ABC transporters in cancer: more than just drug efflux pumps

Abstract

Multidrug transporter proteins are best known for their contributions to chemoresistance through the efflux of anticancer drugs from cancer cells. However, a considerable body of evidence also points to their importance in cancer extending beyond drug transport to fundamental roles in tumour biology. Currently, much of the evidence for these additional roles is correlative and definitive studies are needed to confirm causality. We propose that delineating the precise roles of these transporters in tumorigenesis and treatment response will be important for the development of more effective targeted therapies.

Main

Failure of cancer chemotherapy can occur through increased efflux of chemotherapeutic agents, leading to the reduction of intracellular drug levels and consequent drug insensitivity, often to multiple agents. A well-established cause of multidrug resistance (MDR) involves the increased expression of members of the ATP binding cassette (ABC) transporter superfamily, many of which efflux various chemotherapeutic compounds from cells¹. The most extensively characterized MDR transporters include ABCB1 (also known as MDR1 or P-glycoprotein), ABCC1 (also known as MRP1) and ABCG2 (also known as BCRP or MXR) (Box 1). The discovery of ABCB1 and ABCC1 was a major breakthrough in understanding the mechanisms behind MDR and prompted the identification of numerous proteins with related structures and similar transport capabilities (Table 1). It is now known that these proteins belong to the ABC transporter superfamily, which in humans comprises at least 48 genes with diverse functions². Although it has been more than three decades since the discovery of ABCB1, and third generation inhibitors ofits drug transport function have been developed, clinical trials using these inhibitors have generally produced disappointing results. Although overcoming drug resistance through the inhibition of ABC transporters remains a priority, additional roles are emerging for these proteins that might be important for cancer initiation and progression. Validating these additional functions and elucidating their roles in tumour biology will be central to our understanding of these key molecules and could inform the development of more effective targeted therapies.

Table 1 selected ABC transporters, cancer-related substrates and their expression in tumour stem cell-like populations^*

MDR and the ABC transporter family

The ABC transporters are the largest familyof transmembrane proteins, with seven subfamilies that are designated A to G on the basis of sequence and structural homology, and are responsible for the ATP-dependent movement of a wide variety of (including drugs), lipids and metabolic products across the plasma and intracellular membranes. ABCA family members are predominantly expressed in cells of the central nervous system, as well as the haematopoietic system, and are primarily known for their roles in lipid transport and homeostasis, particularly in lipid trafficking between cellular compartments^3,4,5. In addition to their roles in drug transport (Table 1), several members of the ABCB family are responsible for intracellular peptide transport, including an essential role in major histocompatibility complex (MHC) class I antigen presentation⁶. The ABCC (also known as multidrug resistance protein) family contains the greatest number of known drug transporters (Table 1), but also includes the cystic fibrosistransmembrane conductance regulator (CFTR; also known as ABCC7), the only ABC family member known to function as a channel and which has an essential role in chloride ion efflux⁷. The ABCD gene subfamily encodes peroxisomal half-transporters, some of which have been linked to neurodegenerative diseases⁸. Members of the ABCE and ABCF families seem to have no membrane-spanning domain and are involved in mRNA translation^9,10.

Leukotrienes:

A class of arachidonic acid-derived lipid mediators involved in inflammation and homeostatic biological functions.

Lipidomics:

Systems-level analysis and characterization of cellular lipid profiles using technologies such as mass spectrometry and computational analysis.

Reconstituted vesicle systems:

Artificial membrane systems mimicking natural lipid bilayers. Used to enable the full activity of membrane proteins, which often require correct orientation and bilayer insertion.

Smouldering inflammation:

Localized chronic, and often subclinical, inflammation with few systemic manifestations.

Xenobiotics:

Exogenous chemicals, either naturally occurring or synthetic, found in a living organism.

For many years cancer researchers have focused on members of the ABC superfamily with drug efflux capacity to determine the reasons behind chemotherapeutic treatment failure. Most of these studies have examined the gene or protein expression of only a small number of transporters in tumour specimens that were obtained from patients at diagnosis, and they have correlated the expression data with clinical characteristics. Although these studies have confirmed the clinical relevance of ABCB1 in several cancers, the clinical relevance of other transporters remains uncertain¹¹. However, ABC transporter inhibition remains an attractive potential adjuvant to chemotherapy (reviewed in Refs 1, 12; Box 1).

Clinical trials involving MDR modulators

The goal of reversing MDR in the clinic through the pharmacological inhibition of specific ABC transporters has been pursued for almost two decades. Specific ABC transporters, most notably ABCB1, ABCC1 and ABCG2, are overexpressed in various cancers and can efflux a wide range of chemotherapeutic agents used to treat patients^11,12, making them attractive therapeutic targets.

Despite initial optimism, the results of clinical trials using modulators of multidrug transporters have been disappointing overall¹³. Initial clinical trials in the 1980s used compounds such as verapamil and cyclosporine A that were already approved for clinical use for conditions other than cancer and were also shown to inhibit ABCB1. Unfortunately, these first generation inhibitors had unacceptable levels of toxicity. Second generation ABCB1 inhibitors, exemplified by valspodar (also known as PSC 833), were subsequently developed and had increased potency and decreased toxicity. However, these also generated disappointing results in the clinic. For instance, a recent Phase III study in patients with ovarian or peritoneal cancer, who were being treated with paclitaxel and carboplatin as front-line chemotherapy, failed to show any benefit from the inclusion of valspodar¹⁴. Similar findings have been observed in other cancers¹⁵ and the further development of this compound is unlikely¹¹. Although third generation ABCB1 inhibitors have now been developed that improve many of the shortcomings of earlier modulators, initial results remain disappointing. For example, a recent Phase II study of women with metastatic or locally recurrent breast cancer found no additional benefit in survival with the inclusion of zosuquidar¹⁶.

Despite considerable efforts, the clinical benefit of modulating drug efflux pumps has yet to be realized, leading some investigators to conclude that the contribution of multidrug transporters to clinical drug resistance might not be particularly important¹⁷. However, several complications might have limited the clinical success of ABC transporter inhibitors to date (reviewed in Ref. 11). Foremost among these are alterations to the pharmacokinetics of anticancer agents, with decreased systemic clearance potentially leading to increased severity and incidence of adverse side effects¹⁸. This complication might apply to even the most specific of ABC transporter inhibitors, as genetic disruption of Abcb1 can result in increased drug levels in many tissues¹⁹. Such effects need be taken into consideration with any attempt to develop ABC transporter inhibitors for future clinical applications.

Functional redundancy in transporter families and the frequent overexpression of numerous transporter genes can also complicate attempts to inhibit transporter function. A recent examination of the role of ABC transporters in conferring drug resistance in the US National Cancer Institute's panel of tumour cell lines suggested that more than half the family members might have a role in conferring drug resistance, at least in vitro²⁰. In addition, as members of the ABCB and ABCC families have various roles in the immune system^6,21,22,23, another unwanted consequence of transporter inhibition might be impaired anti-tumour immune responses, as has been recently reviewed²⁴. Further exploration of the physiological roles of ABC transporters will not only raise awareness of potential side effects, but might also identify potential new therapeutic targets and approaches (discussed below).

ABC transporter gene expression, tumour progression and clinical outcome. Several observations from clinical studies correlate ABC transporters with malignant progression and a more aggressive phenotype. Clinical outcome might therefore be influenced by the ABC transporter levels that are found at diagnosis, even in the absence of a correlation between the known drug substrates of a given transporter and the drugs used to treat the cancer in question.

For solid tumours, the grade or degree of differentiation typically reflects the aggressive nature of the tumour, with less differentiated tumours possessing the greatest proliferative potential and a more aggressive phenotype. Various ABC transporters are expressed at higher levels in tumour subtypes or zones that are less differentiated. In primary untreated hepatocellular carcinoma, ABCC1 levels correlate with differentiation grade, tumour size and degree of microvascular invasion²⁵. The same study also found that ABCC1 is expressed in hepatic progenitor cells, but is absent from mature hepatocytes and is strongly upregulated in carcinoma specimens, particularly in tumours classified as having the worst prognosis. Correlation of ABCC1 overexpression with tumour size was also noted in breast cancer^26,27, and ABCC4 levels were found to decrease with differentiation of leukaemia cells to more mature leukocytes²⁸. Similarly, in early-stage lung adenocarcinoma, ABCC3 expression was high in tumours but detectable only at low levels in some precursor lesions and in the corresponding normal lung cell type²⁹. Similar trends were observed in pancreatic cancer specimens, which had significantly higher levels of ABCC3 than normal tissue³⁰, with ABCC3 expression correlating with increasing tumour grade in the 31 tumour specimens assessed.

Although ABCB1 is the most well-known drug transporter and seems to be inducible in response to chemotherapeutic treatment, its level of expression also seems to reflect the tumour phenotype in colorectal carcinomas in which ABCB1 levels correlate with invasion into vessels³¹. For those tumours expressing ABCB1, the cells at the invasive periphery had the highest levels; high ABCB1 levels were also observed in lymph node metastases³¹. ABCB1 is preferentially expressed in poorly differentiated colon tumours, but expression is undetectable in normal colon tissue³². Similarly, ABCB1 expression was highest in the largest and most aggressive tumours in a range of soft tissue sarcomas³³.

Expression of the intracellular lipid transporters ABCA2 and ABCA3 was examined in acute myeloid leukaemia (AML), and although ABCA3 expression levels correlated with poor chemotherapeutic response ABCA2 expression did not. Interestingly, however, high expression of ABCA2 was found in leukaemia cells of patients with the highest white cell counts at disease presentation³⁴, suggesting that ABCA2 levels could influence the disease pathogenesis.

A study of ABC transporter expression in primary untreated tumours from 72 patients with advanced non-small-cell lung cancer revealed that high levels of both ABCC2 and ABCG2 were predictive of poor prognosis following cisplatin treatment, and patients with tumours that stained positive for ABCG2 tended to respond less well to therapy³⁵. Although ABCC2 is a well-established transporter of platinum conjugates, these compounds are not transported by ABCG2, suggesting that ABCG2 could have a role in the response of tumours to chemotherapy that is independent of drug efflux.

The paediatric malignancy neuroblastoma is an embryonic tumour of the neural crest that is often metastatic at presentation³⁶. Poor survival rates are frequently associated with amplification of the MYCN oncogene, and the development of MDR is one of the major causes of treatment failure. Studies quantifying ABCC1 and ABCC4 expression in primary untreated neuroblastoma found that overexpression of either transporter was independently prognostic of event-free survival^37,38. The expression levels of a set of genes that included ABCC1 and ABCC4 were examined in a microarray study of an independent cohort of 251 primary neuroblastomas³⁹. Analysis of these data also showed that high levels of ABCC1 and ABCC4 were strongly predictive of adverse clinical outcome (J.I.F., M.H., M.J.H. and M.D.N., unpublished observations), whereas other ABCC subfamily members were not predictive. ABCC1 is known to efflux several drugs that are used for neuroblastoma treatment. However, the relationship between high ABCC4 expression and poor outcome was unexpected as none of the drugs used to treat the patients in either cohort is a known ABCC4 substrate, suggesting that ABCC4 might transport other molecules relevant to neuroblastoma development and/or progression.

Together, these studies indicate a propensity for high level ABC transporter gene expression in tumour cells of a more malignant, less differentiated nature in multiple primary untreated cancer types; however, these studies are largely correlative to date. One of the few examples in which an in vivo tumour model has been used to address causation comes from disruption of the Abcb1 gene in the Apc^Min/+ mouse model, in which two independent studies have shown a significant reduction in the number of intestinal polyps in the absence of Abcb1 (Refs 40, 41). Further studies in mouse tumour models, coupled with a knockdown approach or a transgenic approach, are required to address whether ABC transporters confer a fundamental biological advantage to tumour cells.

ABC transporters and hallmarks of cancer

In their seminal review in 2000, Hanahan and Weinberg elucidated six essential acquired capabilities of cancer cells: self-sufficiency in growth signals, insensitivity to anti-growth signals, evasion of apoptosis, limitless replicative potential, sustained angiogenesis, and tissue invasion and metastasis⁴². Since then, compelling evidence has led to the proposal of a seventh capability involving cancer-related inflammation⁴³. As outlined below, several observations suggest that ABC transporters might contribute to some of these cancer hallmarks. In addition, signalling lipids with established roles in tumour biology, including prostaglandins, and sphingosine-1-phosphate (S1P), are either known or strongly suspected to be ABC transporter substrates, raising the question of whether ABC transporter-mediated efflux of these molecules is important in cancer.

Leukotrienes:

A class of arachidonic acid-derived lipid mediators involved in inflammation and homeostatic biological functions.

Lipidomics:

Systems-level analysis and characterization of cellular lipid profiles using technologies such as mass spectrometry and computational analysis.

Reconstituted vesicle systems:

Artificial membrane systems mimicking natural lipid bilayers. Used to enable the full activity of membrane proteins, which often require correct orientation and bilayer insertion.

Smouldering inflammation:

Localized chronic, and often subclinical, inflammation with few systemic manifestations.

Xenobiotics:

Exogenous chemicals, either naturally occurring or synthetic, found in a living organism.

Apoptosis and proliferation. The balance between programmed cell death and proliferation regulates cell numbers. Most, if not all, cancers acquire resistance to apoptosis⁴², tipping the balance towards the expansion of tumour populations. Although the ability of ABC transporters to efflux cytotoxic drugs affects the resistance of cancer cells to apoptosis in the context of therapy, several reports provide evidence that ABC transporters might also promote cell survival independently of cytotoxic drug efflux. For example, the ABCB1 inhibitor valspodar can promote cell cycle arrest and apoptosis in human leukaemia cells as a single agent in vitro^44,45,46, and can reduce engraftment and prolong overall survival in a mouse leukaemic xenograft model⁴⁶, suggesting a dependence on ABCB1 for viability in these cells. However, another study indicates that valspodar can induce apoptosis independently of transporter inhibition in some cell lines⁴⁷. Furthermore, expression of ABCB1 has been reported to delay apoptosis in response to apoptotic stimuli in both normal and cancer cells^48,49,50,51, and ABCB1 inhibition reverses this resistance^49,52. Surprisingly, this function was suggested to be independent of active transport⁵³; however, a mechanistic explanation for these observations remains to be established, and at least one other study suggests that these phenotypic changes might not be directly attributable to ABCB1 (Ref. 54).

In neuroblastoma, ABCC1 has also been implicated in promoting cell survival. Knock down of ABCC1 enhances spontaneous cell death in human neuroblastoma cell lines both in vitro⁵⁵ and in a mouse xenograft model⁵⁶. Consistent with these observations, expression of ABCC1 in primary untreated patient samples is a powerful independent prognostic factor in this disease³⁷.

Increased proliferation also contributes to the expansion of tumour populations. Knock down of ABCC4 inhibits proliferation of vascular smooth muscle cells both in vitro and in vivo⁵⁷, and knock down of ABCC1 reduced the mitotic index in neuroblastoma cell line xenografts⁵⁶. Similarly, knockdown of ABCB1 by small interfering RNA (siRNA) suppressed proliferation of a colon cancer cell line in vitro and inhibited tumour expansion in a xenograft model⁵⁸. In late retinal progenitors, proliferation was increased by transduction with a retrovirus that encoded ABCG2 and decreased with siRNA knockdown of ABCG2 (Ref. 59).

Roles in cell differentiation and functional significance in stem cells. Pluripotent stem cells are capable of self-renewal and can give rise to the differentiated cells that constitute tissues. A striking feature of both normal and cancer stem cells is their high level of ABC transporter expression compared with their more differentiated progeny⁶⁰ (Table 1). High ABC transporter expression might confer stem cells with enhanced protection from damage by xenobiotic substances, and it has been suggested that mice deficient in various transporters might be more susceptible to mutagenic chemicals⁶⁰; however, this is yet to be tested experimentally. The expression of ABC transporters in cancer stem cells also has important therapeutic implications, as tumours can have a population of inherently drug resistant pluripotent cells that can survive exposure to chemotherapeutic agents and re-establish the tumour, which is analogous to the stem cell-driven recovery of normal tissue after chemotherapy⁶⁰. However, as for normal stem cells, it is not yet clear whether expression of ABC transporters has a fundamental role in the cancer stem cell phenotype, or occurs as a result of other genetic changes during tumorigenesis.

Whether ABC transporters have any direct role in regulating differentiation remains unclear. Current genetic evidence does not support an absolute requirement for any individual ABC transporter in the maintenance of normal stem cell compartments, and mice deficient in Abcg2, Abcb1 or Abcc1 and mice deficient in both Abcb1 and Abcg2 all develop normally^{19,61,62,63,64,65}. Therefore, these genes do not seem to be responsible for stem cell maintenance or regulation of cellular differentiation (although these studies did not formally investigate the self-renewal capacity of stem cells by the engraftment of secondary recipients⁶⁶). However, given the large number of known ABC transporters² and their high expression levels in stem cells, it also remains conceivable that partial functional redundancy might mask their importance for stem cell maintenance or growth.

In contrast to knockout studies, however, several overexpression studies link ABC transporters with differentiation⁶⁶. For instance, the transduction of mouse bone marrow cells with a retrovirus encoding ABCB1 leads to dramatic ex vivo stem cell expansion and myeloproliferative disorder after engraftment⁶⁷, and enforced expression of ABCG2 in bone marrow cells causes a reduction in mature progeny both in vivo and in vitro⁶⁸. This raises the possibility that although transporters are not required to maintain an undifferentiated state under normal circumstances, they might favour its retention when upregulated in cancer. Alternatively, the expression of ABC transporters in undifferentiated cells might simply be a consequence of their stem cell-like transcriptional programmes rather than a causative factor. For several tumour types, the tumour grade, a low degree of differentiation and poor patient outcome each correlate with the presence of transcriptional programmes closely resembling those seen in embryonic stem cells^69,70 Interestingly, such transcriptional programmes can also be induced by the deregulated expression of MYC⁷⁰. It is therefore essential to determine whether the high expression of ABC transporters in tumours is functionally relevant for their development (aside from drug efflux) or whether they are simply bystanders and a consequence of an altered transcriptional programme in cancer cells, particularly in the case of MYC deregulation. ABC transporter-deficient mouse cancer models are central to answering these questions, as is a better understanding of the transcriptional and post-translational regulation of ABC transporters.

Cell migration, invasion and metastasis. Metastatic disease is the cause of the vast majority of cancer deaths⁷¹. Although direct evidence linking ABC transporters to metastasis is currently lacking, roles are emerging for these proteins in cell migration and invasion.

It is apparent that several ABC transporters have roles in the migration of normal cells. Induced migration of peripheral dendritic cells to lymph nodes is greatly reduced in Abcc1^−/− mice²², and Abcc1- deficient dendritic cells also have markedly attenuated chemotactic responses in vitro. Remarkably, exogenous addition of the ABCC1 substrates leukotriene C₄ (LTC₄) and LTD₄ overcame the absence of the transporter in vivo and in vitro²², suggesting a role for ABCC1 in autocrine signalling in mouse dendritic cell migration. Other ABC transporters might also have a role, as ABCB1-specific neutralizing antibodies and the inhibitor verapamil prevented the migration of dendritic cells from human skin explants²¹. However, comparison of these results across species is complicated by differences in the roles of various family members²⁴. In human dendritic cells, ABCC4 seems to have a more prominent role than ABCC1 in migration, with ABCC4 knockdown or pharmacological inhibition by sildenafil preventing their migration from human skin explants²³.

The apparent roles of ABC transporters in dendritic cell migration are also reflected in other normal and cancer cell lines. Reduction of ABCB1 levels by siRNA reduced the migration of MCF-7 breast cancer cells in transwell migration and Matrigel invasion assays⁷². Using similar assays, a doxorubicin-selected, multidrug-resistant human melanoma line expressing high levels of ABCB1 showed a more invasive phenotype than the parental cell line⁷³. Although it is possible that other genetic changes contributed to this phenotype, knock down of ABCB1 by siRNA substantially reduced the invasiveness of this cell line in vitro. Expression of ABCB1 in the canine kidney cell line MDCK also stimulated increased migration⁷⁴, and inhibition of ABCB1 reduced migration of the rat brain endothelial cell line RBE4 (Ref. 74). Finally, the downregulation of ABCC1 and ABCC4 reduces the migration of neuroblastoma cell lines in wound-closure assays in vitro (J.I.F., M.H., M.J.H. and M.D.N., unpublished observations). Although it remains to be determined whether ABC transporters contribute to metastasis in vivo, several studies indicate such a link. For example, in a breast cancer study, ABCC1 expression was higher in metastatic lymph nodes than in the corresponding primary tumour²⁷, and in melanoma, ABCC1 and ABCC4 were more highly expressed in cell lines derived from metastases than in those derived from primary tumours⁷⁵. More direct lines of investigation will be required to formally investigate these links.

Signalling lipid substrates and cancer

Various signalling molecules with established roles in tumour biology are either known or strongly suspected to be ABC transporter substrates, and for several of these, ABC transporters are either the only known or the most prominent efflux mechanism. These substrates, outlined in Table 1, include prostaglandins, leukotrienes and S1P (discussed below), as well as platelet activating factor, cholesterol metabolites and cyclic nucleotides (Box 2).

Prostaglandins. In some cancer types, tumour cells support an inflammatory microenvironment that promotes their own proliferation and survival, and which aids angiogenesis and metastasis⁷⁶. Among the mediators of chronic inflammation are the eicosanoids (prostanoids and leukotrienes), which are synthesised from arachidonic acid through the cyclooxygenase (COX; also known as prostaglandin endoperoxide synthase or PTGS) and arachidonate 5-lipoxygenase (ALOX5) pathways, respectively (Fig. 1).

Figure 1: ABC transporters in tumour-promoting lipid signalling pathways.

Eicosanoid (prostaglandin and leukotriene) and platelet activating factor (PAF) synthesis is initiated by the release of arachidonic acid (AA) and lysophosphophatidylcholine (lyso-PC) from phospholipids by the action of phospholipase A2 (PLA2). AA is converted into prostaglandin H₂ (PGH₂) by prostaglandin endoperoxide synthase (PTGS; also known as cyclooxygenase (COX)) and then by the action of specific prostaglandin (PG) and thromboxane (Tx) synthases (depending on cell type) to PGE₂, PGD₂, PGF_2α, PGI₂ and thromboxane A₂ (TXA₂). These short-lived molecules are effluxed by members of the ATP binding cassette C (ABCC) subfamily, activate G protein-coupled receptors (GPCRs), including EP₁–EP₄ (PGE₂ receptors), DP (PGD₂ receptor), FP (PGF_2α receptor), IP (PGI₂ receptor) and TP (TXA₂ receptor) and can be subject to re-uptake by members of the solute carrier organic anion transporter (SLC or OATP) family. AA is also converted by 5-lipoxygenase (ALOX5) into leukotriene A₄ (LTA₄), and either hydrolysed by LTA₄ hydrolase (LTA₄H) to LTB₄ or conjugated with glutathione by LTC₄ synthase (LTC₄S) to produce LTC₄, which can be further extracellularly metabolized to LTD₄ and LTE₄. LTs are effluxed by ABCC subfamily members and activate GPCRs including the LTB₄ receptors B-LT₁ and B-LT₂ and the cysteinyl leukotriene receptors CysLT₁ and CysLT₂. PAF is synthesized from lysophosphatidylcholine (LPC) by LPC acetyltransferase (LPCAT), effluxed by ABCB1 and binds to the PAF receptor (PAFR). Sphingosine-1-phosphate (S1P) synthesis begins with the production of ceramide from sphingomyelin by sphingomyelinase (SMase). Ceramide is converted by ceramidase (CDase) to sphingosine, which is subsequently phosphorylated by sphingosine kinases (SK1 and SK2) to produce S1P. S1P is in turn effluxed by ABCC1 and ABCA1 and activates the GPCRs S1PR₁–SIPR₅. Activation of the various GPCRs initiates various autocrine and paracrine effects that are relevant to tumour biology.

Clinical observations and genetic evidence from mouse models demonstrate the importance of prostaglandin-mediated pathways in cancer, particularly in colorectal cancer, but also in several other tumour types, including breast, lung and liver cancers^77,78,79. COX is a crucial component of prostaglandin synthesis, with the COX1 isoform constitutively expressed in many tissues and the COX2 isoform often detected in premalignant and malignant tissues, where it is a major therapeutic target^77,80,81. Downstream enzymes involved in prostaglandin synthesis and the various prostaglandin receptors have also provoked interest as therapeutic targets^79,82.

Prostaglandin signalling culminates in autocrine or paracrine activation of extracellular G protein-coupled receptors (GPCRs) (Fig. 1). The consequences of prostaglandin receptor activation include the activation of the PI3K–Akt pathway⁸³, which triggers various responses that drive tumour progression (such as cell proliferation, survival and motility⁸⁴ and might also contribute to growth factor autonomy⁷⁹; the activation of β-catenin signalling to promote a progenitor-like phenotype⁸⁵; the stimulation of angiogenesis through the expression of angiogenic factors such as vascular endothelial growth factor A (VEGFA)⁸⁶; and the promotion of metastasis⁸⁷. Prostaglandin signalling might also contribute to tumour evasion of immunosurveillance by inhibitingdendritic cell differentiation⁸⁸, and to the recruitment of inflammatory cells that support ⁷⁶ and so further enhancing tumour growth.

Leukotrienes:

A class of arachidonic acid-derived lipid mediators involved in inflammation and homeostatic biological functions.

Lipidomics:

Systems-level analysis and characterization of cellular lipid profiles using technologies such as mass spectrometry and computational analysis.

Reconstituted vesicle systems:

Artificial membrane systems mimicking natural lipid bilayers. Used to enable the full activity of membrane proteins, which often require correct orientation and bilayer insertion.

Smouldering inflammation:

Localized chronic, and often subclinical, inflammation with few systemic manifestations.

Xenobiotics:

Exogenous chemicals, either naturally occurring or synthetic, found in a living organism.

Although the inhibition of both prostaglandin synthesis enzymes and receptors has been the subject of numerous studies, little attention has been given to prostaglandin efflux mechanisms, particularly in the context of cancer. Prostaglandin efflux from cells by simple diffusion is an inefficient process⁸⁹ and it is offset by carrier-mediated re-uptake by the prostaglandin transporter (PGT)⁹⁰, as well as other transporters⁹¹, and by subsequent intracellular oxidation by 15-hydroxyprostaglandin dehydrogenase (15-PGDH)⁹². However, several members of the ABCC family efficiently transport prostaglandins (Fig. 1; Table 1) and in cancer cells in which these transporters are highly upregulated, this is likely to be their principal method of efflux. PGE₂, the prostaglandin most closely linked with cancer biology, is a substrate of ABCC1 (Ref. 93), ABCC2 (Ref. 93) and ABCC4 (Refs 94, 95); Abcc4 deficiency results in markedly reduced extracellular PGE₂ levels in culture and reduced PGE metabolites in vivo⁹⁶. Interestingly, in comparison to normal mucosa, ABCC4 expression is increased in human colorectal cancer specimens and cell lines, as well as in adenomas from the Apc^Min/+ mouse model of spontaneous intestinal tumorigenesis; however, PGT expression is decreased⁹⁷, raising the possibility that prostaglandin signalling might be enhanced by increased efflux and decreased re-uptake. Other prostaglandins are also ABC transporter substrates: PGD₂ is effluxed by ABCC2 (Ref. 93), and PGF_2α can be effluxed by ABCC4 (Refs 94, 95).

The importance of ABC transporters for eicosanoid biology could also extend to the regulation of the cyclopentenone prostaglandins 15d-Δ12,14-PGJ₂ and PGA₂, which are dehydration products of prostaglandins PGD₂ and PGE₂, respectively. The cyclopentenone prostaglandins exhibit potent anti-proliferative and anti-tumorigenic properties, and function intracellularly partly through nuclear receptors such as peroxisome proliferator-activated receptor-γ (PPARγ) rather than through extracellular G protein-coupled prostanoid receptors^{98,99,100,101,102}. Glutathione-conjugated 15d-Δ12,14-PGJ₂ and PGA₂ are substrates for ABCC1 and ABCC3 (Refs 103, 104), which might facilitate their clearance from the cell. ABC transporters may therefore deliver pro-tumorigenic prostaglandins to their extracellular GPCRs, while diverting anti-tumorigenic cyclopentenone prostaglandins from their intracellular targets.

Leukotrienes. Like prostaglandin signalling, leukotriene synthesis also culminates in the activation of extracellular GPCRs (Fig. 1), leading to various effects that include the activation of β-catenin to promote cell survival¹⁰⁵, the promotion of proliferation^106,107, leukocyte invasive behaviour¹⁰⁸ and the secretion of pro-tumorigenic factors by neutrophils¹⁰⁹. The potential of leukotriene pathways as targets in cancer is illustrated by the upregulation of leukotriene biosynthetic enzymes in various cancers^{110,111,112,113}, the effectiveness of ALOX5 inhibition in animal models of pancreatic cancer¹¹⁴, lung cancer^110,115 and chronic myeloid leukaemia (CML)¹¹⁶, and by the recent demonstration that the absence of Alox5 impairs the function of leukaemic stem cells, preventing the development of leukaemia in a mouse model of BCR–ABL-induced CML¹¹⁶.

ABCC1 is a major physiological mediator of LTC₄ efflux and is likely to be the transporter primarily responsible for the release of LTC₄ from leukotriene-producing cells. Bone marrow-derived mast cells from Abcc1-deficient mice exhibit only limited LTC₄ release and instead accumulate LTC₄ intracellularly, and Abcc1-deficient mice show a greatly impaired response to leukotriene-inducing inflammatory stimuli⁶⁵. Furthermore, dendritic cell migration from skin to lymph nodes is impaired in Abcc1-deficient mice and restored by exogenous cysteinyl leukotrienes⁶⁵. Numerous other members of the ABCC subfamily have been implicated in leukotriene transport with ABCC2, ABCC3, ABCC4, ABCC6, ABCC10 and ABCC11 each being able to efflux LTC₄ (Refs 117–124), and ABCC4 has also recently been shown to efflux LTB₄ (Ref. 123) (Fig. 1). The central role of ABCC transporters in leukotriene signalling suggests that they could provide alternative or additional therapeutic targets to enzymes involved in leukotriene synthesis and leukotriene receptors.

Sphingosine-1-phosphate. S1P synthesis and signalling is outlined in Fig. 1 and has recently been comprehensively reviewed¹²⁵. Although S1P can function intracellularly through mechanisms that are currently unclear¹²⁵, its key biological actions are mediated extracellularly through five high-affinity GPCRs (reviewed in Refs 126–129).

S1P is potently angiogenic, stimulating the migration of endothelial cells and promoting blood vessel formation^130,131. The importance of S1P for angiogenesis is illustrated by the S1P₁ receptor-knockout mice¹³² and sphingosine kinase 1 and sphingosine kinase 2 double knockout mice¹³³, which both exhibit lethal defects in vascular maturation. A central role for S1P in tumour biology, particularly in angiogenesis, was recently demonstrated by substantially reduced tumour progression in mouse xenograft and allograft models after treatment with a S1P-specific monoclonal antibody, apparently by the inhibition of tumour-associated angiogenesis¹³⁴. In addition, S1P has been implicated in the proliferation of tumour cells and their evasion of programmed cell death^127,135,136, and it has been shown to stimulate invasiveness in human glioblastoma lines¹³⁷. Cells overexpressing sphingosine kinase 1, the enzyme immediately responsible for S1P production, acquire the ability to form tumours in non-obese–severe combined immunodeficient (NOD–SCID) mice¹³⁸. Furthermore, S1P may induce the expression of COX2 and the production of PGE₂ (Ref. 139).

Recent evidence demonstrates that Abca1-deficient astrocytes have substantially reduced S1P efflux capacity, indicating that ABCA1 may be crucial for efficient S1P export from these cells¹⁴⁰. In addition, on the basis of siRNA-mediated knockdown, ABCC1 has been suggested as a major efflux mechanism for S1P from mast cells¹⁴¹. Multiple export mechanisms seem to exist for S1P, and whether any of these individual mechanisms is of particular relevance to tumour biology remains to be established. Diseases such as neuroblastoma, in which both high ABCC1 levels^37,142 and vascular index¹⁴³ are strongly predictive of poor outcome, might be suitable to address these questions.

Future directions

The range of cellular phenotypes observed with deletion, knock down and expression of ABC transporters, their ability to export known cancer-promoting substrates and their upregulation in cancer cells raise the possibility that ABC transporters have roles in cancer biology beyond the efflux of cytotoxic drugs. However, much of the evidence is circumstantial and few studies have directly addressed these roles from a genetic perspective. Further experiments are clearly required to determine the importance of various ABC transporters in mouse models of diseases for which they are predictive. Particularly enticing candidates include the deletion of Abcc4 in the TH-MYCN mouse model of neuroblastoma¹⁴⁴ given that ABCC4 is strongly predictive of neuroblastoma outcome³⁸ in a manner that seems to be independent of drug transport; and the deletion of Abcc4 in mouse models of colon cancer (such as Apc^Min/+ or ApcΔ⁷¹⁶) given that PGE₂, an ABCC4 substrate, is strongly implicated in colon cancer⁷⁹ and that ABCC4 is upregulated in this disease⁹⁷.

Although several lipid signalling molecules with roles in tumour biology have been identified as substrates of ABC transporters, there are likely to be more to be found. Most of the substrates have been identified from candidate approaches, and in most cases initially tested in . To date, no large-scale analyses have been conducted to identify additional substrates of ABC transporters; however, this is likely to be a fruitful area of research. The emerging field of should prove to be a key technology for these studies¹⁴⁵.

Leukotrienes:

A class of arachidonic acid-derived lipid mediators involved in inflammation and homeostatic biological functions.

Lipidomics:

Systems-level analysis and characterization of cellular lipid profiles using technologies such as mass spectrometry and computational analysis.

Reconstituted vesicle systems:

Artificial membrane systems mimicking natural lipid bilayers. Used to enable the full activity of membrane proteins, which often require correct orientation and bilayer insertion.

Smouldering inflammation:

Localized chronic, and often subclinical, inflammation with few systemic manifestations.

Xenobiotics:

Exogenous chemicals, either naturally occurring or synthetic, found in a living organism.

Leukotrienes:

A class of arachidonic acid-derived lipid mediators involved in inflammation and homeostatic biological functions.

Lipidomics:

Systems-level analysis and characterization of cellular lipid profiles using technologies such as mass spectrometry and computational analysis.

Reconstituted vesicle systems:

Artificial membrane systems mimicking natural lipid bilayers. Used to enable the full activity of membrane proteins, which often require correct orientation and bilayer insertion.

Smouldering inflammation:

Localized chronic, and often subclinical, inflammation with few systemic manifestations.

Xenobiotics:

Exogenous chemicals, either naturally occurring or synthetic, found in a living organism.

The contributions of ABC transporters to cancer hallmarks and the efflux of cancer-related cellular substrates provide an alternative rationale for developing targeted therapies against these proteins. To date, inhibitors of ABC transporters designed to reverse drug resistance have had limited clinical success^11,13. Although some of the same complications might still apply, targeting these alternative functions is different in several respects. Most inhibitors developed to date target ABCB1; however, most known cancer-related cellular ABC transporter substrates are effluxed by the ABCC subfamily (Table 1). Although several ABCC1 inhibitors have been developed or trialled^146,147, few target other members of this subfamily. In addition, failure of a given inhibitor to successfully overcome drug efflux in vivo does not necessarily extend to other substrates, as small molecule inhibitors can have differential, and sometimes opposing effects on the efflux of alternative substrates¹⁴⁸.

Conclusions

ABC transporters are well known to oncologists and cancer researchers for their capacity to efflux an array of therapeutic compounds, which leads to drug resistance and treatment failure. Recent studies suggest that ABC transporters may have other important roles in tumour biology; however, evidence linking them to various cancer hallmarks and to the efflux of tumour-promoting substrates is so far largely correlative. Targeted studies are now needed to address whether increased ABC transporter expression contributes to cancer progression independently of drug efflux. Studies that broaden our understanding of the function of ABC transporters beyond drug efflux could lead to a re-evaluation of their roles in determining clinical outcome. Given the functional redundancy observed for this highly conserved superfamily of transporter proteins, a combination of inhibitors may be required to effectively block a given transport function. Future studies examining their clinical importance should be carefully designed to enable the potential effects of all relevant transporters to be evaluated. Ultimately, this may entail widening the range of transporters screened in such studies. A handful of studies are leading the way in this respect, screening tumour cell lines or small tumour cohorts for expression of the entire ABC transporter family^{20,34,75,149,150}. Initial results suggest important roles for these genes beyond the expected MDR phenotype and argue that these studies should be extended to large tumour cohorts with comprehensive accompanying clinicopathological data.

Box 1 | ABC transporters and multidrug resistance

A major breakthrough in uncovering the mechanisms behind multidrug resistance (MDR) came in 1976, correlating drug resistance with the overexpression of a single protein encoded by the gene ATP binding cassette, subfamily B, member 1 (ABCB1; also known as MDR1). The encoded protein is an ATP-dependent efflux pump and a member of the ABC transporter family¹⁵¹. ABCB1 overexpression is now associated with treatment failure in many cancers, including those of the kidney, liver and colon, as well as lymphoma and leukaemia (reviewed in Ref. 152). Some years later, ABCC1 (also known as multidrug resistance protein 1) was identified as having a role in MDR phenotypes of small-cell lung carcinoma cells¹⁵³. ABCC1 overexpression has been correlated with drug resistance in prostate, lung and breast cancer, as well as in childhood neuroblastoma (reviewed in Refs 11, 154). ABCB1 and ABCC1 efflux overlapping but distinct sets of compounds (Table 1). The third major MDR pump to be discovered was ABCG2 (also known as BCRP), which has a remarkably wide range of substrates and has been associated with drug resistance in breast cancer and leukaemia^155,156,157. A key physiological function of ABC transporters is the protection of cells from many toxic insults from either endogenous or exogenous molecules that can enter the cell by diffusion or active uptake (see the figure). The protective mechanism afforded by ABC transporter-mediated extrusion of such toxic substances, whether they are metabolic waste products, naturally occurring substances or drugs, can make tumour cells resistant to the toxic effects of various chemotherapeutic agents^155,156,157.

Box 2 | Additional ABC transporter substrates with relevance to cancer

Cyclic nucleotides

Cyclic nucleotides are important secondary messengers downstream of G protein-coupled protein receptors with relevance to cancer biology (reviewed in Ref. 158) and are effluxed by ATP binding cassette, subfamily C, member 4 (ABCC4), ABCC5, ABCC11 (Refs 57, 96, 159, 160, 161, 162) and ABCG2 (Ref. 163). The biological significance of these observations is currently unclear as affinities are low, raising the possibility that their transport is relevant under circumstances of high cyclic nucleotide levels only¹⁶⁴. However, a recent study demonstrated that Abcc4 deficiency led to decreased extracellular and increased intracellular cAMP in culture⁹⁶, hinting that ABCC4-mediated cyclic nucleotide transport is of physiological significance at least in some systems.

Platelet activating factor

Platelet activating factor (PAF) has been implicated in various tumour-associated functions and signals and is exported by ABCB1 (Ref. 165). PAF activates its extracellular G protein-coupled receptor PAFR and induces upregulation of the anti-apoptotic proteins BCL-2 and BCL-X_L (Ref. 166). It has also been shown to contribute to angiogenesis in breast cancer¹⁶⁷ and to facilitate metastasis in melanoma¹⁶⁸ and colorectal carcinoma¹⁶⁹.

Cholesterol

Many ABCA family members are involved in regulating lipid transport and metabolism³. Cholesterol is an essential component of cellular membranes, and can be rate-limiting for the rapid growth and division of tumour cells. Furthermore, cholesterol is a key constituent of lipid rafts in cellular membranes, which bring together cooperating membrane-bound proteins for efficient activation of phosphorylation cascades^170,171. Not surprisingly, several cancers are associated with faulty cholesterol metabolism (reviewed in Ref. 4). For example, prostate cancer xenografts propagated in severe combined immunodeficient mice showed a dependency on circulating cholesterol for cell survival and tumour growth¹⁷¹. Alternatively, as intracellular cholesterol levels affect the rate of low density lipoprotein uptake, cholesterol trafficking can influence the availability of intracellular fatty acids, such as arachidonic acid, for the generation of many important signalling intermediates¹⁷² (Fig. 1). The question of whether coordinated regulation of ABC transporters (such as those regulating cholesterol metabolism and those effluxing lipid signalling intermediates) might lead to synergistic effects on tumour phenotype is yet to be addressed.