Ruxolitinib-induced defects in DNA repair cause sensitivity to PARP inhibitors in myeloproliferative neoplasms

Abstract

Myeloproliferative neoplasms (MPNs) often carry JAK2(V617F), MPL(W515L), or CALR(del52) mutations. Current treatment options for MPNs include cytoreduction by hydroxyurea and JAK1/2 inhibition by ruxolitinib, both of which are not curative. We show here that cell lines expressing JAK2(V617F), MPL(W515L), or CALR(del52) accumulated reactive oxygen species-induced DNA double-strand breaks (DSBs) and were modestly sensitive to poly-ADP-ribose polymerase (PARP) inhibitors olaparib and BMN673. At the same time, primary MPN cell samples from individual patients displayed a high degree of variability in sensitivity to these drugs. Ruxolitinib inhibited 2 major DSB repair mechanisms, BRCA-mediated homologous recombination and DNA-dependent protein kinase-mediated nonhomologous end-joining, and, when combined with olaparib, caused abundant accumulation of toxic DSBs resulting in enhanced elimination of MPN primary cells, including the disease-initiating cells from the majority of patients. Moreover, the combination of BMN673, ruxolitinib, and hydroxyurea was highly effective in vivo against JAK2(V617F)⁺ murine MPN-like disease and also against JAK2(V617F)⁺, CALR(del52)⁺, and MPL(W515L)⁺ primary MPN xenografts. In conclusion, we postulate that ruxolitinib-induced deficiencies in DSB repair pathways sensitized MPN cells to synthetic lethality triggered by PARP inhibitors.

Figure 1.

Sensitivity of JAK2(V617F)⁺, CALR(del52)⁺, and MPL(ex10mut)⁺cells to PARPi. (A) Cell lines expressing JAK2(V617F)+EpoR, CALR(del52)+MPL(wt), or MPL(W515L) were incubated with olaparib alone (1.25, 2.5, 5.0 μM) (squares) or BMN673 alone (12.5, 25.0, 50.0 nM) (squares), olaparib plus 200 μM vitamin E (circles), or olaparib plus 2.5 μM SCH51344 (triangles) for 96 hours in the presence of IL-3 plus Epo. Parental cells (diamonds) were incubated with olaparib or BMN673 only. Living cells were counted in Trypan blue. Results represent mean plus or minus SD percentage of living cells in comparison with untreated control from 3 independent experiments. (B) Western analysis of the indicated proteins in parental cells (P) and in isogenic cells expressing JAK2(V617F)+EpoR, CALR(del52)+MPL(wt), and MPL(W515L).

Figure 2.

Sensitivity of individual MPN samples expressing JAK2(V617F), CALR(del52), and MPL(ex10mut) to PARPi. Lin⁻CD34⁺ primary cells from (A) healthy donors (n = 3) and from (B) JAK2(V617F)⁺, (C) CALR(del52)⁺, (D) MPL(ex10mut)⁺ MPN patients were incubated with olaparib (1.25, 2.5, 5.0 μM) or BMN673 (12.5, 25.0, 50.0 nM) for 96 hours in the presence of growth factors (100 ng/mL SCF; 10 ng/mL Flt3 ligand; 20 ng/mL IL-3, IL-6, granulocyte colony-stimulating factor, and GM-CSF; 12 U/mL Epo; 2.5 ng/mL thrombopoietin) followed by plating in methylcellulose. Colonies were counted after 7 to 10 days. Results represent the percentage of colonies in comparison with untreated control.

Figure 2.

Sensitivity of individual MPN samples expressing JAK2(V617F), CALR(del52), and MPL(ex10mut) to PARPi. Lin⁻CD34⁺ primary cells from (A) healthy donors (n = 3) and from (B) JAK2(V617F)⁺, (C) CALR(del52)⁺, (D) MPL(ex10mut)⁺ MPN patients were incubated with olaparib (1.25, 2.5, 5.0 μM) or BMN673 (12.5, 25.0, 50.0 nM) for 96 hours in the presence of growth factors (100 ng/mL SCF; 10 ng/mL Flt3 ligand; 20 ng/mL IL-3, IL-6, granulocyte colony-stimulating factor, and GM-CSF; 12 U/mL Epo; 2.5 ng/mL thrombopoietin) followed by plating in methylcellulose. Colonies were counted after 7 to 10 days. Results represent the percentage of colonies in comparison with untreated control.

Figure 3.

JAK2i ruxolitinib reduced HRR and D-NHEJ activity and enhanced the anti-MPN effect of PARPi olaparib. (Panel A) Parental cell lines and those expressing JAK2(V617F)+EpoR, CALR(del52)+MPL(wt), or MPL(W515L) were untreated (C) or treated with 5 μM olaparib (O) or 400 nM ruxolitinib (R) or ruxolitinib plus olaparib (R+O) in the presence of IL-3 plus Epo for 24 hours (γ-H2AX) and 96 hours (cell survival). DSBs were detected by γ-H2AX immunofluorescence overlapping with 4′,6-diamidino-2-phenylindole (DAPI) (top panel), and living cells were counted in Trypan blue (bottom panel; percentage of living cells in comparison with untreated control). Results represent means plus or minus SD from 3 independent experiments. *P < .05 in comparison with C using the Student t test; **P ≤ .001 in comparison with R and O groups using the response additivity approach. (Panel B) Western analysis of the indicated proteins in cells expressing JAK2(V617F)+EpoR, CALR(del52)+MPL and MPL(W515L), and in BaF3-EpoR cells (Parental) after 24-hour incubation with 400 nM ruxolitinib in the presence of IL-3 plus Epo. Proteins downregulated by ruxolitinib are in red boxes. (Panel C) HRR and D-NHEJ activities in JAK2(V617F)⁺ cells untreated (−) or treated for 24 hours with 400 nM ruxolitinib (+). Top panel, Western blots, Middle panel, D-NHEJ activity. S indicates linearized plasmid substrate; P indicates ligated plasmid products; results show the percentage of P. Bottom panel, HRR activity measured by restoration of EGFP expression. Results show the percentage of GFP⁺ cells; *P ≤ .01. (Panel D) Number of proliferating Lin⁻CD34⁺CD38⁻CTV^low and quiescent Lin⁻CD34⁺CD38⁻CTV^max cells from individual JAK2(V617F)⁺ MPN samples left untreated (C) or treated with ruxolitinib (R; 25 nM), olaparib (O; 1.25 μM), and ruxolitinib plus olaparib (R+O). (Panel E) Cumulative percentages from samples examined in panel D. *P < .001 in comparison with R or O groups using the Student t test; **P < .01 in comparison with R and O groups using the response additivity approach.

Figure 4.

The effect of ruxolitinib on the sensitivity of JAK2(V617F), CALR(del52), and MPL(ex10mut) MPN cells to PARPi. Lin⁻CD34⁺ cells from PV, ET, and PMF patients carrying (A) JAK2(V617F), (B) CALR(del52), and (C) MPL(ex10mut) were incubated with olaparib (O; 1.25 μM), hydroxyurea (H; 10 μM), and/or ruxolitinib (R; 25 nM) for 72 hours in the presence of growth factors (see Figure 2) and plated in methylcellulose. Colonies were counted after 7 to 10 days. Results represent mean number of colonies plus or minus SD from triplicates. (D) Ruxolitinib (R)-treated Lin⁻CD34⁺ cells from cohorts of MPN samples (steel blue bars) and healthy donors (gray bars) displayed heterogenic sensitivity to PARPi. *P < .05 in comparison with O, and **P < .05 in comparison with HO using the Student t test.

Figure 5.

BMN673 exerted an anti-MPN effect in vivo. (Panel A) Experimental model. Lethally irradiated C57BL/6 recipient mice were injected with a 1:1 mixture of 10⁶ GFP+JAK2(V671F) and 10⁶ wild-type bone marrow cells. Five weeks later, mice were treated with vehicle (C), hydroxyurea (H; 30 mg/kg twice daily IP), ruxolitinib (R; 30 mg/kg twice daily oral gavage), BMN673 (B; 0.33 mg/kg IV), H+R, H+B, R+B, and H+R+B for 3 weeks. Percentage of GFP+JAK2(V617F) was measured in (panel B) bone marrow cells, (panel C) splenocytes, and (panel D) peripheral blood leukocytes; (panel E) number of GFP+JAK2(V617F) Lin⁻Sca1⁺c-Kit⁺ (LSK) cells per 10⁶ bone marrow cells was calculated, too. *P < .05, **P < .05, and ***P < .05 when compared with control, single treatment, and double treatment, respectively, from 6 to 7 mice using the Student t test.

Figure 6.

BMN673 exerted an anti-MPN xenograft effect in vivo. (Panel A) Experimental model. Sublethally irradiated NRGS recipient mice were injected with 10⁶ primary Lin⁻ MPN cells from individual MPN patients expressing JAK2(V617F), CALR(del52), or MPL(W515L). One week later, mice were treated with vehicle (C), hydroxyurea (H; 30 mg/kg twice daily IP) plus ruxolitinib (R; 30 mg/kg twice daily oral gavage) (HR), BMN673 (B; 0.33 mg/kg IV), and HR+B (HRB) for 3 weeks (3-5 mice per group). Indicated cells were detected by immunofluorescence. (Panel B) Percentage of hCD45⁺ cells was measured in peripheral blood leukocytes (PBLs), splenocytes (SPLs), and bone marrow cells (BMCs). Number of hCD45⁺ BMCs expressing Lin⁻CD34⁺ and Lin⁻CD34⁺CD38⁻ per 10⁶ cells was determined. *P < .05 and **P < .05 in comparison with C and all other groups, respectively, using the Student t test.