ABCB1 Overexpression Is a Key Initiator of Resistance to Tyrosine Kinase Inhibitors in CML Cell Lines

The tyrosine kinase inhibitor (TKI) imatinib has resulted in excellent responses in the majority of Chronic Myeloid Leukaemia (CML) patients; however, resistance is observed in 20–30% of patients. More recently, resistance to the second generation TKIs, nilotinib and dasatinib, has also been observed albeit at a lower incidence. ABCB1 has previously been implicated in TKI export and its overexpression linked to TKI resistance. In this study the dynamics of nilotinib resistance was studied in CML cell lines with particular focus on ABCB1 expression levels during development of resistance. Results revealed ABCB1 overexpression is likely an important initiator of nilotinib resistance in vitro. ABCB1 overexpression was also observed in cell lines as an intermediate step during development of resistance to imatinib and dasatinib in vitro. We conclude that ABCB1 overexpression may provide an initial platform to facilitate development of additional mechanisms for resistance to TKIs. This provides a rationale for investigating this phenomenon in patients undergoing TKI therapy.


Introduction
Chronic myeloid leukemia (CML) is a malignant disorder of the blood stem cells characterized by the presence of the Philadelphia chromosome which gives rise to the BCR-ABL tyrosine kinase [1,2]. Tyrosine kinase inhibitors (TKIs), such as imatinib [3], and the second generation inhibitors nilotinib [4] and dasatinib [5], act by selectively binding to the BCR-ABL kinase domain to inhibit kinase activity and subsequent downstream oncogenic signaling [6]. The advent of TKIs resulted in significantly improved treatment outcomes for the majority of patients with chronic phase(CP)-CML [7]; however, an estimated 20-30% of patients treated with first line imatinib therapy will exhibit primary or secondary resistance [8,9]. More recently, resistance to second generation TKIs has also been observed [10,11].
Intracellular drug concentrations are dependent on numerous parameters including absorption, distribution, metabolism, excretion and toxicity (ADME-Tox). One of the major groups of transporters involved in ADME-Tox is the ATP-Binding Cassette (ABC) superfamily. The interaction of ABCB1 and ABCG2 with TKIs has been extensively reviewed with the consensus indicating that ABCB1 is involved in TKI export at clinically relevant concentrations [34].
With the emerging interest in treating CML patients with customized therapeutic regimes, to reduce TKI resistance and maximize treatment outcomes, there will be a need for implementation of screening procedures in order to identify patients most at risk of developing resistance, thus allowing preventative clinical intervention. However, in order to determine which patients are more likely to develop TKI resistance it is necessary to understand the kinetics of resistance emergence. Generation of TKI-resistant cell lines via in vitro exposure to gradually increasing concentrations of imatinib [13-16, 28, 29, 36, 39-41], nilotinib [12,30,41] and dasatinib [15,40] provides an experimental system resembling the in vivo situation. However, in the majority of previous studies, resistance mechanisms were determined only once cells demonstrated overt resistance to the end-point concentration of TKI [12, 13, 16, 28-30, 36, 39, 41, 42]; researchers did not generally examine modes of resistance in the intermediary cells created during incremental increases of drug. Those studies that did examine the intermediate stages of resistance focused on the generation of imatinib and dasatinib resistance only [14,15,35,40]. Thus, in this study, two BCR-ABL1-expressing cell lines were cultured in increasing concentrations of nilotinib with samples examined at every stage of resistance generation, allowing interrogation of resistance development dynamics.
In an in vitro model of imatinib resistance, ABCB1 overexpression appeared most relevant during early stages of resistance development [40]. Thus, we postulated that increased ABCB1, leading to reduced intracellular levels of TKI, created a favorable environment for development of additional resistance mechanisms. In addition to the two nilotinib resistant cell lines generated as part of the current study, we also had access to imatinib-and dasatinib-resistant cells previously generated in our laboratory. In this prior study, kinase domain mutations and BCR-ABL overexpression were identified as end-point resistance mechanisms, however, ABCB1 expression was not examined over the course of resistance generation [15]. In the current study we now investigate the kinetics of ABCB1 expression and thus are able to evaluate the interplay of ABCB1 expression and secondary resistance mechanisms in vitro across three FDA approved TKIs. Additionally, in vivo evidence from two separate studies suggest that ABCB1 overexpression may also precede mutation development, disease progression and treatment failure in patients undergoing imatinib therapy [43][44][45]. Taken together, these results support ABCB1 overexpression as a likely initiator of resistance to imatinib, nilotinib and dasatinib and provide a strong rationale for investigations into the use of ABCB1 as a prognostic biomarker for identification of patients likely to respond poorly to TKI therapy. receives honoraria and research funds from Novartis Pharmaceuticals, BMS and Ariad and is a member of Advisory Boards for Novartis, BMS and Ariad. However, Novartis, BMS and Ariad had no role in the design of the study, collection and analysis of data nor the decision to publish. This does not alter the authors' adherence to PLOS ONE policies on sharing data and materials.

Materials and Methods
Cell lines BCR-ABL1-expressing cell lines K562 (American Type Culture Collection (ATCC) Manassas, VA, USA) and the ABCB1 overexpressing variant, K562-Dox, (Leonie Ashman, University of Newcastle, Callaghan, NSW) were cultured as described previously [46]. K562-Dox cells stably overexpress ABCB1 as a result of long-term exposure of the parental K562 cell line to the ABCB1 substrate doxorubicin.

Generation of TKI-resistant cell lines
Cell lines maintained in liquid culture were gradually exposed to escalating concentrations of nilotinib (Tasigna; kindly provided by Novartis Pharmaceuticals, Basel, Switzerland) as described in Supplemental methods. In washout experiments, cells were washed 3× in drug free media, left to equilibrate overnight, washed 3× in drug free media and cells lysed. Control cell lines cultured in 0.1% DMSO were maintained in parallel. Imatinib and dasatinib resistant cells were generated and cultured as described previously [15].
IC50 assays and western blotting 2×10 5 BCR-ABL1-expressing cells were incubated for 2 h at 37°C/5%CO 2 with concentrations of nilotinib ranging 0-100 μM, imatinib ranging 0-100 μM and dasatinib ranging 0-5000 nM. Following incubation, cells were lysed in Laemmli's buffer [47] before resolution by 12% SDS-PAGE and electrophoretic transfer to PVDF membrane (GE Healthcare, Buckinghamshire, UK) at 65 mA overnight. Western blotting for phosphorylated CT10 regulator of kinase like (p-CRKL) was performed as previously described [48]. IC50 values were determined as the dose of drug required to reduce p-CRKL levels by 50% and are presented as mean ± SEM. Cyclosporine A (Sigma-Aldrich, St. Louis, MO, USA) is an inhibitor of ABCB1 and was used at 10 μM; PSC-833 is a cyclosporine A derivative kindly provided by Novartis Pharmaceuticals and was used at 10 μM. Western blotting for other proteins was performed as described in Supplemental methods.

Cell viability assays
Cells were washed and resuspended in fresh culture media before culture in 24-well plates (Thermo Fisher Scientific, Waltham, MA, USA) in the presence of TKI at a density of 2×10 5 cells/mL. Plates were seeded with 1 mL of cell suspension in triplicate, placed in sterilised containers and incubated for 72 h before cell viability determination with 7-aminoactinomycin D Flow cytometry and fluorescent substrate efflux studies ABCB1 cell surface expression and function was measured as described previously [46]. Mean Fluorescence Intensity was used to calculate fold changes in ABCB1 expression in resistant vs. control cells.

Real time quantitative polymerase chain reaction (RQ-PCR)
1×10 7 intermediately resistant cells were stored in TRIzol stabilization solution (Invitrogen Life Technologies) at -80°C. RNA was extracted using the phenol/chloroform method [49] and cDNA synthesized using random hexamers (GeneWorks, Hindmarsh, SA, Australia) and Superscript II reverse transcriptase (Invitrogen Life Technologies). Primers were designed using Primer Express software v2.0 (Applied Biosystems, Foster City, CA, USA), and sequences were as described in Supplemental methods. Amplification was performed using RT 2 real-time SYBR Green/ROX PCR Master Mix (SuperArray Bioscience, Frederick, MD, USA) on a Rotor-Gene real time PCR machine (Corbett Research, San Francisco, CA, USA). Results were analysed using Rotor-Gene 6000 Series software (Corbett Research) and the relative expression levels of the genes of interest were calculated by the comparative Ct method using the 2 ΔΔCt formula to achieve results for relative quantification. Control cell lines were used as calibrators and all samples were normalized to house keeping genes (BCR for ABCB1, GUSB for LYN).

BCR-ABL1 quantitation and mutation analysis
BCR-ABL1 mRNA expression levels were quantitated using the TaqMan Universal PCR Master Mix (Applied Biosystems) and the ABI Prism 7500 Sequence Detection System (Applied Biosystems) as described previously [15]. BCR-ABL1 kinase domain sequencing was performed as described previously [15].

Statistics
Statistical tests were performed using the GraphPad Prism 5 statistical software (GraphPad Prism Inc, La Jolla, CA, USA). Normality tests were performed on each data set using the D'Agostino & Pearson omnibus normality test. The Mann-Whitney Rank Sum or the Student's t-test were used to determine differences between experimental groups depending on whether the data sets failed or passed the normality test, respectively. Differences were considered to be statistically significant when the probability value (p-value) was <0.05.

Generation of nilotinib resistant cells lines
Nilotinib resistant K562 and K562-Dox (ABCB1 overexpressing) cells were generated; nilotinib concentrations up to 100 nM were easily tolerated, however, upon exposure to 125 nM nilotinib, cell growth stalled and viability significantly decreased (S1 Table). After prolonged exposure to 125 nM nilotinib (K562 nilotinib intermediate #6), cell viability increased to~80% and resistance to nilotinib was evaluated by IC50 based on p-CRKL protein expression. These assays rely on BCR-ABL kinase inhibition as the key readout and thus enable a distinction between BCR-ABL dependent and BCR-ABL independent resistance; an increase in IC50 value indicates on target, BCR-ABL dependent resistance. K562 #6 NIL cells demonstrated significantly increased IC50 NIL when compared with control cells (2066 nM vs. 199 nM, p<0.001; Fig  1A, S1A Fig). Resistance was confirmed by cell viability assays (87% vs. 19% survival in 1000 nM nilotinib, p<0.001; Fig 1A). K562 #6 NIL cells were assessed for resistance to imatinib and dasatinib and again demonstrated significantly increased IC50 when compared with control cells (20.5 μM vs. 3.3 μM, p = 0.004 and 11.7 nM vs. 5.6 nM, p<0.001 for imatinib and dasatinib respectively; Fig 1B and 1C, S1B and S1C Fig). These results were confirmed by cell viability assays (92% vs. 29% in 5 μM imatinib and 74% vs. 21% in 500 nM dasatinib, p<0.001; Fig 1B  and 1C). In order to replicate the in vivo concentration of nilotinib to which a patient's leukemic cells are exposed [50], K562 #6 NIL cells were cultured in increasing concentration of nilotinib for a further two months until a concentration of 2 μM nilotinib was reached. Once this concentration was tolerated, the resultant K562 #10 NIL cells were assessed for maintenance of resistance compared with control cells (IC50 NIL = 1661 nM; 85% survival in 1000 nM nilotinib, p<0.001 for all nilotinib concentrations; Fig 1A). K562 #10 NIL cells also maintained resistance to imatinib (IC50 IM = 21.2 μM, 89% survival in 5 μM imatinib, p<0.001; Fig 1B) and dasatinib (IC50 DAS = 7.6 nM, p = 0.038; 74% survival in 500 nM dasatinib, p<0.001; Fig 1C). Thus the generation of nilotinib resistance in K562 led to a pan-resistant phenotype in all cases.
However, when resistance to TKI was assessed by p-CRKL dependent IC50, unexpected results were observed: IC50 to all three TKIs decreased. IC50 NIL decreased from 507 nM to 219 nM (p<0.001; Fig 1D,  A further decrease in IC50 was observed in K562-Dox #8 NIL cells despite the obvious resistance to all three TKIs measured in cell viability assays (Fig 1D-1F, S1D, S1E and S1F Fig). Given that p-CRKL is a downstream target of BCR-ABL, a decrease in IC50 indicated that these cells had developed a BCR-ABL independent mechanism of resistance.

ABCB1 overexpression is the initiator of resistance to nilotinib in vitro
Once overt TKI resistance had been established, potential mechanisms of resistance were interrogated. Due to the likelihood of ABCB1 involvement in nilotinib transport [34], ABCB1 protein and mRNA expression was monitored over the course of resistance development. The onset of resistance within K562 #6 NIL cells coincided with a significant increase in ABCB1 mRNA expression: resistant cells demonstrated up to 4.7-fold greater levels of ABCB1 mRNA compared with control cells (p<0.01; Fig 2A). Surface ABCB1 protein expression was also significantly increased in K562 #6 NIL cells compared with control (up to 56% ABCB1 vs. 8%; p = 0.002). However, upon continued culture to 2 μM nilotinib (intermediate #10), both ABCB1 mRNA and protein expression reduced to levels comparable with that of control cells (Fig 2A).
In order to determine whether increased expression of ABCB1 had a direct impact on nilotinib resistance, p-Crkl dependent IC50 NIL experiments were performed in the presence and absence of two ABCB1 inhibitors (cyclosporine and PSC-833). Indeed, a significant reduction in IC50 NIL was observed in resistant cells in the presence of both inhibitors: K562 #6 NIL IC50 NIL reduced from 2066 nM to 1522 nM in the presence of cyclosporine (p = 0.043) and to 1155 nM in the presence of PSC-833 (p = 0.009). No significant effect of ABCB1 inhibition on IC50 NIL was observed in K562 control cells (Fig 3). Interestingly, even in the presence of complete ABCB1 inhibition, the IC50 NIL of K562 #6 NIL cells remained significantly increased compared with K562 control cells indicating the presence of an additional resistance mechanism in this cell line. Further evidence for the direct involvement of ABCB1 overexpression in nilotinib resistance was provided when IC50 NIL was examined over a seven-day period in K562 #6 NIL cells overexpressing ABCB1. During this time period, ABCB1 expression decreased from 85% to 72% to 61% with a corresponding decrease in IC50 NIL from 2700 nM to 2120 nM to 1570 nM (S2 Fig). Taken together, these data indicate a key role for ABCB1 overexpression in the nilotinib resistance observed in this cell line model.
The phenomenon of initial ABCB1 overexpression followed by a significant reduction to almost negligible levels was also observed during development of TKI resistance in K562-Dox cells; an intriguing observation given that this cell line has stably expressed ABCB1 for over a decade. Initially, elevated expression of both ABCB1 mRNA (3.5-fold increase, p = 0.002) and protein (MFI 487 vs. 213, p<0.001) was observed when compared with levels in K562-Dox control cells (Fig 2B). However, upon prolonged culture (when cell growth stalled and viability decreased, S1 Table) two populations of cells became apparent, an ABCB1-positive population and an ABCB1-negative population. Interestingly, the ing that those cells with decreased ABCB1 expression had a growth advantage and were clonally selected over time. Following continued escalation of nilotinib to 2 μM (#8 NIL resistance intermediate), ABCB1 mRNA and protein expression was negligible and comparable to levels observed in parental K562 cells (Fig 2B).

K562-Dox cells lose all functional ABCB1 following continued culture in nilotinib
To our knowledge this is the first report documenting loss of ABCB1 expression in K562-Dox cells due to continued culture in TKI. Thus, it was important to determine whether there was also loss of ABCB1 function. Therefore, K562-Dox #8 NIL cells were assessed for ability to effectively efflux the ABCB1 substrate rhodamine-123. Results demonstrated a complete loss of ABCB1 function with the resistant cell line demonstrating comparable rhodamine-123 levels to the parental K562 cell line. Rhodamine-123 levels in K562-Dox control cells were as expected (Fig 2C).

ABCB1 overexpression precedes increased LYN phosphorylation in nilotinib resistant K562 cells
Because BCR-ABL1 overexpression is a common form of TKI resistance observed in patients [21], resistant K562 cells were screened for increased expression of BCR-ABL protein. Results demonstrated this possibility unlikely since no overexpression or increased kinase activity (as measured by phospho-BCR-ABL) was observed in early or late stage resistance intermediates ( Fig 4A). No kinase domain mutations were detected in any K562 resistance intermediates (S1 Appendix). Overexpression/increased activity of LYN was also investigated: results demonstrated increased LYN mRNA in K562 #6 NIL cells that remained high for the duration of nilotinib escalation (15-18 fold increase compared with control cells, p<0.05; Fig 2A). Additionally, catalytic activation of LYN increased significantly as measured by autophosphorylation at tyrosine 396 (Y396, p = 0.008; Fig 4B).

ABCB1 overexpression precedes Bcr-Abl independent resistance in nilotinib resistant K562-Dox cells
BCR-ABL kinase activity was assessed in nilotinib resistant K562-Dox cells to confirm adequate inhibition of phosphorylation. Nilotinib was removed from K562-Dox #8 NIL culture media by thorough washing followed by overnight equilibration. Total and phospho-BCR-ABL levels were determined and compared with the levels in cells that remained in continuous culture with nilotinib. Results demonstrated complete inhibition of kinase activity in the presence of nilotinib. However, upon drug washout, reactivation of BCR-ABL occurred resulting in similar levels of phospho-BCR-ABL in both resistant and control cells (p>0.05). Importantly, total BCR-ABL levels remained unaffected in the absence and presence of nilotinib ( Fig 4C). Thus, while nilotinib was effective at inhibiting BCR-ABL kinase activity, cells demonstrated complete resistance indicating the presence of an independent resistance mechanism. Additionally, no increase in LYN expression or activity was observed (S4 Fig); in fact, resistant cells demonstrated significantly decreased levels of LYN compared with control cells excluding LYN as a likely effector of resistance as has previously been reported [30,41].

ABCB1 overexpression precedes resistance to imatinib and dasatinib in vitro
Given that ABCB1 is likely to play a role in the transport of, and resistance to, imatinib and dasatinib [34], ABCB1 expression was examined in previously established imatinib and dasatinib resistant cell lines and correlated with already defined mechanisms of resistance [15]. Indeed, ABCB1 mRNA expression increased initially in response to both imatinib and dasatinib exposure, and then decreased once further resistance mechanisms predominated. When compared with control cells, imatinib-resistant K562 cells demonstrated up to 2.8-fold greater levels of ABCB1 mRNA; a significant decrease in expression of ABCB1 mRNA was associated with an increase in LYN mRNA expression (Fig 5A). Similarly, a peak of 8.6-fold greater ABCB1 mRNA levels in KU812 imatinib resistant cells was observed prior to a concomitant decrease in expression following development of the imatinib resistant F359C kinase domain mutation (Fig 5B). The same initial increase in ABCB1 mRNA followed by a decrease once additional resistance mechanisms predominated was observed in two separate dasatinib resistant K562-Dox cell lines (DAS1 and DAS2; Fig 5C and 5D). ABCB1 protein expression confirmed these data (S5 Fig). Taken together, these results indicate that ABCB1 overexpression likely acts as an initial mediator of resistance to nilotinib, imatinib and dasatinib providing favourable conditions for further resistance development.

Discussion
Nilotinib-resistant cell lines have previously been generated as a means of recapitulating in vivo resistance, however, this is the first study in which the resistance mechanisms were studied in a sequential fashion to better define the dynamics of resistance development. The model described here allowed an interrogation of the interplay between various mechanisms in the development of frank resistance. In addition, previous studies have generated cells resistant to 20 nM [30] and 50 nM [12] nilotinib whereas the end-point concentration of nilotinib chosen in this study (2 μM) better reflects the trough plasma level observed in patients [50] and thus is a more accurate representation of the nilotinib concentration to which a patient's leukaemic cells are exposed.
Even though TKIs such as imatinib, and more recently nilotinib and dasatinib, have resulted in significantly better treatment outcomes for the majority of patients with CML, resistance to therapy is still a problem for a significant proportion of patients (up to 30%) [8,9]. Additionally, BCR-ABL independent mechanisms of resistance are poorly understood. Accordingly, a greater understanding of the kinetics of resistance development, especially in the BCR-ABL independent setting, is required. As we move into the era of customised treatment regimes, it becomes important to predict which patients are likely to respond to a given TKI such that treatment plans are tailored to the individual. Thus, in vitro cell line models are key in determining, not only common modes of TKI resistance, but also kinetics of resistance development. In this study, two cell line models of nilotinib resistance were generated and resistance emergence examined; resistance kinetics in response to imatinib and dasatinib were assessed in four additional cell lines. Results highlighted ABCB1 overexpression as a likely initiator of resistance to all three TKIs.
Notably, this is the first report of ABCB1 overexpression in the K562 cell line in response to exposure to nilotinib, which decreased following a concordant increase in LYN kinase activity. LYN overexpression has previously been reported as a BCR-ABL independent mode of resistance [27]. However, in the current study, K562 #10 NIL cells with increased LYN activity also demonstrated increased p-CRKL dependent IC50, which was unexpected if the resistance was BCR-ABL-independent. This, combined with the fact that LYN reportedly regulates BCR-ABL activity [33,51,52], suggests that LYN overexpression is in fact a BCR-ABL dependent mechanism of resistance, which challenges current dogma. More intriguing is the fact that these cells remain resistant to dasatinib, which has activity against Src Family Kinases including Lyn. This raises the possibility that an additional resistance mechanism/s is present (eg: overexpression of an efflux transporter, activation of an alternate signalling pathway) and warrants further investigation.
Exposure of K562-Dox cells to nilotinib resulted in a complete loss of ABCB1 expression and function following initial ABCB1 overexpression. K562-Dox cells have stably expressed ABCB1 for over a decade and this is the first report of complete loss of ABCB1 expression back to levels observed in parental K562 cells. No overexpression or increased activity of LYN was observed. Instead, resistance is likely attributable to a novel BCR-ABL independent mechanism of resistance, an area of research under current investigation.
Examination of imatinib-and dasatinib-resistant cell lines generated within our laboratory and characterised previously [15] also highlighted ABCB1 overexpression as an initiating mechanism of resistance to these TKIs. Again, most previous studies examined resistance in cells surviving end-point TKI concentration only [12, 13, 16, 28-30, 36, 39, 41, 42]. Two studies in which intermediately resistant cells were studied either failed to examine ABCB1 expression [14], or did not correlate the increased ABCB1 expression observed with TKI resistance [35]. One recent study, however, did confirm our findings with results demonstrating exposure of K562 cells lines to imatinib and dasatinib resulted in increased ABCB1 mRNA. Authors also hypothesised that increases in transporter expression may be important in the early stages of resistance development but not as crucial after prolonged exposure to TKI [40]. The involvement of ABCB1 in development of resistance to nilotinib, imatinib and dasatinib is further substantiated when the rate of resistance acquisition in K562-Dox cells, which already overexpress ABCB1, is considered. When compared with K562 and KU812 cells (which express negligible basal levels of ABCB1) K562-Dox cells develop resistance to nilotinib, imatinib and dasatinib at a faster rate, withstanding higher concentrations of TKIs earlier in the dose escalation process. K562-Dox cells also required less time to tolerate a given concentration of TKI, reaching end point concentrations faster. These data confirm ABCB1 overexpression as a key initiator of resistance to all three TKIs (S2 Table). Thus, it appears likely that ABCB1 overexpression causes increased export of TKIs leading to lower intracellular concentrations. This then results in suboptimal BCR-ABL inhibition, which presumably creates a favourable environment for development of other more efficient resistance mechanisms. Given the variety of resistance mechanisms observed, both in our studies and those from others, it is likely that the events following ABCB1 up regulation, are stochastic in nature and dependent on the cell lines studied.
In conclusion, the findings detailed here suggest that ABCB1 overexpression likely provides an initial platform required for development of additional mechanisms of resistance to TKIs in vitro. The cell line models of resistance described provide a valuable tool for studying resistance in patients. We observed a consistent upregulation of ABCB1 as a result of exposure to three FDA approved TKIs; we are currently investigating whether this phenomenon also occurs in patients receiving TKI therapy. With further in vivo validation, the results presented here suggest monitoring of ABCB1 levels may potentially identify patients likely to develop additional resistance and/or respond poorly to their current therapy regime.
Supporting Information S1 Appendix. K562 nilotinib resistance intermediates do not harbour any Bcr-Abl kinase domain mutations. DNA sequencing of the Bcr-Abl kinase domain of K562 control cells and nilotinib 4resistance intermediates was conducted and compared with the GenBank ABL reference sequence. Data demonstrate that neither the control cells nor any of the resistance intermediates contain mutations. Data were analysed using Mutation Surveyor Version 3.24 with each peak representing a DNA base (A = Adenine; C = Cytosine; G = Guanine; or T = Thymine) in the 5'-3' direction of the ABL sequence. The base number for both the reference sequence and the sequences of interest are indicated. The amino acid and corresponding residue number are also indicated. The amino acid sequences exactly match the reference sequence with no kinase domain mutations present. NIL = nilotinib. (PDF) S1 Fig. K562 and K562-Dox cells cultured long term in nilotinib demonstrate cross-resistance to imatinib and dasatinib.  Fig. ABCB1 mRNA levels increase initially in imatinib resistant KU812 cells then decrease following emergence of kinase domain mutations. Expression levels of ABCB1 mRNA were assessed in KU812 cells resistant to imatinib. Expression levels were then correlated with other, previously defined, resistance mechanisms [15]. Specifically, % of BCR-ABL1 mRNA (maroon line) and % of various kinase domain mutations (orange, yellow, green, blue, purple lines) are indicated. mRNA expression represents the mean of at least three independent experiments performed in triplicate. Error bars represent SEM. IM = imatinib. (TIF) S1 Table. Summary of nilotinib (NIL) concentrations to which cell line resistance intermediates were exposed and the corresponding number of days before dose was increased. (DOCX) S2 Table. Summary of imatinib (IM) and dasatinib (DAS) concentrations to which cell line resistance intermediates were exposed and the corresponding number of days before dose was increased. Cells lines shown in bold have been assessed for ABCB1 expression in the current manuscript. Note that cell lines expressing negligible levels of ABCB1 (K562 and KU812) required longer periods of time to develop resistance to IM and DAS compared with K562-Dox cells which demonstrate overexpression of ABCB1 initially. Additionally, generation of a DAS resistant K562 cell line was extremely difficult (cells kept dying at the 1 nM DAS stage, intermediate #2) and was attempted three times before successful dose escalation occurred. A DAS resistant KU812 cell line could not be generated due to the inherent sensitivity to TKIs of this cell line.