ABCG2 is an efflux transporter responsible for inducing multidrug resistance (MDR) in leukemic cells; through its ability to extrude many antineoplastic drugs, it leads to AML resistance and/or relapse. Moreover, ABCG2 may be co-expressed with other MDR-related proteins and is finely regulated by epigenetic mechanisms.
1. Introduction
Acute myeloid leukemia (AML) is an aggressive heterogenous disease arising from the accumulation and clonal expansion of somatic-driven mutations in CD34+/CD38- hematopoietic progenitors, which demonstrate increased proliferation, survival, and impaired maturation capacity
[1][2]. It is the most common form of acute leukemia in adults, with a median age at diagnosis of 68 years and a sharp increase in incidence in the following decades
[3]. In addition, AML prevalence is increased by therapy-related AML, which accounts for 10–15% of newly diagnosed AML, an effect of improved survival after anti-cancer therapies
[4]. The estimated 5-year overall survival (OS) of AML is around 30%, with great differences between age groups (≈50% in younger patients and only ≈10% in elderly patients) and with disappointing progress over the past five decades, particularly in the older population
[5][6].
Despite this, the occurrence of intrinsic or acquired drug resistance still results in AML being an aggressive disease and a major challenge for clinicians. Clinical multidrug resistance (MDR) is a multi-factorial phenomenon depending on host variations, drug–drug interaction, deregulation of cell death mechanisms, failure of DNA damage response and repair, epigenetic alterations, intratumor heterogeneity, and microenvironment alteration. MDR protects leukemia cell from immune surveillance and from the alteration in intracellular drug concentrations due to overexpression of membrane drug transporter proteins
[7]. Acquired MDR has been intensively studied, and the molecular basis of this phenomenon is well established
[8]. The cross-resistance to different, structurally unrelated anti-cancer drugs, such as vinblastine, vincristine, and daunomycin, developed by Chinese hamster lung cells grown in actinomycin D to select resistant cells, was first described more than 50 years ago
[9]. A few years later, another study reported that daunomycin was actively transported out of multidrug-resistant cells
[10]. The hypothesis of a promiscuous membrane transporter able to confer MDR was confirmed by the identification in Chinese hamster ovary cells of the “P-glycoprotein”, so called for the altered membrane permeability associated with its expression in resistant cells
[11] and by characterization of its encoding gene
[12]. The human homologue gene was soon identified and referred to as ATP-binding cassette (ABC) subfamily B1, ABCB1
[13]. Its recognition paved the way to studies of ABC transporters, leading to the identification of 48 human membrane proteins, grouped into 7 subfamilies, involved in different physiological biochemical and developmental processes beyond cancer drug transport
[14][15]. The ABC superfamily is highly conserved among plant and animal species, mainly acting as import pumps in prokaryotes
[16]. In eukaryotic cells, they act as exporters, pumping out substances from the cytoplasm or entrapping them into intracellular organelles, such as peroxisome, endoplasmic reticulum, or lysosomes, using an energy-dependent process involving binding and hydrolysis of ATP
[17][18].
The general architecture of ABC proteins consists of two cytoplasmic nucleotide-binding domains (NBDs), which bind and hydrolyze ATP, and of two sets of hydrophobic transmembrane domains (TMDs), which transport substrates. ABC genes encode either a full transporter or a half transporter with a single TMD domain and a single NBD. Half transporters must dimerize as either homo- or heterodimers to form an active protein
[15]. Whereas the structure and function of NBDs are similar throughout ABC subfamilies, TMDs are highly heterogeneous, thus permitting the recognition of different substrates and their translocation across membranes, irrespective of concentration gradient
[19]. Despite the huge amount of data on ABC structure obtained by electron microscopy, the precise translocation mechanism remains elusive. It should be underlined that among the 48 ABC members, some have “narrow” substrate specificity, while others (19 of the 48) have broad specificity and are able to transport a wide range of anticancer drugs and to cause drug resistance if overexpressed in tumor cells
[20]. The study of “null” mutants established via germline mutations has underscored the diversity of their physiological role and the consequences of their dysfunction. So, at present, more than 20 ABC proteins, belonging to all the identified sub-families, have been associated with human diseases
[15][17]. The most important steps in ABC family discoveries are reported in
Figure 1.
Figure 1. A schematic chronology of significant discoveries of ABC transporters involved in multidrug resistance.
2. ABCG Subfamily
2.1. ABCG1
ABCG1 is coded on chromosome 21q.22.3 and contributes to cholesterol transport and to cellular cholesterol homeostasis
[21]. The protein is expressed in many cell types, including endothelial cells, lymphocytes, and myeloid cells on the cell membrane and in endosomes
[22]. ACBG1 seems to regulate T cell development in the thymus by regulating intracellular cholesterol levels
[23][24]. Moreover, ABCG1 is involved in innate immune response by the regulation of inflammation via reduction of inflammatory cytokines, and in anti-tumor immunity by favoring IL-4–mediated macrophage M2 polarization, producing a pro-cancer effect
[25]. Moreover, it seems to be upregulated in lung cancer tissue, and aberrant expression of ABCG1 in lung cancer cells promotes proliferation, migration, and tumor invasion
[26].
2.2. ABCG4
ABCG4 is mainly localized in the central nervous system (CNS) and seems to have a protective role in Alzheimer’s disease through inhibitory effects on amyloid β production
[27]. Furthermore, ABCG4 overexpression seems to confer drug resistance, despite the mechanism not being completely understood
[28].
2.3. ABCG5 and ABCG8
ABCG5 and
ABCG8 form an obligated heterodimeric complex (G5/G8), highly expressed on epithelial cells of the intestine and liver, where it mediates sterol transport
[29]. Gene variations are associated with hypercholesterolemia, platelet dysfunction, sitosterolemia, cardiovascular disease, and gallstones
[30][31][32].
2.4. ABCG2
ABCG2 is the most studied among ABCG members. In the early 1990s, the observation of MDR in cell lines selected with mitoxantrone lacking MDR1 and MRP1 expression led to identification of a new transporter protein. The gene responsible for the novel resistance phenotype, first cloned by Doyle et al., in the MCF-7 Adr/Vp cell line, was named
BCRP for breast cancer resistance protein, from the cell line origin
[33][34][35], and further designed as
ABCG2 by the Human Genome Organization Committee. The
ABCG2 gene is highly conserved among species, most of which have a single gene present
[36]. The exceptions are rodent and fish, which have more
ABCG2 genes
[37]. The human ABCG2 gene is located on chromosome 4, band 4q21–4q22, and extend over 66 kb containing 16 exons (range size from 60 to 532 bp) and 15 introns. The translational start site is on exon 2, ABC signature motif in exon 6, and the ATP binding sites (Walker motif A and B) in exon 3 and exon 6. The promoter region is located approximately 312 bp from the transcriptional start site
[38].
In vitro studies demonstrated the upregulation of the
ABCG2 gene under hypoxic conditions by estradiol progesterone and by aryl hydrocarbon receptor agonists
[39][40][41][42].
ABCG2 expression can also be induced via peroxisome proliferator-activated receptor γ (PPARγ)
[43] and downregulated by dexamethasone via the glucocorticoid receptor (GR)
[44]. However, data on
ABCG2 regulation are often controversial, and it has been hypothesized that the observed contradiction may be due to cell- or organ-specific regulation. Epigenetic regulation has been observed in overexpressing cell lines, where elevated
ABCG2 levels were associated with hypomethylation or unmethylation of the CpG island and with hyperacetylation of the
ABCG2 promoter
[45].
At least one dimerization is required to produce a functional protein; dimerization is under the control of the endoplasmic reticulum quality control (ERQC) network. If passing the ERQC, protein traffics to the Golgi apparatus, where the protein becomes fully glycosylated, passes a further quality control, and can be delivered to its destination, the plasma membrane. Mechanisms by which the protein is transferred to the cell membrane are not completely elucidated and may include direct delivery, trafficking via the endosomal pool or trans-cytotic pathway via the basolateral membrane. Surplus protein is proteolytically degraded in lysosomes, and misfolded proteins become ubiquitinated and degraded in proteosome. Interestingly, 40–60% of the produced protein, even in wild form, does not pass ERQC and is eliminated
[46]. Differently to ABCG5/ABCG8, ABCG2 may work only as homodimer; however, several studies have reported that it can assemble in higher oligomeric forms, from tetramers to dodecamers, a process sometimes favored by the presence of single-nucleotide polymorphisms
[46][47].
2.5. ABCG2 Substrates
Since ABCG2 was first described, the list of its substrates has been steadily expanding. The substrate specificity of ABCG2 is highly overlapping to that of ABCB1, and like ABCB1, ABCG2 preferentially targets hydrophobic and lipophilic compounds with planar aromatic systems. Transfer across the membrane is associated with conformation changes of the protein (“in-facing”, with the substrate binding site open in the cytoplasm, and “out-facing” open in the extracellular space). Recently, Gyöngy et al., and Yu et al., proposed two molecular models to explain drug transport, highlighting the crucial role of ATP binding to modulate ABCG2 conformation
[48][49]. As for the other ABC members, the molecular basis of ABCG2 substrate specificity is not fully elucidated. It has been hypothesized that substrate binding depends on the formation of a “membrane entrance” in the lipid bilayer by hydrophobic amino-acid residuals, available in the “in-facing” protein conformation, and that the different combination of these residuals provides the substrate specificity
[50].
Mitoxantrone transport is the hallmark of the cells expressing ABCG2; thus, the first recognized substrates were predominantly chemotherapy agents. More recently, other classes of substrates have been identified, including antivirals, HMG CoA inhibitors, flavonoids, carcinogens, and calcium channel blockers
[51][52].
3. Expression and Clinical Significance of ABCG2 in AML
Despite the role of ABC proteins in determining drug resistance in hematologic and solid cancers, it is still a matter of controversy, and many studies in the past decades have shown a relationship between ABCG2 overexpression and poor clinical outcome in AML. The heterogeneity of the employed methods, including ABCG2 m-RNA expression, protein evaluation by flow cytometry or immune-cytochemistry, ABCG2 efflux of fluorescent substrates, and the lack of standardization may, in part, explain the confounding results and make data comparison difficult.
In 2001, Sargent et al., evaluated ABCG2 expression by immunocytochemistry, using the anti ABCG2 BXP-34 monoclonal antibody, in 20 samples of de novo AML (12 previously treated and 8 untreated). ABCG2 positivity showed high variability, but 27% patients had more than 10% positive cells and were considered as ABCG2 overexpressing. There were no differences between pretreated and naïve patients with regard to FAB cytotype or other clinical/biological characteristics.
Van den Heuvel-Eibrink et al., compared ABCB1, ABCC1, LRP, and ABCG2 mRNA levels in 20 leukemia samples at diagnosis or relapse, observing significantly higher ABCG2 mRNA levels at relapse (median 1.7-fold,
p = 0.04). On the contrary, expression levels of the other tested ABC proteins did not change. They hypothesized that only ABCG2 accounted for drug resistance at relapse
[53]. In 2007, the same group explored the relevance of the same genes’ (ABCB1, ABCC1, LRP, and ABCG2) expression in a cohort of 154 elderly patients, observing a negative correlation between ABCB1 and ABCG2 expression and WBC count (
p = 0.001) and a positive association between ABCB1 and ABCG2 and CD34 blast expression (
p = 0.001). Moreover, high ABCB1/ABCG2 expression significantly reduced complete remission (CR) rate (
p = 0.03) and seemed to be associated with a reduced event-free survival (EFS) (
p = 0.05)
[54].
The prognostic significance of ABCG2 expression in pediatric AML was assessed by Steinbach et al., The authors evaluated ABCG2 expression by real time PCR in 59 pediatric cases, reporting 10-fold higher mRNA levels in patients not in remission after induction therapy compared to those achieving CR (
p = 0.012). The highest expression was observed in the M1-M2 FAB subtype and the lowest in M5 (
p = 0.004). Worse OS was reported in ABCG2 overexpressing cases, irrespective of disease risk (
p = 0.023)
[55]. The same group also investigated the prognostic relevance of the co-expression of ABCB1, ABCG2, ABCC3, and ABCA3 in 112 children with AML treated according to the AML-BFM 2004 protocol. Patients with high levels of ABCG2 and ABCC3 had reduced CR rate, and those with ABCG2 overexpression also had lower DFS. As for CR probability in adults (Liu et al.
[56]), DFS in this pediatric cohort was negatively affected by the number of overexpressed ABC transporters (
p < 0.001)
[57].
Taken together, this data supports a role of ABCG2 in affecting AML outcome, both in adults and in children. Poor response to induction therapy may be also influenced by the frequent co-expression of ABCB1, but the stem cell–like properties conferred to leukemic cells by ABCG2 overexpression may account for high relapse rate and for poor survival when induction therapy is intensified or despite allogeneic stem cell transplantation, even if performed in CR.
4. ABCG2 Polymorphisms in AML
The cloning of ABCG2 DNA from drug-selected cell lines and from normal tissues revealed many amino-acid substitutions able to alter the protein function and the substrate preference. Moreover, mutational analysis identified more than 80 ethnic-associated synonymous and non-synonymous single-nucleotide polymorphisms (SNPs) potentially influencing ABCG2 expression and function and affecting drug absorption, plasma concentration, and distribution and elimination (
Figure 1)
[58].
In drug-selected cell lines such as S1-M1-80 and MCF7/AdVp3000, unique mutations in amino acid position 482 makes cells highly resistant to mitoxantrone and doxorubicin. The replacement of Arg with Gly or Thr at position 482 increases rhodamine and anthracycline efflux compared to the wild-type counterpart
[59]. In contrast, the R482G and R482T variants negatively affect ABCG2’s ability to transport methotrexate but confer increased methotrexate resistance
[60]. At least 13 ABCG2 variants with substitution at R482 have been described in cell lines, all associated with strong resistance to mitoxantrone. The COO- terminus of TMD3, near position 482, and 3D homology models suggest that R482 is the central cavity of the binding pocket, with a crucial function in drug transmembrane translocation
[61]. Mutations at N557 and H630 confer lower resistance to SN-38, although mitoxantrone resistance is maintained
[62]. Mutations at C603 may impair the homodimer formation
[47], and substitutions at N596 may affect N-linked glycosylation, reducing the amount of ABCG2 on the cell membrane
[63]. Among synonymous SNPs, Q141K has been associated with increased risk of gout in Asian
[64] and American populations
[65].
5. ABCG2 and Extracellular Vesicles
Among the factors mediating the acquisition of the MDR phenotype, the role of intercellular communications has recently emerged. In this setting, extracellular vesicles (EVs) play a pivotal role
[66]. EVs are a heterogenous group of lipid bilayer structures derived from either endosomal multivesicular bodies (exosomes) or from the plasma membrane (micro vesicles, also called ectosomes). Cells that secrete more vesicles show a greater level of resistance
[67]. Many studies have shown that cytotoxic drugs may be sequestered into EVs and released from the cells, thus preventing their accumulation in the nucleus. This mechanism has been demonstrated in MCF7 breast cancer–resistant cells, in which ABCG2 efflux protein localizes to EVs and mediates the uptake of drugs in the vesicles before their release
[68]. The compartmentalization of ABCG2 in EVs and not in other intracellular compartments depends on the activation of the PI3K/AKT pathway, suggesting a potential therapeutic target to overcome drug resistance by the inhibition of EVs biogenesis. Preliminary in vitro studies using the specific PI3K-AKT axis results in the reduction of EV number and volume. ABCG2 is relocated to the intracellular compartment and loses the ability to concentrate anticancer drugs, thus restoring cell sensitivity
[68]. EVs may act also by transferring between cells the efflux transporters or mi-RNAs involved in the regulation of efflux proteins’ expression
[69].
6. ABCG2 Inhibition
Different approaches to overcome ABCG2-mediated MDR have been proposed. Despite some positive results obtained in preventing ABCB1 binding
[70][71], the attempt to develop chemotherapy agents that are not recognized by ABCG2 remains challenging, due to the molecular variety of transported compounds and the still incomplete knowledge of binding mechanisms
[72]. The attractive strategy to induce collateral sensitivity (CS), a well-known phenomenon in which, due to overexpression of ABC transporters, a MDR cell becomes hypersensitive to some unrelated anticancer drugs, is difficult to pursue due to the variety of mechanisms involved in drug resistance
[73].
So far, the most used method remains an inhibition of the efflux function, re-sensitizing resistant cells to conventional anticancer drugs. The first identified functional inhibitor of ABCG2 was fumitremorgin C (FTC), a mycotoxin produced by
Aspergillus fumigatus, extensively used in experimental settings to test drug sensitivity in ABCG2 overexpressing cells
[74].
It must be underlined that many of the molecules with inhibitory activity work only at high concentrations, often not attainable in clinical use. Among novel molecules, Ko143 (an FTC derivative), chromone derivatives, and a recently discovered indenoindol derivative are the most promising agents, having an IC50 in the nanomolar range
[75]. Many ABCG2 inhibitors, like calcium channel blockers, anti-HIV drugs, and xanthine-oxidase inhibitors, were selected by drug repurposing, with the potential advantage of shortening the drug development process. The most interesting are tivozanib, fostamatinib, ponatinib, and febuxostat, all active at nanomolar IC50
[76][77][78][79].
Other new anticancer drugs have recently demonstrated an ABCG2 inhibitory effect in vitro. Sorf et al., demonstrated that ribociclib, a C4C/6 inhibitor approved for the treatment of locally advanced/metastatic breast cancer, inhibits ABCB1- and ABCG2-mediated daunorubicin and mitoxantrone efflux in AML cell lines at IC50 ranging between 1,4–3 μM, suggesting that combination therapy can revert MDR, especially in CD34+/FLT3-WT cases
[80]. In vitro inhibition of ABCG2- and ABCC1-mediated daunorubicin and mitoxantrone efflux was demonstrated also by talazoparib, a drug approved for metastatic BCRA1/2 mutated breast cancer
[81].
Despite the discovery of several molecules with an inhibitory effect, the interest in the clinical development of effective ABCG2 inhibitors has been deprioritized, likely due to the disappointing results attained in ABCB1 inhibition, and at present no clinical trials including ABCG2 inhibitors are ongoing. This is in contrast with the evident contribution of ABC transporters in chemotherapy failure, at least in AML. In this setting, ABCG2 overexpression not only accounts for an increased relapse risk and poor survival after conventional therapy, but also identifies a subset of patients at higher risk of relapse after allogeneic transplantation, which is still recognized as the only “curative” option for high-risk disease
[82]. It is possible that the acquisition by ABCG2-overexpressing leukemic cells of stem-cell-like properties eventually favors their survival in transplant preparative regimens and their escape from post-transplant graft versus the leukemia effect. On this basis, strategies to counteract ABCG2 should be adopted.
This entry is adapted from the peer-reviewed paper 10.3390/ijms24087147