 5-fluorouracil–treated p27^+/+, p27^+/−, and p27^−/− mice (N = 3 per group) was analyzed by FACS to determine the proportions of Lin⁻/c-Kit⁺/Sca1⁺ (LKS) cells. The LKS population was gated for CD34 expression to distinguish between progenitor cells (LKS CD34⁺ for MPP) and the stem cell–enriched fraction (LKS CD34⁻ for HSC). (B) Homing capacity was assessed in recipients of BM LKS cells from p27^+/+, p27^+/−, and p27^−/− mice (N = 3 per group) using CFSE labeling. (C) Long-term engraftment was assessed in CD45.1⁺ recipients of CD45.2⁺ BM cells from p27^+/+, p27^+/− and p27^−/− mice (N = 3 per group) after injecting all mice with MNCs (3 × 10⁵ cells per mouse) and LKS cells (3000 cells per mouse). Starting 4 weeks after transplantation, the contribution of CD45.2⁺ cells in the peripheral blood was measured by weekly FACS. The data shown represent engraftment after 6 weeks. (D) BCR-ABL1–expressing p27^+/+ and p27^−/− BM cells were subjected to an in vivo competition experiment. To adjust for the 50% reduced LKS population in p27^−/− mice, twice the number of BCR-ABL1–transduced BM cells from p27^−/− CD45.2 mice (1.4 × 10⁵ cells per mouse) mixed with BCR-ABL1–transduced BM cells from p27^+/+ CD45.2/CD45.1 mice (0.7 × 10⁵ cells per mouse) were injected into lethally irradiated wild-type congenic recipient mice (CD45.1). At the time of autopsy, BM cells were analyzed by FACS for the presence of GFP⁺ p27^−/− CD45.2 or p27^+/+ CD45.1 cells. (E-F) p27^+/+ mice were transplanted with BCR-ABL1–transduced GFP⁺ BM LKS cells (5000 or 500 per mouse) from p27^+/+, p27^+/− and p27^−/− mice and compared for survival using Kaplan-Meier statistics. (G-H) Spleen and liver weights of mice transplanted with BCR-ABL1–transduced LKS cells were compared according to genotype. (I) Representative histological sections of BM, liver, lung, and spleen from mice transplanted with BCR-ABL1–transduced LKS cells. Scale bars represent 100 μm *P < .050; **P < .010; **P < .001. CFSE, carboxyfluorescein succinimidyl ester; HSC, hematopoietic stem cells; MNC, mononuclear cell; MPP, multipotent progenitor.

Figure 4

Stabilization of p27 in the nucleus attenuates BCR-ABL1–induced leukemia. (A) BM cells from p27^+/+ and p27^T187A mice were transduced with BCR-ABL1 retrovirus, sorted by FACS for GFP⁺/Lin⁻ cells, and plated in methylcellulose in the presence or absence of cytokines. Myeloid colony formation of BCR-ABL1–transduced cells was assessed after 8 days. (B) p27^+/+ mice were transplanted with BCR-ABL1–transduced BM cells from p27^+/+ and p27^T187A mice and survival was analyzed using Kaplan-Meier statistics. (C) WBC and PLT counts and (D) spleen and liver weights of leukemic mice were compared according to genotype. (E) Lineage distributions of GFP⁺ cells in the bone marrow and spleen of leukemic mice: granulocytes (Gr1⁺CD11b⁺), B cells (CD19⁺), T cells (CD3⁺), and LKS cells (Lin⁻c-Kit⁺Sca1⁺). Error bars represent standard deviation. (F) Representative histological sections of BM, liver, lung, and spleen from mice transplanted with BCR-ABL1–transduced BM cells from p27^+/+ and p27^T187A mice. Scale bars represent 100 μm. In the above experiments, *P < .050, **P < .010. PLT, platelet.

Figure 5

Retention of p27 in the nucleus attenuates BCR-ABL1–induced leukemia. (A) BM cells from p27^+/+ and p27^S10A mice were transduced with BCR-ABL1 retrovirus, sorted by FACS for GFP⁺/Lin^– cells, and plated in methylcellulose in absence of cytokines. Myeloid colony formation of BCR-ABL1–transduced cells was assessed after 8 days. (B) Kaplan-Meier survival analysis of p27^+/+ mice transplanted with BCR-ABL1–transduced BM cells from p27^+/+ and p27^S10A mice. (C) WBC and PLT counts and (D) spleen and liver weights of leukemic mice were compared according to genotype. (E) Lineage distributions of GFP⁺ cells in the BM and spleen of leukemic mice: granulocytes (Gr1⁺CD11b⁺), B cells (CD19⁺), T cells (CD3⁺), and LKS cells (Lin⁻c-Kit⁺Sca1⁺). Error bars represent standard deviation. (F) Representative histological sections of BM, liver, lung, and spleen from mice transplanted with BCR-ABL1–transduced BM cells from p27^+/+ and p27^S10A mice. Scale bars represent 100 μm. In above experiments, *P < .050, **P < .010.

Figure 6

Cytoplasmic p27 promotes BCR-ABL1–induced leukemia. (A) BM cells from p27^−/− and p27^CK− mice were transduced with BCR-ABL1 retrovirus, sorted by FACS for GFP⁺/Lin⁻ cells and plated in methylcellulose in absence of cytokines. Myeloid colony formation of BCR-ABL1–transduced cells was assessed after 8 days. (B) Kaplan-Meier survival analysis of p27^+/+ mice transplanted with BCR-ABL1–transduced BM cells from p27^−/− and p27^CK− mice. (C) WBC and PLT counts and (D) spleen and liver weights of leukemic mice were compared according to genotype. (E) Lineage distributions of GFP⁺ cells in the BM and spleen of leukemic mice: granulocytes (Gr1⁺CD11b⁺), B cells (CD19⁺), T cells (CD3⁺), and LKS cells (Lin⁻c-Kit⁺Sca1⁺). Error bars represent standard deviation. (F) Representative histological sections of BM, liver, lung, and spleen from mice transplanted with BCR-ABL1–transduced BM cells from p27^−/− and p27^CK− mice. Scale bars represent 100 μm. In above experiments, **P < .010.

Figure 7

Schematic describing the function of p27 in CML. (A) p27 distribution in CML cells. BCR-ABL1 degrades nuclear p27 in a kinase-dependent manner and promotes its cytoplasmic localization in a kinase-independent manner. (B) Complete loss of p27 (p27-null mutant). The absence of p27 promotes leukemogenesis, consistent with a net tumor-suppressive function of p27. (C) Nuclear stabilization of p27 (p27^T187A mutant). Stabilization of nuclear p27 as a result of the p27^T187A mutation only moderately attenuates leukemia because the oncogenic effect of cytoplasmic p27 persists. (D) Decreased p27 export to the cytoplasm (p27^S10A mutant). Reduced cytoplasmic p27 and increased nuclear p27 retention due to the p27^S10A mutation significantly reduces leukemogenesis of p27^S10A mice. (E) Functional cytoplasmic and nonfunctional nuclear p27 (p27^CK− mutant). The presence of functional cytoplasmic p27 and the complete absence of functional nuclear p27 in leukemic cells as a result of the p27^CK− mutation promote leukemogenesis. As we do not provide direct evidence for trafficking of nuclear p27 to the cytoplasm, a dashed line was used to suggest nuclear-cytoplasmic shuttling of p27.