Clonal evolution in tyrosine kinase inhibitor-resistance: lessons from in vitro-models

Abstract

Introduction: Resistance in anti-cancer treatment is a result of clonal evolution and clonal selection. In chronic myeloid leukemia (CML), the hematopoietic neoplasm is predominantly caused by the formation of the BCR::ABL1 kinase. Evidently, treatment with tyrosine kinase inhibitors (TKIs) is tremendously successful. It has become the role model of targeted therapy. However, therapy resistance to TKIs leads to loss of molecular remission in about 25% of CML patients being partially due to BCR::ABL1 kinase mutations, while for the remaining cases, various other mechanisms are discussed.

Methods: Here, we established an in vitro-TKI resistance model against the TKIs imatinib and nilotinib and performed exome sequencing.

Results: In this model, acquired sequence variants in NRAS, KRAS, PTPN11, and PDGFRB were identified in TKI resistance. The well-known pathogenic NRAS p.(Gln61Lys) variant provided a strong benefit for CML cells under TKI exposure visible by increased cell number (6.2-fold, p < 0.001) and decreased apoptosis (-25%, p < 0.001), proving the functionality of our approach. The transfection of PTPN11 p.(Tyr279Cys) led to increased cell number (1.7-fold, p = 0.03) and proliferation (2.0-fold, p < 0.001) under imatinib treatment.

Discussion: Our data demonstrate that our in vitro-model can be used to study the effect of specific variants on TKI resistance and to identify new driver mutations and genes playing a role in TKI resistance. The established pipeline can be used to study candidates acquired in TKI-resistant patients, thereby providing new options for the development of new therapy strategies to overcome resistance.

Keywords: KRAS; NRAS; PDGFRB; PTPN11; chronic myeloid leukemia; drug resistance; imatinib; nilotinib.

Figure 1

Analyses of genetic aberrations in TKI-resistant sublines. (A) Schematic representation of the generation of TKI-resistant cell lines in vitro. TKI-sensitive cells were exposed to an initial drug concentration (IM: 0,1 µM; N: 0.01 µM). When the cellular proliferation rate was restored, the drug concentration was stepwise increased (IM: 0.3, 0.5, 0.7. 1.0, 1.5 and 2 µM; N: 0.02, 0.05, 0.07 and 0.1 µM). (B) Overview of the TKI-resistant sublines used for the present study: Four imatinib-resistant sublines, resistant against low (0.5) and high (2 µM) imatinib, and two nilotinib-resistant sublines, resistant against low (0.05) and high (0.1 µM) nilotinib, were analyzed and compared to TKI-sensitive K-562 cells. (C) Analysis pipeline for the TKI-resistant cell lines. Using a coverage of >10%, the removal of SNVs already present in TKI-sensitive cells (VAF < 0.05) and removal of deep intronic SNVs, SNVs with a difference in the variant allele frequency (VAF) >15% between TKI-sensitive and resistant cell lines were obtained. The numbers indicate the SNVs clustered into variants acquired in highTKI/absent in lowTKI, variants with reduced VAF in highTKI and variants with constant or reduced VAF the high TKI-resistant cell lines compared to low TKI-resistant cells. (D) Total number of mutations in the TKI-resistant sublines. IM, imatinib; N, nilotinib; TKI, tyrosine kinase inhibitor; R, replicate.

Figure 2

Pathway networks in TKI resistance. (A) Network propagation of SNVs with ΔVAF>0.15 in the TKI-resistant sublines compared to TKI-sensitive cells. (B) Scaled enrichment scores of the gene set variation analysis (GSVA) in TKI resistance. Genome-wide gene expression data was obtained from Clariom S arrays, as well as the dataset GSE203442. IM, imatinib; N, nilotinib; R, replicate.

Figure 3

Identification of putative candidate driver variants in TKI resistance. (A) Oncoplot showing potential candidate driver mutations in the TKI-resistant cell lines obtained from association with a list of mutational oncogenes from Martínez-Jiménez et al. (23). Dark blue: missense mutations, light blue: truncating mutations, Red: ClinVar mutation, orange: multihit variants. (B, C) Proportion of SNVs in the TKI-resistant replicate cell lines shown as variant allele frequencies (VAFs). Red indicates ClinVar, orange multihit variants. IM, imatinib; N, nilotinib; R, replicate.

Figure 4

Effect of candidate variants NRAS p.(Gln61Lys), PTPN11 p.(Tyr279Cys) and PDGFRB p.(Glu578Gln) on the response to TKI treatment. (A) Graphical representation of the pathways affected by variants in the candidate genes NRAS/KRAS, PTPN11 (encoding SHP2), PDGFRB and KMT2D. (B–D) Top left: Western Blot of successful transfection of wild-type (WT) and variant into TKI-sensitive K-562 cells compared to GAPDH. Cellular fitness after WT and variant transfection and 48 h nilotinib exposure (0.1 µM) for (B) NRAS WT and p.(Gln61Lys), as well as imatinib exposure (2 µM) for (C) PTPN11 WT and p.(Tyr279Cys) and (D) PDGFRB WT and p.(Glu578Gln). Top right: Total cell number analyzed using trypan blue staining. Bottom left: Metabolic activity measured by WST assay. Bottom middle: Caspase 9 activity analyzed by caspase 9-Glo assay. Bottom right: Ki-67 expression to investigate cellular proliferation. Data was normalized to respective negative control (NC) and analyzed using Two-way ANOVA followed by Dunnett’s test. N = 3. Error bars indicate standard deviation. *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 5

Influence of PTPN11 p.(Tyr279Cys) and PDGFRB p.(Glu578Gln) on the development of imatinib resistance. Stably transfected cells expressing either PTPN11 wild-type (WT), p.(Tyr279Cys), or PDGFRB WT or p.(Glu578Gln) were exposed to increasing concentrations of imatinib. (A) Cells were cultivated with the respective imatinib concentration and the total cell number was analyzed using trypan blue staining for 0.1, 0.2, and 0.3 µM imatinib within 21 days. Black: Negative control (NC); dark grey: PTPN11; light grey: PDGFRB; solid line: mutation; dashed line: WT. (B) Ki-67 expression to analyze proliferation and (C) Caspase 9 activity of PTPN11 WT and p.(Tyr279Cys) transfected cells after 21 days of treatment with the respective imatinib concentration measured by caspase 9-Glo assay. Data were normalized to NC. Statistical analysis was performed using two-way ANOVA followed by Dunnett’s test. N = 3. Error bars indicate standard deviation. ***p < 0.001.