Protein tyrosine kinase 2b inhibition reverts niche-associated resistance to tyrosine kinase inhibitors in AML

Abstract

FLT3 tyrosine kinase inhibitor (TKI) therapy evolved into a standard therapy in FLT3-mutated AML. TKI resistance, however, develops frequently with poor outcomes. We analyzed acquired TKI resistance in AML cell lines by multilayered proteome analyses. Leupaxin (LPXN), a regulator of cell migration and adhesion, was induced during early resistance development, alongside the tyrosine kinase PTK2B which phosphorylated LPXN. Resistant cells differed in cell adhesion and migration, indicating altered niche interactions. PTK2B and LPXN were highly expressed in leukemic stem cells in FLT3-ITD patients. PTK2B/FAK inhibition abrogated resistance-associated phenotypes, such as enhanced cell migration. Altered pathways in resistant cells, assessed by nascent proteomics, were largely reverted upon PTK2B/FAK inhibition. PTK2B/FAK inhibitors PF-431396 and defactinib synergized with different TKIs or daunorubicin in FLT3-mutated AML. Midostaurin-resistant and AML cells co-cultured with mesenchymal stroma cells responded particularly well to PTK2B/FAK inhibitor addition. Xenograft mouse models showed significant longer time to leukemia symptom-related endpoint upon gilteritinib/defactinib combination treatment in comparison to treatment with either drug alone. Our data suggest that the leupaxin-PTK2B axis plays an important role in acquired TKI resistance in AML. PTK2B/FAK inhibitors act synergistically with currently used therapeutics and may overcome emerging TKI resistance in FLT3-mutated AML at an early timepoint.

Fig. 1. LPXN is induced during acquired midostaurin-resistance and is phosphorylated by PTK2B.

A MV4-11 cells were exposed to increasing concentrations of midostaurin for several weeks. Acquired midostaurin-resistance was analyzed by MTS assays. MV4-11 cells with approximately doubled IC₅₀ were termed MV4-11R early (left), MV4-11 cells with approximately 7-fold higher IC₅₀ are referred to as MV4-11R late (middle). Depicted are means from technical triplicates ± SD. Proteomics workflow scheme of MV4-11R vs WT cell analyses (right). B Total proteome analysis by mass spectrometry of MV4-11 parental and MV4-11R (early). Volcano plot of differentially expressed proteins in MV4-11R versus parental cells, statistically significant proteins are labelled in orange (log2FC >0.65, adj. p-value< 0.05). LPXN is highlighted in red. C Co-immunoprecipitation of endogenous LPXN in MV4-11R cells. GO pathway analysis of molecular function (upper) and cellular component (lower) of proteins interacting with LPXN in MV4-11R cells. D Co-immunoprecipitation of endogenous LPXN using two different LPXN antibodies (Abcam, LSBio) in MV4-11 WT, R, and OCI-AML3 cells. Co-purified proteins were analyzed by mass spectrometry. IP with IgG serving as control. Significantly enriched proteins/overlap in all three cell lines are shown. E HEK293T cells were transfected with plasmids encoding V5-LPXN, PTK2B, or empty vector. After immunoprecipitation of LPXN with a V5-tag antibody, tyrosine phosphorylation of LPXN was assessed by western blotting in the presence or absence of PTK2B co-expression. V5-LPXN and empty vector-transfected cells served as control. F Total proteome analyses in different AML cell lines: FLT3-mutated (MOLM-13, MV4-11) and FLT3 wildtype (Kasumi-1, HL-60, OCI-AML2, OCI-AML3). Left: Heatmap of LPXN and PTK2B expression across all cell lines. Right: Correlation of LPXN and PTK2B expression in the proteomic datasets of the different AML cell lines (Pearson r = 0.7680, p < 0.0001). G Kaplan–Meier plot for overall survival of patients with low and high PTK2B mRNA expression (TCGA LAML data set, UCSC Xena browser). Statistical significance was determined by log-rank test.

Fig. 2. Alterations in protein homeostasis upon midostaurin-resistance and PTK2B inhibition.

A Left: LPXN and PTK2B upregulation in MV4-11R (early) cells was confirmed by western blot. Data are representative for two independent experiments. Right: mRNA levels of LPXN and PTK2B in MV4-11 parental and MV4-11R (early) cells as analyzed by RT-qPCR. Depicted are means from triplicates ± SD. Statistical significance was assessed using unpaired two-tailed students t-test. **p = 0.0041. B Analyses of protein stability in leukemia cells. Outline (upper): MV4-11 parental and MV4-11R cells were labeled with SILAC medium (intermediate (IM) and heavy (HV)) for six passages. After switching back to light-medium, cells were harvested at different time points and HV/IM ratios were assessed by LC-MS/MS analysis over time to assess for differences in protein stability/degradation. Lower: Correlation of degradome with total proteome of MV4-11R (late) cells. Degradation slope was calculated as turnover ratio over the different harvesting time points and compared to total proteome (all 2043 proteins are labeled in grey, significant proteins (164) were labeled in black). Pearson correlation coefficient r and p-values are depicted. C Resistant/parental turnover ratios for LPXN and PTK2B at different harvesting time points, normalized to 1 at 0 h. D Nascent proteomics of MV4-11 parental and resistant cells. Left: schematic depiction of AHA-SILAC pulse labeling and enrichment of newly synthesized proteins. Middle: volcano plot of differentially translated proteins, significantly altered proteins are labelled in orange (log2FC >1, p-value < 0.05). Right: GSEA for proteins involved in extracellular matrix organization and collagen formation in MV4-11R vs WT cells. E Nascent proteomics of MV4-11R cells with or without PF-431396 treatment (16 h, 300 nM). Left: GSEA of changes in translation upon PTK2B inhibition. Right: GSEA for proteins involved in degradation of extracellular matrix organization and integrin-mediated cell adhesion in MV4-11R treated with 300 nM PF-431396 compared to MV4-11R cells with mock treatment (R). F Left: Comparison of up- and downregulated gene sets of nascent proteomics experiment of MV4-11R vs WT (grey) and MV4-11R+ inhibitor vs R (blue). PTK2B inhibitor treatment reverts major effects of R vs. WT on nascent proteome level. Right: Spearman correlation of nascent proteome of MV4-11R+ inhibitor vs R with MV4-11R vs WT.

Fig. 3. PTK2B/FAK inhibitor PF-431396 abolishes enhanced cell migration in midostaurin-resistant cells.

A Adhesion to ECM proteins in MV4-11 parental and resistant cells was assessed by fibronectin adhesion assays. Depicted are means from triplicates of three biological replicates ± SD. Statistical significance was assessed using unpaired two-tailed students t-test. ***p = 0.0006. BSA-coated wells served as control. B Adhesion assays were performed with MV4-11 parental cells which were treated with different concentrations of PF-431396 for 24 h prior to the assay at equal density. Untreated cells serving as control. Depicted are means from triplicates of at least two biological replicates ± SD. Statistical significance was assessed using unpaired two-tailed students t-test. ***p = 0.0005, **p = 0.003, ns = not significant. C Cell migration assays were performed with MV4-11 parental and MV4-11R cells which migrated for 24 h. After staining and washing, migratory cells were measured by cell counting (Supplementary Fig. S3A) and after extraction by colorimetry/optical density (upper). Depicted are means from duplicates of three biological replicates ± SD. Pictures were taken after staining of migratory cells (lower). Scale bars represent a length of 0.12 mm. Statistical significance was assessed using unpaired two-tailed students t-test. **p = 0.0043. D Cell migration assays were performed as in 3C with MV4-11 parental and resistant cells which were treated with 300 nM PF-431396. After staining and washing, migratory cells were measured by cell counting (Supplementary Fig. S3B) and colorimetry/optical density (upper). Depicted are means from duplicates of two biological replicates ± SD. Pictures were taken after staining of migratory cells (lower). Scale bars represent a length of 0.12 mm. Statistical significance was assessed using unpaired two-tailed students t-test. *p = 0.0250, ns not significant.

Fig. 4. PF-431396 and midostaurin are synergistic in FLT3-mutated cells, particularly in midostaurin-resistant cells.

A IC₅₀s for PF-431396 were determined by MTS assays in different FLT3-mutated (red: MV4-11, MOLM-13) and FLT3-wildtype (blue: HL-60, OCI-AML2, OCI-AML3) cells. Depicted are means from technical triplicates ± SD. Statistical significance was assessed using unpaired two-tailed students t-test. ***p = 0.0007. B Dose-response assays for MV4-11 WT and MV4-11R cells treated with midostaurin or daunorubicin combined with PF-431396 for 72 h. Viability was assessed by staining with MTS reagent. Depicted are means from technical triplicates ± SD. C Dose response matrix depicting Bliss scores for the midostaurin/PF-431396 and daunorubicin/PF-431396 combination. Bliss <0 indicates antagonism, Bliss = 0 indicates the two drugs act independent; Bliss >0 indicates synergy. Bliss scores were calculated from dose response assays from Fig. 4B. D Synergy between midostaurin and PF-431396, as well as daunorubicin and PF-431396 in MV4-11 WT and MV4-11R cells. Depicted are Bliss average synergy scores from dose-response assays performed with the same drug concentrations as in 4B and C. E Dose-response assays for MV4-11 WT and MV4-11R cells treated with midostaurin combined with PF-431396 for 72 h. Viability was assessed by staining with MTS reagent. Depicted are means from technical triplicates ± SD.

Fig. 5. Defactinib and gilteritinib act synergistically in FLT3-ITD mutated cell lines.

A IC₅₀s for defactinib were determined by MTS assays in FLT3-mutated (red: MV4-11, MOLM-13) and FLT3-wildtype (blue: Kasumi-1, OCI-AML2, OCI-AML3) cells. Depicted are means from technical triplicates ± SD. Statistical significance was assessed using unpaired two-tailed students t-test. ****p = <0.0001. B Colony forming unit assay of MV4-11 WT cells treated with 7.5 nM gilteritinib, 200 nM defactinib and the combination. Colonies were counted after eight days of culturing. Depicted are data from three biological replicates. Scale bars represent a length of 0.4 mm. Statistical significance was assessed using unpaired two-tailed students t-test. ****p < 0.0001. C Dose response matrix depicting Bliss scores for the gilteritinib/defactinib combination. Bliss scores were calculated from dose response assays shown in Figure S5A. D Bliss average synergy scores of MV4-11 WT and R, MOLM-13 WT and R, and HL-60 WT with different drug combinations. E Left: Confirmation of single and simultaneous knockout of FAK and PTK2B in MV4-11R cells by western blot. Right: Synergy assays of MV4-11R control, as well as single knockout (FAK and PTK2B) or simultaneous knockout (FAK + PTK2B). Cells were treated with gilteritinib and defactinib for 72 h. Depicted are Bliss average synergy scores calculated from dose response assays from three technical replicates of two biological replicates. Statistical significance was assessed using unpaired two-tailed students t-test. *p = 0.0102 ns = not significant. F Colony forming unit assay of MV4-11R control and simultaneous FAK and PTK2B knockout cells treated with 15 nM gilteritinib, 500 nM defactinib or the combination of both. Depicted are triplicates of two biological replicates (third biological replicate is shown in Supplementary Fig. S5C). Statistical significance was assessed using unpaired two-tailed students t-test. G Three FLT3-ITD mutated patient samples (red) and three FLT3-WT patient samples (blue) were treated with defactinib for 72 h and viability was measured by cell counting. Depicted are three technical replicates. Statistical significance was assessed using unpaired two-tailed students t-test. **p = 0.0039. H. Dose-response assays for FLT3-ITD mutated patient samples #4, #5, and #6 exposed to gilteritinib and defactinib for 72 h. Viability was assessed by cell counting and normalized to untreated cells. Depicted are means from technical triplicates ± SD. Statistical significance was assessed using unpaired two-tailed students t-test. Patient 4: ****p < 0.0001, *p = 0.0141, ***p = 0.0002; patient 5: ***p = 0.0004, *p = 0.0119, ***p = 0.0002; patient 6: **p = 0.0039, *p = 0.0116, ***p = 0.0007.

Fig. 6. PTK2B/FAK inhibitors and TKIs synergize in AML-niche models and in-vivo.

A Workflow scheme co-culture assay with MV4-11 WT and HS-5 stroma cells. B Dose response assays for MV4-11 WT cells cultured alone or together with HS-5 and treated with 15/25 nM midostaurin or 200/300 nM PF-431396 or the combination. Depicted are technical triplicates of four independent biological replicates for the first combination and technical duplicates of three biological replicates for the second combination. Statistical significance was assessed using unpaired two-tailed students t-test. ****p < 0.0001. C Workflow scheme of MV4-11R-derived xenograft experiments with gilteritinib/defactinib combination treatment or treatment with either drug alone. Vehicle only treated mice serving as control. D Leukemia burden and change over time in the different treatment groups of mice calculated from bioluminescence measurement. E Kaplan-Meier plot for time to leukemia symptom related endpoint. p-values were calculated by log-rank test.