Molecular Pathways: Regulation and Therapeutic Implications of Multidrug Resistance

The expression, function, and modulation of the MDR transporter gene ABCB1 is one of the most extensively studied mechanisms of drug resistance in human cancers (1–5). The role of P-glycoprotein (P-gp) is well established in hepatic drug excretion, limitation of gastrointestinal absorption of substrate drugs, and as a key component of the blood–brain, blood–testicular, and blood–placental barriers (3, 6–9). ABCB1 transfection and its inhibition by gene knockouts, RNA silencing, and small-molecule transport inhibitors have proven its function as a determinant of resistance to various anticancer drugs, such as anthracyclines, taxanes, vinca alkaloids, and others, including a large fraction of new drugs in development (1, 8, 9). Among many clinical trials with various inhibitors of P-gp in attempts to reverse drug resistance, the large majority have been negative, and only a few show some evidence of clinical benefit (2).

Despite the significant advances in our understanding of the functional aspects of P-gp, however, the regulation of the ABCB1 gene and its impacts on clinical trials are not well elucidated. In this review, we mainly focus on our recent understanding of genetic and epigenetic aspects of ABCB1 expression, and their cooperation with transcription factors on the ABCB1 gene. We shed light on how molecular signaling coordinated with epigenetic mechanisms controls ABCB1 expression. These regulatory insights derived from the ABCB1 model may also be useful for deciphering non-ABCB1–mediated drug resistance and for prevention and molecular intervention of MDR in the clinic.

Mechanisms underlying the intrinsic genomic instability of cancers have been an important area of research. Using Luria-Delbruck fluctuation analysis in the human uterine sarcoma cell line MES-SA, we showed that spontaneous mutational events underlie acquired MDR in these cells, and activation of ABCB1 is a frequent mechanism for cellular resistance to doxorubicin, paclitaxel, and vinblastine (10–13), but not to etoposide (14). Structurally, Huff and colleagues proposed that gene rearrangements serve as an important mechanism to control ABCB1 expression in many cases by juxtaposition of the native ABCB1 promoter to transcriptionally active promoters of many unrelated genes (15). This mechanism was also found in samples from both leukemias and lymphomas in some patients with cancer, suggesting its potential clinical importance (16). Several oncogenes including Ras, c-Raf, and c-Raf kinase have been implicated in ABCB1 regulation (17–19). These studies suggest that oncogenes play a role in regulating the expression of ABCB1, possibly via malignant transformation processes.

During cellular transformation and tumor progression, one of the most commonly observed alterations is mutation in the TP53 tumor suppressor gene. There is substantial evidence that p53 is involved in regulation of ABCB1. The general observation is that the wild-type p53 represses and that p53 mutations enhance ABCB1 promoter activity (5, 17, 20, 21). Moreover, the interacting site of p53 on the ABCB1 promoter has been identified (22). Hence, alterations or interactions in the p53 system have linked MDR to the p53-mediated signal pathway. Suppression of Mdm2 p53-binding protein homolog (MDM2), a potent inhibitor of p53, by Nutlin-3 inhibits the functions of P-gp and ABCC1 (MRP1) in drug-resistant neuroblastoma and rhabdomyosarcoma cell lines (23). These studies suggest that the p53 system is involved in the regulation of ABCB1, possibly by direct interaction with the ABCB1 promoter. Thus, restoration or mimicry of wild-type p53 function may be a strategy to reduce or prevent MDR and sensitize cancer cells to anticancer drugs.

ABCB1 expression was also shown to be associated with demethylation of the ABCB1 promoter in the promyelotic leukemia cell line HL-60 after drug selection (24), human T-cell leukemia cells treated with the demethylating agent 5′-azadeoxycytidine (25), acute myelogenous leukemia (AML) clinical samples (26), adult acute lymphocytic leukemias (27), and bladder cancers (28). These data suggest that demethylation of the ABCB1 promoter activates ABCB1 expression in cancer cell lines and also in clinical cancer patients.

Jin and Scotto further showed that both histone acetyltransferases (HAT) and histone deacetylases (HDAC) regulate ABCB1 mRNA levels in SW620 (human colon carcinoma) cells, when the cells were transiently treated with the HDAC inhibitor trichostatin A (29). We have shown a 20-fold increase in acetylated H3 (acetyl-H3) in the nucleosomes within the 968-bp region of the MDR1 far upstream promoter (P2), but not within the native ABCB1 promoter (P1), in doxorubicin single-step selected mutants of MES-SA cells (13). Reciprocally, a 35-fold increase in acetyl-H3 was found in the ABCB1 proximal promoter (ABCB1 P1), but not in the P2 promoter, in step-wise selected MES-SA/Dx5 cells (13). The detailed molecular events that are responsible for this differential utilization of the ABCB1 promoters are unclear. However, this is the first epigenetic evidence showing that different cytotoxic regimens could result in distinct routes that lead to differential ABCB1 activation. The importance of acetyl-H3 in the regulation of ABCB1 was also confirmed by an independent study, in which Baker and colleagues showed that upregulation of ABCB1 by anticancer drugs was associated with a dramatically induced acetyl-H3, but not acetyl-H4, on the MDR1 P1 promoter in CEM-Bcl2 and in SW620 cells by 24-hour drug induction (30). Similarly, a repressive histone marker, trimethylated H3-Lys9, was found to be associated with the expression of another multidrug transporter ABCG2 (31). These studies indicate that the acetylation status of H3 plays an important role in the regulation of some ABC transporters.

A wide-ranging evaluation of the roles of transcription factors in the regulation of ABCB1 was provided by Scotto (5). Here, we focus on several transcription factors that are associated with either chromatin remodeling or epigenetic interaction with the ABCB1 promoter. In general, a chromatin-remodeling process is believed to be the initial step to induce gene expression. We showed, by transient electroporation of C/EBPβ expression plasmids into the cells, that C/EBPβ [nuclear factor for interleukin-6 (NFIL6)], a member of the CCAAT/enhancer-binding protein family, can induce expression of the silent ABCB1 gene in MCF-7 breast cancer cells (32). C/EBPβ has been reported to upregulate ABCB1 promoter activity via a sequence motif resembling the C/EBPβ DNA-binding sequence (33). However, we revealed a novel mode of action for this protein in the regulation of ABCB1 expression by showing that C/EBPβ interacts with the ABCB1 chromatin that contains an inverted CCAAT-box (Y-box) in a cell-type–dependent manner (32). The behavior of C/EBPβ on the ABCB1 promoter is consistent with the capacity of its N-terminal transactivation domain to recruit the human SWI/SNF chromatin-remodeling complex (34). Our study suggests that some transcription factors, such as C/EBPβ, are directly involved in the development of acquired MDR by transactivating endogenous ABCB1 (32).

Moreover, the human Brahma protein (hBrm), a component of the human SWI/SNF complex, is also implicated in the upregulation of ABCB1 promoter activity, possibly via cooperation with C/EBPβ on the Y-box in MCF-7 cells (32). Whether the hBrm-C/EBPβ complex represents a positive or negative regulator for the chromatin-embedded MDR1 gene was not determined in MCF-7 cells in the previous study (32). The association of the SWI/SNF complex (containing hBrm) on the ABCB1 promoter was confirmed by Harikrishnan and colleagues (35). In their study, hBrm seems to be complexed with the methyl-CpG–binding protein (MeCP2) to repress the endogenous ABCB1 promoter in CEM-CCRF cells. In general, we believe that the regulation of ABCB1 expression either by C/EBPβ or hBrm alone or by the C/EBPβ-hBrm complex is both cellular-context and chromatin-status dependent. The regulatory complex may be controlled indirectly by autocrine production of interleukin (IL)-6, which was shown to be associated with ABCB1 activation via a C/EBPβ pathway in breast cancer cells (36).

The interactions among DNA methylation, histone modification, chromosomal remodeling, and transcription factor positioning are highly regulated to achieve a desired phenotype. In the case of ABCB1, a hypermethylated promoter with deacetylated H3 represents a doubly repressive status of the ABCB1 promoter, which needs to be released prior to achieving ABCB1 expression. Indeed, in P-gp–negative MES-SA and MCF-7 cells, the ABCB1 chromatin was deacetylated (13, 37). The occupancy of both MeCP2 and the corepressor MBD2 would reinforce the repressive status of the ABCB1 chromatin (37). Hence, release of both MeCP2 and HDAC1 from the methylated ABCB1 promoter modulates ABCB1 expression (38). Such a derepression mechanism facilitates acetylation of a preferred histone species (e.g., H3) at a specific residue (e.g., Lys9), thus enabling the opening of compact chromatin and positioning of general transcription factors as well as RNA polymerase II at the ABCB1 promoter (Fig. 1).

The Y-box of the ABCB1 promoter is likely a central docking site to recruit transcription factors such as p300/CREB-binding protein–associated factor (p/CAF), which has HAT activity, and to recruit C/EBPβ, which has chromatin-remodeling ability (29, 32). HAT complexes were shown to stabilize the SWI/SNF complex binding to promoter nucleosomes with specificity (39). Thus, these studies illustrate a structural and functional link among the Y-box, Y-box–binding proteins, C/EBPβ, HATs, HADCs, MeCP2, and the SWI/SNF complex on the ABCB1 promoter. They also suggest a complex interaction pattern that is crucial for the initiation of a cluster of genes, including ABCB1, which contribute to the intractability of cancer cells to various therapies (Fig. 1). Y-box–binding proteins thus represent important markers for MDR (40, 41). Disruption of the YB-1 also caused suppression of EGFR and HER-2/neu, reinforcing the putative association among the YB-1–Her2–ABCB1 network (Fig. 1).

The clinical relevance of ABCB1 expression was predominantly focused on the ABCB1-encoded P-gp, which has been used as a therapeutic target to enhance chemotherapies by inhibiting the transporter function. However, the clinical significance of P-gp inhibition as a therapeutic strategy has proved to be difficult to establish. A major confounding issue in clinical trials of MDR modulation has been the altered pharmacokinetics of chemotherapeutic drugs such as doxorubicin, daunorubicin, etoposide, and taxanes, both as a result of P-gp inhibition and as a result of inhibition of other drug transporters and CYP 3A4 (6, 42–46). In some cases, such as the use of the cyclosporine analogue valspodar (PSC-833) in AML, substantially increased toxicity was observed, resulting in an increased toxic death rate (47).

The most positive data for MDR modulation was the Southwest Oncology Group (SWOG) 9126 trial using high-dose cyclosporine infusion in relapsed and high-risk AML (48). However, other phase III studies in AML using cyclosporine or valspodar have been negative (47, 49). The discrepancies in results between these studies may be due to different patient populations, with SWOG 9126 containing a large proportion of secondary AML expressing high levels of P-gp. Patient selection based on screening for functional P-gp expression may be important. Baer and colleagues reported a trend toward benefit in a subset of AML patients with PSC-modulatable efflux, who had an increase from 5 to 14 months of median survival when valspodar was added to induction chemotherapy (P = 0.07; ref. 47). Recently, a phase III trial of the potent and specific P-gp inhibitor zosuquidar was reported as negative in older patients with AML (50). In addition to studying an unselected AML population, the schedule of zosuquidar in this Eastern Cooperative Oncology Group study was suboptimal, given the short half life of the drug (51, 52).

Incomplete P-gp inhibition, drug interactions leading to excessive toxicities, and suboptimal clinical trial designs are all reasons for the lack of success of P-gp inhibition as a therapeutic approach. However, the most significant obstacle may be the expression of multiple, redundant mechanisms of resistance in human cancers. In this regard, relatively nonspecific drugs, such as cyclosporine, may offer an advantage by inhibiting several drug transporters, although this advantage is counterbalanced by an increased potential for drug interactions and toxicities (53). Coexpression or selection for other drug transporters as alternative MDR mechanisms has been reported in the setting of inhibition of P-gp function in AML. These other transporters include ABCC1 (MRP1), ABCC2 (MRP2), and ABCG2 (BCRP/MXR; refs. 54–57). Coexpression of P-gp and other transporters has also been reported in normal and leukemic stem cells (58, 59). These data imply that the coexistence of P-gp with other transporters such as ABCG2 is a default stem cell program in AML, and perhaps in other cancers. Coexpression of ABCB1 and ABCG2 is obviously associated with the acetylation of histones, particularly H3 (13, 30, 31, 60). Thus, the role of epigenetic-based therapies in drug resistance should be further investigated using HAT (rather than HDAC) inhibitors, targeting acetyl-H3.

The use of HDAC inhibitors in treating human cancer patients may result in expression of ABCB1 or ABCG2 or both. Thus, induction of ABCB1 expression by FK228, a depsipeptide that specifically inhibits class I HDACs, was reported in normal peripheral blood mononuclear cells and in circulating tumor cells, acute promyelocytic leukemia (APL) cells, and osteosarcoma (61–63). Coadministration of all-trans retinoic acid with FK228 increased both acetyl-H4 and acetyl-H3-Lys9 at the ABCB1 promoter in acute APL cells (62). Hence, upregulation of ABCB1 by HDAC inhibitors, such as FK228, could reduce its antitumor efficacy through the classical MDR mechanisms.

Many survival signals that are associated with ABCB1 activation have been endogenously active in AML, including phosphoinositide 3-kinase (PI3K)/Akt, Raf, and PKC. These factors can also contribute to ABC transporter-independent drug resistance mechanisms. Thus, coexistence of other non-ABCB1 resistance mechanisms in AMLs elucidates the limitations of P-gp inhibitors as a treatment strategy and underlies the need both for a deeper understanding of the interrelatedness of drug resistance mechanisms and for novel strategies to deal with resistance. Such strategies might include identification of shared regulators of multiple resistance pathways and concurrent targeting of such pathways. Pharmacologic reactivation of p53 in human cancer with a mutant p53 status is one such example. In murine tumor models, restoration of functional p53 induces a senescence program (in liver carcinomas and in sarcomas) and apoptosis (in lymphoma), which consequently leads to tumor clearance or regression (64, 65). These genetic experiments in mice have proved the principle of p53-based therapy. Restoration of functional p53 might also increase therapeutic efficacy of some anticancer drugs through the suppression of ABCB1. Acetylation of p53 results in destabilization of the p53–MDM2 interaction loop, thus activating p53-mediated stress response (66). The SNP309 G/G genotype that influences MDM2 transcription levels is considered an independent risk factor in B-cell chronic lymphocytic leukemia (67) and is associated with poor survival among early-stage non–small cell lung cancer patients with squamous cell histology (68). Inhibitors that target MDM2–p53 interactions might also concomitantly target MDR mechanisms.

Multiple genetic and epigenetic regulatory mechanisms constitute a tightly coordinated network that determines predominant drug-resistant phenotypes, including classic P-gp–related MDR. We elucidate the complexities of MDR in human malignancies. The role of P-gp expression in clinical drug resistance and drug disposition is well established, and transport by P-gp is an important consideration in the development of new anticancer drugs. However, P-gp targeting to sensitize cancers to therapies has not been successful in clinical trials. The negative clinical trials are linked to limited inhibition of P-gp, excessive toxicities from inhibition of drug disposition, alternative MDR and other drug resistance mechanisms, and suboptimal clinical trial designs. Targeting the regulation of ABCB1 and related resistance mechanisms by new therapeutic approaches, including epigenetic modulators, will be subject to some of the same constraints as P-gp inhibitors. These constraints include the complexity and redundancy of drug resistance mechanisms, effects on cytotoxic drug distribution and excretion, as well as the fundamental importance of these MDR mechanisms in stem cell biology and survival.

No potential conflicts of interest were disclosed.

This work was supported by National Cancer Institute grant CA09302 (K.G. Chen), NIH grants R01 CA52168 and R01 CA92474 (B.I. Sikic), and M01 RR0070 (General Clinical Research Center, Stanford University).