Secondary fusion proteins as a mechanism of BCR::ABL1 kinase-independent resistance in chronic myeloid leukemia

Abstract

Drug resistance in chronic myeloid leukemia (CML) may occur via mutations in the causative BCR::ABL1 fusion or BCR::ABL1-independent mechanisms. We analyzed 48 patients with BCR::ABL1-independent resistance for the presence of secondary fusion genes by RNA sequencing. We identified 10 of the most frequently detected secondary fusions in 21 patients. Validation studies, cell line models, gene expression analysis, and drug screening revealed differences with respect to proliferation rate, differentiation, and drug sensitivity. Notably, expression of RUNX1::MECOM led to resistance to ABL1 TKIs in vitro. These results suggest secondary fusions contribute to BCR::ABL1-independent resistance and may be amenable to combination therapies.

Although most chronic myeloid leukemia (CML) patients in chronic phase are successfully treated with ABL1 tyrosine kinase inhibitors (TKIs), approximately a third will develop resistance.¹ Resistance mechanisms may occur in a BCR::ABL1 kinase-dependent manner due to reactivation of BCR::ABL1 kinase activity, most commonly through acquired mutations in the ABL1 kinase domain which compromise drug binding. Alternatively, less well characterized BCR::ABL1 kinase-independent resistance mechanisms may occur through secondary molecular changes resulting in persistent activation of alternative pathways, despite effective BCR::ABL1 kinase inhibition.^1-3 While second-line TKI therapies exist to effectively treat BCR::ABL1-dependent resistance, efficacy is more limited in BCR::ABL1-independent resistance,^1,2 increasing risk of progression to accelerated phase (AP) and blast crisis (BC) disease stages, which are associated with poor survival.³

Previous research has identified several changes that occur in the advanced phases of CML, including the appearance of secondary fusion proteins, which may contribute to disease progression. Some such fusions involve genes such as RUNX1,^4,5 CBFB,^6,7 and KMT2A⁸ that are associated with different cancers. However, the significance of secondary fusions and their ability to mediate resistance remains unclear. We hypothesized that these fusions cause aberrant cellular signaling which contributes to BCR::ABL1 kinase-independent resistance.

To identify candidate fusions, total RNA was extracted from primary patient mononuclear cells from 48 patients with BCR::ABL1 kinase-independent resistance (loss of achieved molecular and/or cytogenetic response to ≥1 TKI without an explanatory BCR::ABL1 kinase domain mutation) and subjected to paired-end RNA-seq analysis (Figure 1A). Beyond detection of the expected BCR::ABL1 fusion, 10 of the most commonly detected in-frame secondary fusions included: TTYH3::LFNG, KDM7A::MKRN1, LRRC6::TMEM71, ZNF292::PNRC1, TRDV2::TRAC, TPM4::ACTB, RUNX1::MECOM, PDPK1::ATP6V0C, HNRPU::ZBTB18, and CBFB::MYH11 (Figure 1B, Supplemental Table 1-2). Among the 21 of 48 patients (43.8%) harboring one of these 10 fusions, 15 patients exhibited a mix of either CML-CP or more advanced disease, with most showing significant disease burden (median BCR::ABL1 transcript levels: 37% FISH and 26.8%, qPCR), and having previously failed at least 2 TKIs before sample collection (Supplemental Table 3).

Figure 1: Several recurrent fusions are present in patients with BCR::ABL1 kinase-independent resistance and associate with resistance to ABL1 TKIs.

A. Schematic for fusion identification from patient samples. B. Left panel: Circos plot illustrating all fusions found in from a cohort of 48 patients with BCR::ABL1 kinase-independent resistance. Right panel: Detected frequencies for 10 of the most frequent in-frame fusions identified by RNA-seq. C. Example electrophoresis of PCR fusion breakpoint amplification and verification for four fusions (RUNX1::MECOM, CBFB::MYH11, KDM7A::MKRN1, and PDPK1::ATP6V0C). The deidentified number for each sample is provided above each lane. D. Comparison of sensitivity to imatinib for Ba/F3 cells expressing BCR::ABL1 only, BCR::ABL1 with a secondary fusion or empty vector, and Ba/F3 parental cells. E. Fold-change in imatinib IC50 for all fusion cell lines is recorded across all trials. P-values were calculated using one-way ANOVA with a Bonferroni correction for multiple comparisons. Error bars represent mean plus or minus the standard error of the mean. F. Relative proliferation rates from all MTS assays for all fusion cell lines normalized to BCR::ABL1 only. P-values were calculated using one-way ANOVA with a Bonferroni correction for multiple comparisons. Error bars represent mean plus or minus standard error of the mean. G. Flow cytometry-based profiling of changes in expression of CD11b across cell lines. Fluorescence minus one (FMO) control included for gating reference.

When compared to cohorts of BCR::ABL1-dependent resistance patients and those with newly diagnosed disease, 5/10 fusions (ZNF292::PNRC1, PDPK1::ATP6V0C, RUNX1::MECOM, TPM4::ACTB, CBFB::MYH11) were found exclusively in BCR::ABL1-independent patients; the remaining were found at low levels in at least one of the other cohorts (Supplemental Table 4). These findings are consistent with recent genomic studies suggesting that some patients with resistance/progression may potentially use both BCR::ABL1-independent and -dependent mechanism.^3,9 PCR analysis confirmed nine predicted breakpoints (Figure 1C, Supplemental Figure 1, Supplemental Table 5), six of which were Sanger verified (Supplemental Figure 2) and analyzed further to explore their potential contribution to resistance. Among these, RUNX1::MECOM has been identified in poor prognosis, therapy-associated AML and associated with megakaryoblastic leukemia in murine models.^5,10 RUNX1 translocations involving other partner genes (e.g. RUNX1T1) are also found in AML.⁴ Additionally, the CBFB::MYH11 fusion defines a diagnostic subset of AML associated with good outcomes^7,11, and KDM7A fusions with BRAF have been previously reported in gastric cancers.¹²

Most of the fusions appeared to possess a DNA binding domain combined with several binding/association domains (Supplemental Figure 2A-D). RUNX1::MECOM features the RUNT binding domain of RUNX1, which is important in the transcriptional control of hematopoiesis, combined with the SET domain and several zinc finger domains of MECOM (Supplemental Figure 2A). In general, these fusion could alter DNA binding and chromatin remodeling, which may result in transcriptional changes that contribute to resistance phenotypes.

To characterize the secondary fusions in the context of BCR::ABL1-positive cells, murine Ba/F3 cell lines were generated co-expressing BCR::ABL1 and one of four secondary fusions: RUNX1::MECOM, KDM7A::MKRN1, CBFB::MYH11, and TRDV2::TRAC (Supplemental Figure 3A, Supplemental Table 6). Cell proliferation assays were used to evaluate sensitivity of BCR::ABL1 Ba/F3 cells co-expressing secondary fusions to a panel of ABL1 TKIs (Supplemental Figure 3B). RUNX1::MECOM co-expressing cells demonstrated increased resistance to ABL1 TKIs, most prominently to imatinib (Figure 1D and E). Ba/F3 BCR::ABL1 cells co-expressing RUNX1::MECOM grew approximately 2-fold slower than BCR::ABL1 only cells (p=0.0068); a similar trend was seen for cells co-expressing BCR::ABL1 and CBFB::MYH11 (Figure 1F). Given the association between proliferation rate and differentiation, and previous work indicating BCR::ABL1 cells co-expressing RUNX1 mutations have increased expression of CD11b and CD19,⁹ we evaluated changes in CD11b, CD19, and CD45 expression. While Ba/F3 BCR::ABL1 cell lines co-expressing secondary fusions did not demonstrate significant changes in CD19 or CD45 expression, those co-expressing RUNX1::MECOM had a significant increase in CD11b expression compared to BCR::ABL1 only and MIG empty vector controls (Figure 1G). The combination of reduced proliferation and increased expression of CD11b may result from RUNX1::MECOM effecting an adaptation to a monocytic-like differentiation state, which may contribute to the resistance phenotype. Notably, cell differentiation state in CML has previously been associated with decreased sensitivity to ABL1 TKIs.^9,13

To uncover mechanisms underlying resistance phenotypes, we compared gene expression profiles in Ba/F3 cells expressing BCR::ABL1 alone or with empty MIG vector, RUNX1::MECOM, or KDM7A::MKRN1 after treatment overnight with or without 250 nM imatinib (Supplemental Methods). Several clusters of differentially expressed genes became further altered upon treatment with imatinib (Figure 2A and Supplemental Table 7). Pathway analysis using the Reactome knowledgebase and the Pathway Regulation Score (PRS) method indicated that RUNX1::MECOM expressing cells treated with imatinib demonstrated significant enrichment of pathways involving FGFR, PI3K, MET, TNF-alpha, RUNX1, and NF-kB compared to imatinib-treated MIG empty vector control cells (Figure 2B, Supplemental Table 8). Most genes in these pathways were not highly expressed in RUNX1::MECOM cells before treatment. KDM7A::MKRN1 expressing cell lines upregulated pathways involving MET and PKA in response to treatment with imatinib, while downregulating pathways involving NF-kB (Figure 2B).

Figure 2: Secondary fusions activate select pathways following imatinib treatment, which are targetable through combination therapy.

A. Unsupervised clustering of differentially expressed genes for cells expressing BCR::ABL1 and RUNX1::MECOM (red) or KDM7A::MKRN1 (purple) compared to MIG empty vector control (blue), with or without 250 nM imatinib treatment (grey). Only genes with FDR-adjusted p<0.05 and a fold-change of ≥2.5 in either direction were included, and unsupervised hierarchical clustering was performed using the Canberra distance calculations. Scaled gene expression (z-score) is shown for each gene, with upregulation in orange and downregulation are in blue. B. Volcano plot of differentially expressed pathways for Ba/F3 BCR::ABL1 cells co-expressing RUNX1::MECOM (top) and KDM7A::MKRN1 (bottom) when compared to MIG empty vector expressing cells without treatment (left) and with imatinib (right). Size of the dot corresponds to the overall PRS enrichment score, with select highlighted pathways of interest. Examples of differentially expressed genes for RUNX1::MECOM expressing cells from C. the PI3K-Akt-mTOR pathway, D. the FGFR signaling cascade, and E. p38 MAPK signaling associated genes. F. Examples of differentially expressed genes associated with integrin signaling and transcription for KDM7A::MKRN1 co-expressing cells. All p-values for differentially expressed genes reflect FDR-adjustment followed by gene-specific Type I error rate control for multiple pairwise comparisons. Ends of the box and whisker plot indicate the range of observed values. G. Left panel: Plot of sensitivity to imatinib for Ba/F3 BCR::ABL1 cells co-expressing RUNX1::MECOM when treated alone (black), with 25 nM of dactolisib (pink), and with 5000 nM idelalsib (green). Right panel: IC50 values for each Ba/F3 cell line, including BCR::ABL1 only, MIG empty vector, and Ba/F3 parental cell controls, are plotted. H. Surface 2D plot of imatinib and idelalisib synergy (ZIP score) generated from a 7x7 dose matrix with Ba/F3 cell lines of interest. Synergy was calculated using the R ‘synergyfinder’ package.

Differentially expressed genes within enriched pathways include genes associated with canonical BCR::ABL1 signaling.² For RUNX1::MECOM, increased expression of the PI3K-Akt-mTOR pathway was attributable to Pik3cb, Akt3, and Prkca, upon treatment with imatinib (Figure 2C). Notably, PI3K cascade activation has been associated with increased protein synthesis, cell growth, and cell survival in CML.^2,14 In addition, RUNX1::MECOM expressing cells treated with imatinib had elevated expression of genes involved in FGFR (Sos2, Ptpn11, and Fgf23; Figure 2D) and MAPK signaling (Mapk14, Irak1, and Traf6; Figure 2E). Consistently, among patient samples with BCR::ABL1-independent resistance, patients harboring RUNX::MECOM exhibited increased expression of many of these genes (Supplemental Figure 4). This evidence suggests that many of the pathways critical for CML pathophysiology and survival may be persistently activated in BCR::ABL1 cells despite ABL1 kinase inhibition.

For KDM7A::MKRN1 expressing cells, elevated gene expression in MET signaling pathways was observed upon imatinib treatment (Figure 2B). Furthermore, genes related to integrin signaling, such as Itgb1 (CD29) and Itgae (CD103), along with several genes related to chromatin remodeling and transcription, such as H2afx, were also significantly upregulated (Figure 2F). These findings suggest that KDM7A::MKRN1 may result in transcriptional changes that correlate with its phenotype.

Lastly, given evidence of upregulation of the PI3K-Akt-mTOR pathway in the Ba/F3 BCR::ABL1 cells co-expressing RUNX1::MECOM, cell lines were tested with imatinib alone or in combination with the PI3K inhibitor idelalisib (5000 nM) or Akt inhibitor dactolisib (25 nM). Consistent with PI3K-Akt-mTOR as one of the many downstream pathways activated by BCR::ABL1, the addition of idelalisib resulted in a significant reduction in the IC₅₀ values for imatinib across all BCR::ABL1 cell lines, though the reduction was the most pronounced for RUNX1::MECOM co-expressing cells (Figure 2G). These results were mirrored across an expanded dose matrix, where idelalisib plus imatinib combinations demonstrated greatest synergy in RUNX1::MECOM co-expressing cells (Figure 2H). Notably, Ba/F3 parental cells showed little sensitivity at doses of 5000 nM of idelalisib, and based on previous studies of the reported drug levels of idelalisib in patients (C_max for 200mg BID: 1790 ng/mL),¹⁵ required idelalisib levels should be achievable in patients. These data provide strong evidence that cells co-expressing BCR::ABL1 and RUNX1::MECOM are amenable to combination therapy with PI3K inhibitors and ABL1 TKIs as a result of upregulated PI3K signaling.

Although certain fusions showed phenotypic alterations, co-expression of them did not result in significant ABL1 TKI resistance in our models. Many of the fusions, such as CBFB::MYH11, showed some increased resistance to various ABL1 TKIs but did not achieve statistical significance. Importantly, the Ba/F3 system evaluates these fusions in isolation, whereas in the majority of patients, the chromosomal and genomic instability that generated the fusions may also produce additional genetic alterations which could be required for certain fusions to express their resistance phenotype.¹¹ Additionally, some potential fusions may not lead to significant expression changes. KDM7A::MRKN1, for example, may represent a tandem repeat that causes insufficient changes necessary for resistance. TRDV2::TRAC, which did not lead to a significant phenotype in our experiments, may represent a T-cell receptor variation previously observed in T-ALL. Furthermore, some of the fusions which run in the same direction (e.g. ZNF292::PNRC1, PDPK1::ATP6V0C) could potentially be the product of circular RNA, resulting in low-level transcriptional events without intracellular consequence. Notably, for the 6 patients from our newly diagnosed/TKI-naïve cohort with a secondary fusion detected and follow-up response data available, 5 of these patients (1 with TTYH3::LFNG, 1 with LRRC6::TMEM71, 3 with TRDV2::TRAC) achieved subsequent molecular remission on ABL1 TKI-based therapy; the remaining patient (harboring HNRNPU::ZBTB18) achieved hematologic but not cytogenetic response to imatinib and subsequently underwent bone marrow transplant. Future research into discernment of these fusion candidates may require alternate assays and cellular models.

In summary, we validated multiple secondary fusions in patients with BCR::ABL1 kinase-independent resistance and demonstrated that expression of RUNX1::MECOM associates with PI3K-Akt signaling-mediated decreased sensitivity to ABL1 TKIs and an altered differentiation state. These findings suggest that such patients may be amenable to combination therapy with PI3K inhibitors.

Data Availability Statement

All data from experiments supporting the findings of this study are available from the corresponding authors upon reasonable request.