ATP Binding Cassette C1 (ABCC1/MRP1)-mediated drug efflux contributes to disease progression in T-lineage acute lymphoblastic leukemia

Purpose— In acute lymphoblastic leukemia (ALL), multidrug resistance is often mediated by ATPase Binding Cassette (ABC) proteins, which principally involve ABCB1 (multidrug resistance 1, MDR1 ) and ABCC1 (multidrug resistance protein 1, MRP1 ). However, direct comparisons between the differential effects of ABCB1 and ABCC1 have been difficult, since identical cell lines with differential expression of these transporters have not been developed. Experimental Design— In this study, we developed and compared the biological profiles of Jurkat cell lines that selectively over-expressed ABCB1 and ABCC1. Vincristine (VCR) plays an important role in the treatment of T-lineage ALL (T-ALL), and is often the first drug given to newly-diagnosed patients. Because of its importance in treatment, we provided escalating, sub-lethal doses of VCR to Jurkat cells, and extended our observations to expression profiling of newly diagnosed patients with T-ALL. Results— We found that VCR-resistant cells over-expressed ABCC1 nearly 30-fold. The calcein AM assay confirmed that VCR-resistant cells actively extruded VCR, and that ABCC1-mediated drug resistance conferred a different spectrum of multidrug resistance than other T-ALL induction agents. siRNA experiments that blocked ABCC1 export confirmed that VCR resistance could be reversed in vitro. Analyses of T-lymphoblasts obtained from 92 newly diagnosed T-ALL patients treated on Children's Oncology Group Phase III studies 8704/9404 showed that induction failure could be explained in all but one case by the over-expression of ABCB1 or ABCC1. Conclusions— Taken together, these results suggest that over-expression of ABC transporters plays a contributing role in mediating treatment failure in T-ALL, and underscore the need to employ alternate treatment approaches in patients for whom induction failed or for those with relapsed disease.


Introduction
With modern, dose-intensive chemotherapeutic approaches, nearly 80% of children, adolescents and young adults with acute lymphoblastic leukemia are now being cured of their diseases [1,2]. Despite this steady progress towards improved outcome, leukemic relapse remains a common problem. Inherently involved in relapse are a super-family of ATP Binding Cassette (ABC) drug transport proteins that actively extrude many chemically unrelated compounds out of dividing cancer cells [3][4][5]. The ABC transport proteins have been widely implicated in disease relapse in a variety of hematopoietic malignancies, including T-lineage acute lymphoblastic leukemia (T-ALL) [6][7][8].
Malignant T-lymphoblast survival in the presence of chemotherapeutic agents is dependent on cellular and molecular alterations that are not present in chemotherapy-sensitive cells. The resistance of tumor cells to multiple, structurally or functionally dissimilar drugs is classified as multi-drug resistance (MDR) [9]. Experimentally-induced drug resistance in leukemic cell lines has shown that acquired MDR is commonly mediated by ATP-binding cassette (ABC) proteins ABCB1 (MDR1), and ABCC1 (MRP1) [7,8]. In T-ALL, overexpression of ABCB1 has not been uniformly associated with a poor outcome [10,11], but we found that intrinsic up-regulation of ABCB1 did correlate with induction failure within a subset of patients who were treated on Children's Oncology Group Phase III studies 8704 and 9404 [8]. The role for ABCC1 in mediating therapy resistance for T-ALL has been even less clear. While the family of multidrug resistance-associated proteins has been implicated in causing chemotherapy resistance in T-ALL [6], others found that ABCC1 did not independently predict poor outcome [12]. The roles played by ABC transport proteins in T-ALL have yet to be fully understood, and may have differential effects on treatment within biologically-defined subsets of patients.
While treatment approaches have varied significantly over the preceding decades, induction therapy for T-ALL commonly begins with vincristine (VCR), which is typically given as 1.5 mg/m 2 (maximum 2 mg) IVP once weekly [13][14][15][16]. Because of this consistent approach in VCR dosing, we developed a VCR-resistance cell line to investigate its resistance profile. We found that T-ALL cells acquire resistance to VCR through an up-regulation of ABCC1, and that the multi-drug resistance generated from ABCC1 expression is distinctly different from ABCB1 over-expression. We found that intrinsic over-expression of ABCB1 and ABCC1 was predictive of induction failure in all most cases, suggesting that additional resistance mechanism(s) remain to be defined. Taken together, these findings call for the identification of individual patients with ABCB1/C1 up-regulation, as they might benefit from an alternate approach to induction therapy.

Quantification of drug resistance
As previously described [7,8], drug resistance was quantified first by simply measuring duration of survival in 300 nM VCR. Progenitor (parental) and VCR-resistant cells were maintained separately at a concentration of 2 × 10 5 cells/ml with chemotherapy added at Day 0 and percent viability estimated with a hemacytometer (Hausser Scientific, Horsham, PA) and 0.4% trypan blue dye (Sigma). Secondly, the effective concentration resulting in 50% cell death (EC 50 ) was measured by culturing 1 × 10 5 cells/ml with logarithmically increasing concentrations of VCR and assessing cell viability for parental and resistant cell lines after 7 days.

Microarray analyses of Jurkat cells
Total RNA was extracted from parental and resistant cell lines according to RNeasy Mini Kit manufacturer's protocol (Qiagen, Baltimore, MD). Hybridization of RNA to the Human Genome U 133 Plus 2.0 oligonucleotide microarray (Affymetrix, Santa Clara, CA) and data were normalized using the methods of the Robust Microarray Analyses, as previously described [17]. The differences in gene expression between parental and resistant cell lines were computed using GeneSpring 7.3.1 (Agilent Technologies, Inc., SantaClara, CA).

Flow cytometry analysis for surface expression and functional analysis of transport activity
The ABCC1 transporter pump was detected at the cell surface by flow cytometry using PElabeled mouse anti-human ABCC1 monoclonal antibody (clone QRCL-3, BD Biosciences, San Jose, CA). The antibody was incubated at 4° C for 30 minutes in the dark with 1 × 10 6 parental and resistant cell lines and then were fixed and permeabilized according to the Cytofix/Cytoperm Kit (BD Biosciences) alternative protocol. In addition, to detect that the ABCC1 pumps were functional, we employed a calcein acetoxymethyl ester (calcein AM) duplex assay [18]. In living cells, calcein AM is cell permeant and non-fluorescent until it is hydrolyzed by intracellular esterases into a strongly green fluorescent anion, calcein. Calcein is retained within the cell cytoplasm, unless it is extruded by active protein transporters, such as ABCB1 or ABCC1. The Vybrant Multidrug Resistance Assay Kit (Invitrogen-Molecular Probes, Eugene, OR) was used to functionally test ABCC1 activity. Experiments were run according to manufacturer's suggested protocol and at least 10,000 events were counted (FACscan cytometer) and analyzed with IDLeQuery software [19].

siRNA-mediated suppression of gene expression
To investigate the effects of siRNA on ABCC1-mediated VCR resistance, 2 × 10 5 Jurkat cells were centrifuged and re-suspended in 75 μl siPORT electroporation buffer (Ambion, Austin, TX) and transferred into 1-mm gap electroporation cuvettes (Bio-Rad Laboratories, Hercules, CA). According to the manufacturer's suggestions, 2.25 μg of siRNA was added and cells were then electroporated with a single square-wave pulse for 280 μs at 220 V delivered by a Gene Pulser XCell (Bio-Rad). To preserve cell wall integrity, cells were incubated in the cuvette for 10 minutes at 37° C before transfer. The siRNA sequence used to suppress ABCC1 transporter RNA was 5'-GGUGUUAUACAAGACCUUUtt-3' (sense) and 5'-AAAGGUCUUGUAUAACACCtt-3" (antisense) at the optimal amount of 1ug/ sample (ABCC1-siRNA, #AM51334; Ambion). VCR resistant cell lines were used to establish experimental points, with and without GAPDH siRNA (Ambion) was used as a control. After 48 hours of maintaining viability, 500 nM VCR was added to each sample with percent viability and absolute cell concentration recorded on subsequent days. Total RNA was isolated with the RNeasy Mini Kit (Qiagen) and qRT-PCR performed using the SuperScript III Platinum Two-Step qRT-PCR Kit with SYBR Green (Invitrogen, Carlsbad, CA). The relative rate of gene expression was compared using the delta-delta method [20].

Clinical case preparation and microarray analyses
Children's Oncology Group (COG) Cell Bank newly-diagnostic samples from ALL Classification studies 9900/9904 and AALL03B1/AALL0434 were enriched to > 90% T-ALL blasts with ficoll-hypaque density centrifugation prior to cryopreservation. RNA processing and hybridization to the Human Genome U133 Plus 2.0 oligonucleotide microarray (Affymetrix) were performed as previously described [17], and 100 cases were examined for their relative expression of ABC proteins based on their outcome into three groups: 1) induction failure (n= 11); 2) post-remission relapse (n= 20); and 3) complete continuous remission (CCR, n= 69). Patient subset comparisons were performed using twotailed Student's t-test function in Prism 4.0 (GraphPad, La Jolla, CA). Expression profiles for the 9404 profiles are publically available under GEO accession number GSE14618 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE14618).

Development of a VCR-resistant T-ALL cell line
To investigate our hypothesis that VCR-resistance is an important underlying feature in disease progression/relpase in T-ALL, it was necessary to first develop a VCR resistance in a T-ALL cell line. This goal was accomplished by incrementally increasing VCR doses on a weekly basis over a time-span of approximately nine months; these conditions mimic those associated with leukemic relapse. Once the cell line became tolerant to VCR concentrations that are found within first and second order elimination therapeutic ranges in patient sera [21], we next tested Jurkat cell viability in analogous VCR concentrations (300 nM) and found 100% of the cells remained viable for one week in a tissue culture setting (Fig 1A). Following these experiments, we measured the ex vivo inhibitory concentration causing a 50% diminution of cell viability (EC 50 ) in the wildtype, parental cells and the VCR-resistant clone. Compared to the parental cell line, the VCR-resistant cells showed a 155-fold increase in EC 50 from 0.96 nM to 149 nM (Supplemental Table, and Fig. 1B). These data show the Jurkat T-ALL cells are dynamically able to acquire a high level of VCR resistance when maintained in escalating doses of sub-lethal VCR concentrations, and at levels greater than the mean therapeutic concentration typically achieved during therapy.

ABCC1 up-regulation in VCR-resistant cells
To assess whether over-expression of a membrane transporter protein might be responsible for VCR resistance, we next tested whether ABCC1 surface protein expression might be inducible under conditions of cytotoxic stress (Figure 2, A1 and A2). Wildtype Jurkat cells did not show baseline expression of the ABCC1 protein relative to a control mAb (Panel A1), but cells maintained in long-term co-culture with VCR demonstrated an approximate 100-fold increase in expression of ABCC1 (Panel A2). We then tested whether additional ABC genes might be differentially regulated under the same experimental conditions. To accomplish this goal, we compared the RNA expression level of all the ABC-transporters in both the parental and VCR-resistant cells by gene expression profiling. ABCC1 was the only transporter that was significantly up-regulated out of the 48 ABC-transporters that were tested (Fig. 2B). Specifically, ABCC1 was increased nearly 30-fold in VCR-resistant cells as compared to the parental controls, but no other ABC transporters genes were differentially over-expressed. Interestingly, ABCB1 (MDR1) was not up-regulated in response to VCRdependent disease selection, as we previously described in response to DNR exposure [8].

Up-regulated ABCC1 is an effective drug efflux protein
In order to demonstrate that ABCC1 expression is responsible for VCR resistance, we examined the efflux capability of parental and resistant cells using a chemical VCR surrogate, calcein AM (Fig 3). Calcein AM did not accumulate in VCR-resistant cells, but did so in parental, drug sensitive cells. In addition, the efflux in ABCC1-expressing, drug resistant cells could be blocked by verapamil, a known inhibitor of ABCC1 (Fig. 3) [22]. These results show that VCR-induced chemo-resistance in Jurkat is mediated by the specific up-regulation of ABCC1, which then results in increased surface expression of function ABCC1 efflux protein.

Silent interfering RNA inhibition of ABCC1 expression restores drug sensitivity
We performed siRNA experiments to show that the elimination of ABCC1 expression would return drug sensitivity. We exposed cells to siRNA directed against ABCC1 and quantified their growth over 6 days. Our measurements of ABCC1 mRNA levels in resistant cells showed that ABCC1 expression remained high, allowing cells to maintain proliferative growth. When ABCC1expression was markedly reduced, sensitivity to the cytotoxiic effects of VCR was restored (Fig. 4A). To confirm the knock-out effects of siRNA, we measured the expression level of the ABCC1 mRNA over an 8-day period. Following a siRNA hiatus of 6 days, ABCC1 mRNA expression increased to pre-siRNA levels (Fig. 4B). These results suggest that a single multidrug resistance transport protein (ABCC1) can be selectively upregulated by VCR, which has been previously shown for ABCB1 in other tumor models. [23]

VCR resistant cells show multidrug resistance
Previously, it has been shown that a single agent can induce resistance to several dissimilar drugs. Therefore, we tested parental and drug-resistant cell lines for resistance to several common chemotherapeutic agents. We confirmed that VCR-resistant cells also displayed increased resistance to daunorubicin and prednisone, but not to L-asparaginase (Supplemental Table and Supplemental Figure 1A and B). In addition, the resistance to daunorubicin and prednisone increased mildly from 3.11 nM to 8.06 nM, and 196 μM to 545 μM, respectively, from VCR naïve to VCR-resistant Jurkat cells. These findings are consistent with the extensive observations, that ABC proteins can extrude a wide variety of molecules, hence the earlier designated name for ABCC1, "multi-resistance protein 1(MRP1)" [3,8,24]. In the resistant cells, L-asparaginase shows very minimal change, as it is not transported by ABC proteins [8]. Taken together, the results described for Figs 2 -5 demonstrate two important findings: first, resistance to VCR was selectively mediated by ABCC1, and secondly, up-regulation of ABCC1 resulted in multi-drug resistance that included DNR and PRED, but not L-ASP. Importantly, these drugs are commonly employed in the induction phases for newly-diagnosed and relapsed T-ALL patients.

ABCC1 up-regulation is associated with treatment failure in patients treated for T-ALL
To investigate the clinical significance of ABCC1 up-regulation in a cohort of patients with newly diagnosed T-ALL, we next investigated de novo ABCC1 RNA levels in a cohort of 100 patients treated on COG T-ALL Phase III studies 9404 and AALL0434. The patient's outcome status was characterized by those for whom induction failed, those who achieved a first remission, but later relapsed, and those who remained in complete continuous remission for at least four years from initial diagnosis. As shown in Figure 5, a subset of 11 patients for whom induction failed had an RMA-normalized ABCC1 level of 9.695 ± 0.1646. This level was significantly higher than the 20 patients who relapsed, having a mean ABCC1 level of 8.883 ± 0.1573 (unpaired, two-tailed t test, P = 0.0025), or in comparison to 69 patients who remained in long-term remission, expressing a group mean level of 9.141 ± 0.0687 (unpaired, two-tailed t test, P = 0.0033). While over-expression of ABCC1 could not alone explain disease progression amongst individual patients, these data show that de novo drug resistance can be explained in part by a group mean over-expression of ABCC1 within these outcome subsets.

Discussion
ABC transport proteins mediate drug resistance by exporting chemotherapeutic agents out of proliferating cancer cells, including malignant T-lymphoblasts [3,5,[7][8][9]. We have previously reported that ABCB1 (MDR1) is selectively upregulated in T-ALL cells that have been exposed to sub-lethal concentrations of anthracyclines [8]. To further investigate the roles of structurally and functionally dissimilar ABC transporters in mediating drug resistance, we maintained cells in sub-lethal concentrations of VCR, a chemotherapeutic agent that is universally included during induction therapy for T-ALL and T non-Hodgkins Lymphoma [14,16,25]. Using molecular, cellular and functional assays, we identified that ABCC1 (MRP1) was specifically upregulated under VCR-induced cytotoxic pressure, but conferred resistance to other induction phase drugs, including PRED, and to a lesser extent, DNR. Our siRNA analyses showed that ABCC1 up-regulation in T-ALL cell lines resulted in profound resistance to VCR, suggesting that under conditions of that mimic sub-lethal drug exposure, ABCC1 may play a key role in mediating disease progression. In this study, our subsequent analyses of the patient samples indicated that ABCC1 was over-expressed in patients who suffered disease progression with induction therapy. Because second-line, nonablative therapies for induction failure are often unsuccessful [12, 26,27], our findings underscore the importance of early identification of such patients so that their treatment might be modified to utilize agents that bypass ABCB1 and ABCC1-mediated drug efflux.
For patients who do not achieve a first remission, de novo resistance is thought to be mediated by a number of biological events that control cell proliferation, including a disruption of cell division at the G 0 /G 1 checkpoint, [17] molecular aberrations that prevent therapy-induced apoptosis [28], or by extrusion of chemotherapeutic drugs by ABC transporters [8]. Overexpressed genes within our previously-described 116 member classifier and those over-expressed within the Jurkat VCR-resistant cell line included TMSB15A (205347_s_at), TJP2 (202085_at) and CDK1 (23153_at). These genes are involved in mediating cytoskeletal integrity (TMSB15A), tight junction protein assembly (TJP2) and cell cycle progression (CDK1), all of which are important in dividing cells. Our experiments with sub-lethal VCR dosing resulted in highly VCR-resistant T-ALL cells, suggesting that in some patients for whom VCR dosing is not optimized, or for patients having primary VCR resistance, relapse or disease progression may be further complicated with the development of a multi-drug resistance phenotype early in treatment. Others have suggested that VCR under-dosing may be an important contributor to relapse, calling for further efforts to optimize VCR dosing in the treatment of ALL [29,30]. In contrast to the findings of Efferth et al [7], we found that under conditions of selective VCR exposure only ABCC1, and not ABCB1, was significantly up-regulated in our T-ALL cell line model. We are the first to demonstrate that ABCC1 contributes to de novo resistance against VCR in T-ALL, leading us to investigate its role in combination with other ABC transporters.
Multidrug resistance in T-ALL is conferred by a number of ABC transporters, which can mediate resistance to anthracyclines (ABCB1, ABCC1), methotrexate (ABCC1, ABCC2, and ABCG2) [31,32] and VCR (ABCC1). Compared to other ABC transporters, ABCC1 has several unique properties. While there is overlap between the xenobiotic substrates that are transported by ABCB1 and ABCC1, ABCC1 transports relatively broader ranges of compounds, including organic anions, heavy metals and vinca-alkaloids [31,33]. In a recent study of 72 T-ALL patients treated on DCOG ALL-7 -8 and -9, van Grotel and co-workers found CD34 expression was an important predictor of a poor 5 year disease free survival, but did not find an association between outcome and greater than median quantitative RT-PCR expression of MDR1/ABCCB1 and MRP1/ABCC1 [12]. In their study, patient outcomes were not described for early and late relapse, and since induction failure occurs in small subset of 2 to 5% of patients, over-expression of ABCB1 and ABCC1 in patients with early treatment failure may have been missed. Plasschaert and colleagues found that ABCB1-dependent disease resistance was prevalent in adults with T-ALL, but less so in children [10]. Similarly, Wuchter and coworkers [11] found no evidence for ABCB1mediated chemotherapy resistance in nine children having T-ALL for whom induction failed when treated on ALL-BFM 90/95. All patients in our studies were age 18 or less, and we used expression profiling to identify over-expression of ABCB1 and ABCC1, not the rhodamine-123 efflux assay [34]. Differences in patient age and assay sensitivity may account for discrepancies between our study and the findings of others. Among the patients with over-expression of ABCC1 transporter (Fig. 6), one later relapsed on therapy, and three remained in CCR for more than four years, suggesting that additional molecular or phenotypic features may modulate ABC-mediated therapy resistance.
Patients with a phenotype characterized by CD1a (−), CD8(−), CD5(weak) with stem-cell or myeloid markers have been designated to have an 'early T-precursor' (ETP) T-ALL [35]. Such patients have been found to have very poor outcomes, and either fail to achieve a first remission or relapse early in the treatment course. In most cases, they have had high MRD levels at the end of induction. The molecular features of ETP patients have yet to be completely defined, but may be characterized by specific expression profiling features, an absence of biallelic TCRgamma deletions (ABD), or have mutations of the MEF2C and PTEN genes [15,28,36]. We did not have access to the surface phenotypic features of those patients. However, a sub-set of our patient cohort were analyzed for molecular mutations by Gutierrez and colleagues [15]; interestingly, all patients for whom induction failed and had molecular evidence for ABCC1 over-expression were also found to have an ABD and an ETP gene expression signature. One patient with ABCC1 over-expression and yet remained in CCR had an ABD, but did not have an ETP expression profile. In some cases, patients with an ETP expression signature did not overexpress ABCB1 or ABCC1, and did not fail induction therapy. These findings suggest that ETP patients have a variety of molecular aberrations, as recently extensively documented by Zhang and co-workers [37]. In several tumor models, cells arrested at early stages of development (e.g. patients with ETP) have been reported to have ABC transport activities, possibly to protect developing cells from cytotoxic damage, including chemotherapeutic agents [38,39]. We speculate that the genotypic mutations that result in ETP or induction failure T-ALL may also maintain affected progenitor T-lymphoblasts in a stem-cell like state, in which ABC transport molecules are highly expressed, resulting in de novo chemotherapy resistance. The collaborating roles between ABC transport proteins and other molecular features associated with induction failure or early relapse will require further study as potential biomarkers for high-risk disease.
Until recently, re-induction regimens containing VCR, DNR and other ABCC1 substrates produced only transient remissions in a fraction of patients, calling for the introduction of a novel set of agents that were not ABC substrates. Nelarabine, an RNA nucleoside analog that is not known to be a substrate of ABCC1, induced remissions that were achieved and maintained in up to 50% of first-time relapse patients, suggesting that relapse regimens must utilize cytotoxic drugs that might escape the transporting effects of ABC proteins [40]. Our findings suggest that patients who are at risk to not achieve a first remission should be identified early in the course of induction, as they may benefit from alternate treatment strategies that bypass ABC-mediated drug efflux.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material.