Regulation of ABC Transporters at the Blood-Brain Barrier

Abstract

ATP Binding Cassette (ABC) transporters at the blood-brain barrier function as ATP-driven xenobiotic efflux pumps and limit delivery of small molecule drugs to the brain. Here I review recent progress in understanding the regulation of the expression and transport activity of these transporters and comment on how this new information might aid in improving drug delivery to the brain.

THE BLOOD-BRAIN BARRIER

The blood-brain barrier resides within the 5–8 µm diameter microvessels of the brain. Brain capillary endothelial cells express a unique phenotype that is dependent on signals that come from their immediate environment. In situ, their development, maintenance and function are dependent on other elements of the neurovascular unit, including pericytes, astrocytes and likely microglia and neurons (1, 2). The brain capillary endothelium differs from peripheral microvessels in three important ways (1, 3). First, it contains remarkably tight junctional complexes between endothelial cells. These complexes severely restrict paracellular movement of both small molecules and macromolecules. Second, the endothelial cells exhibit limited transendothelial vesicular trafficking, further restricting movement of large solutes. Third, the cells express a particularly rich mixture of plasma membrane transporters that selectively facilitate the entry of the nutrients, ions and metabolites required for CNS homeostasis and optimal neuronal function. In addition, multiple ABC transporters drive efflux of potentially toxic metabolites generated within the CNS and limit accumulation of foreign chemicals (xenobiotics, including neurotoxicants and therapeutic drugs) coming from the periphery.

Available evidence indicates similar rosters of tight junction and transporter function proteins also function at the blood-spinal cord barrier; recent reports suggest similar modes of ABC transporter regulation (4–6). In addition to these barrier tissues, the brain’s ventricular system, including the choroid plexus epithelium and the lining of the subarachnoid space produces the cerebrospinal fluid (CSF) that bathes the surface of the brain and the spinal cord. By making and altering CSF, these tissues function to move signaling molecules through the CNS and to remove metabolic wastes and toxic proteins through regulated convective mechanisms.

ABC Transporters at the Blood-Brain Barrier

The blood-brain barrier’s efflux machinery does a superb job of recognizing xenobiotics, but a poor job of distinguishing between toxicants and therapeutic drugs. Thus the same mechanisms that protect against neurotoxicants also limit drug access to the CNS and in doing so present a serious obstacle to treatment of for example, brain cancer, epilepsy and neuroAIDS. ATP-driven, plasma membrane transporters that function as unidirectional efflux pumps are a major element of the barrier’s xenobiotic efflux machinery. These transporters are members of the ABC family. The human genome contains 49 genes encoding ABC transporters, divided into 7 different subfamilies, A-G, based on evolutionary divergence (7). ABC family members function as ATP-driven transporters on surface and intracellular membranes, as ion channels and as receptors. Mutations in ABC genes result in genetic disorders such as, cystic fibrosis (ABCC7, CFTR, a chloride channel), Dubin Johnson's syndrome (ABCC2, MRP2, a metabolite and drug transporter), progressive familial intrahepatic cholestasis (ABCB11, BSEP, a bile salt efflux pump) and retinal degeneration (ABCA4, a lipid flippase) (7).

For vertebrates, three ABC subfamilies, B, C and G, contain transporters that function as multispecific, ATP-driven efflux pumps and handle both metabolites and xenobiotics (7). These transporters are expressed in most cells, but they are most highly expressed in barrier and excretory tissues. As a result, they importantly influence the peripheral and CNS pharmacokinetics of many signaling molecules, waste products of normal metabolism, therapeutic drugs, environmental toxicants and drug and toxicant metabolites (Fig. 1). Because of their well-documented roles in limiting access of small molecule drugs to the CNS, the function of these ABC transporters at brain barriers is well documented (8). Through primary active transport, they are capable of generating and maintaining substantial drug concentration gradients across cellular membranes and tissues. This seems also to be the case for drugs that are lipophilic and that would normally diffuse readily across membranes. For those drugs, ATP-driven pumping is able to overcome diffusive back leak.

Figure 1.

The distribution of ABC transporters that handle foreign chemicals, e.g., drugs and toxicants, within the brain capillary endothelium. Note that for some of the MRPs subcellular distribution is species-dependent and still unresolved. The table summarizes the involvement of P-glycoprotein and BCRP in the handling of major classes of drugs.

Brain capillaries express multiple ABC transporters that are capable of handling a remarkably wide range of therapeutic drugs, including many chemotherapeutics (Fig. 1) (8). Transporters localized to the luminal plasma membrane contribute directly to barrier function, actively pumping back into the blood chemicals that diffuse into the cells. For P-glycoprotein (MDR1), one accepted view is that hydrophobic substrates for transport are extracted from the plasma membrane’s lipid bilayer (hydrophobic vacuum cleaner or flippase models) (9). Thus drugs that are both lipophilic and P-glycoprotein substrates do not need to access the endothelial cell interior to be available for transport. For ABC transporters localized to the abluminal plasma membrane, e.g., MRP1, or that are expressed at both membranes, e.g., MRP4, effects on drug entry into the CNS can be more complicated. One would think that abluminal ABC transporters would be positioned to pump substrates into the brain parenchyma. However, drugs that come from the periphery must first pass the battery of luminal efflux transporters, which often handle many of the same drugs as the abluminal transporters. Thus, the extent to which abluminal ABC transporters can facilitate drug delivery to the CNS is a complex function of many factors, including drug affinities for uptake and efflux transporters, transporter protein expression levels at each surface membrane and drug passive permeability, e.g., potential for back flux.

High ABC transporter expression on the luminal membrane of brain capillary endothelial cells is the major reason why it is such a challenge to deliver small molecule drugs to the brain. Moreover, numerous studies have shown that increased ABC transporter expression in the tissue leads to reduced drug accumulation in the brain and that decreased expression/activity leads to increased drug accumulation (10, 11). In addition, the blood-brain barrier and its transporters have been implicated in CNS disease progression (12–15), suggesting that these transporters are not just bystanders, but rather active participants and thus potential targets for therapy. We need a full understanding of ABC transporter function and regulation in health and disease is needed to improve the delivery of small molecule therapeutics to the CNS and to identify new ways to treat CNS diseases.

Measuring ABC Transporter Activity and Protein Expression

Assessing efflux function and regulation of blood-brain barrier ABC transporters poses unique challenges at the physiological, biochemical and molecular levels. Measurement of transport function, whether in vivo or in vitro, requires use of specific substrates and inhibitors. Because of overlapping substrate and inhibitor specificities, this is a serious problem for the MRP family of transporters, and still a problem for P-glycoprotein and breast cancer resistance protein (BCRP, ABCG2).

ABC transporters are unidirectional, efflux pumps and direct measurements of substrate efflux rates would be the best measure of function. These are difficult to make in vivo. A simpler approach is to measure drug uptake from the circulation using either radioisotopes (scintillation counting of brain homogenates or imaging using positron emission spectroscopy, PET (16)), fluorescent or luminescent probes (imaging (17)) or chemical analyses (usually mass spectroscopy-based). PET imaging with ¹¹C-labeled probes generates low resolution images, but it is one technique that can provide data on blood-brain barrier transporter function in humans. Although currently used to assess tight junction permeability in single vessels in situ, multiphoton imaging of fluorescent probes through a cranial window could be used to assess efflux transporter function in single brain capillaries in situ. Finally, new proteomic-based analytical techniques permit quantitative assessment of ABC transporter expression levels in brain capillaries and brain capillary endothelial cells from animal models and humans (18).

Mechanistic and molecular details of ABC transporter function and regulation can be investigated in vivo using pharmacological tools and knock out and transgenic animals. However, such questions often best lend themselves to investigation using in vitro systems at the intact capillary and endothelial cell levels. With freshly isolated capillaries, confocal imaging with fluorescent substrates provides direct measurements of ABC transporter activity (19). However, capillaries make up about 1% of brain volume and the low yield of pure capillaries limits the use of biochemical and molecular tools. Although we routinely assess gene expression at the mRNA and protein levels in isolated brain capillaries, we have not successfully used transfection or viral infection to alter gene expression. With intact capillaries, macromolecules and viruses do not readily penetrate endothelial cells over the 24 h during which they are fully viable and techniques used to hasten penetration and expression seem to reduce capillary viability (Miller, unpublished data).

In contrast, primary cell cultures and cell lines derived from brain capillary endothelial cells provide substantial material and the capability to make measurements of transport and gene expression (20). They respond to procedures commonly used to alter gene expression. Importantly, they provide access to cells of human origin. In this regard, an alternative, human stem cell-based approach to the use of endothelial cell lines shows great promise. This involves differentiating human pluripotent stem cells to endothelial cells with extensive blood-brain barrier characteristics and co-culturing them with other elements of the neurovascular unit, derived from neural progenitor cells (21, 22).

In the end, each approach has inherent strengths and weaknesses, requiring one to balance tradeoffs. In general, moving away from the in vivo situation increases the potential to bring powerful molecular tools to bear on underlying mechanisms of transport and their regulation. In doing that one has to be concerned about altered expression of key proteins, altered signaling and loss of critical cell-cell interactions within the endothelium and the larger neurovascular unit, and thus physiological relevance. For these reasons, it is critical to validate important in vitro findings with in vivo measurements.

In addition to the technical concerns discussed above, additional factors complicate activity-expression relationships and thus mechanisms of regulation. First, at least for P-glycoprotein, it appears that a finite fraction of transporter protein does not reside in the luminal plasma membrane of the endothelial cell and thus cannot contribute to efflux transport activity. Both immuno-electron microscopy and biochemical measurements suggest that 30–50% of P-glycoprotein in the brain capillary endothelium is localized to structures, e.g., internal and abluminal membranes, where it cannot contribute to drug efflux from the endothelium, (23, 24). Whether the internal pool of transport protein can be rapidly mobilized to the plasma membrane to enhance transport remains to be seen (24), as does the possibility that the transporters provide intracellular protection through sequestration of xenobiotics within vesicular compartments. Second, recent evidence indicates that P-glycoprotein activity in the plasma membrane can be modulated through formation of oligomeric structures within the membrane, either through self-association or association with other membrane proteins in microdomains, such as, caveoli and lipid rafts (9, 25–27). Such associations could have a profound effect on the activity of transporter within the luminal membrane. Finally, although the topic is complex, ABC transporters are intimately associated with membrane lipids and shifts in plasma membrane lipid composition can certainly influence transporter activity (9, 28). If a substantial fraction of a given ABC transporter protein were not in the luminal membrane, were positioned incorrectly within the membrane or were inactive within the membrane, measurements of protein expression based on Western blots or specific luminal membrane localization (biotin labeling, immunostaining) would overestimate functional protein.

REGULATION OF ABC TRANSPORTERS AT THE BARRIER

This section of the review contains four sub-sections. The first three sub-sections are focused on signaling mechanisms by which expression and activity of ABC transporters are regulated. The fourth deals with alterations in transporter activity in CNS disease.

Recent research has defined three signaling patterns that alter ABC transporter expression/activity at the blood-brain barrier: 1) Brain capillary endothelial cells express multiple nuclear receptors that can be activated by endogenous metabolites, nutrients and xenobiotics, with receptor activation directly driving increases in expression of multiple ABC transporters. 2) Upon ligand binding, certain ligand-activated receptors turn on signaling pathways that alter transporter expression though other, downstream transcription factors. They use signaling to indirectly drive increases in transporter expression. In many cases, translocation of nuclear factor kappa-light-chain-enhancer of activated B cells (NF-kB) to the nucleus is the final step. 3) Certain signaling pathways target basal transport activity of P-glycoprotein and BCRP. When activated, they rapidly and reversibly reduce transport activity. Targeting elements of these latter pathways has the potential to improve drug delivery to the CNS.

In the discussion that follows, I will focus primarily on two ABC transporters, P-glycoprotein and BCRP. Both are critical gatekeepers for many CNS-acting drugs and drug candidates (29). Both are highly expressed at the blood-brain barrier and both are multispecific, handling a large number of prescribed drugs. Studies from a number of laboratories now show that the two transporters work in concert to deny certain drugs access to the CNS. That is, the increase in brain accumulation of several drugs is much greater than additive when both transporters are knocked out or blocked by specific inhibitors of transport (29). Thus, for drugs that are modest substrates for both P-glycoprotein and BCRP, e.g., tyrosine kinase inhibitors like lapatinib, a large benefit for drug delivery to the CNS may be obtained by reducing transport on both transporters in concert. Certainly, for those drugs, being able to target both transporters at the same time would provide the greatest benefit to CNS pharmacotherapy.

Increased Expression by Direct Action of Ligand-Activated Receptors

In barrier and excretory tissues, multiple ligand-activated receptors function as sensors for metabolites, drugs and toxicants. Receptor proteins contain multiple functional domains, including ligand and DNA binding domains and often activation function domains. For many ligand and receptor pairs, induced transporter gene expression is part of a coordinated response to foreign chemicals that involves increased capability to metabolize xenobiotics (increased expression of Phase 1 and Phase 2 xenobiotic metabolizing enzymes) and then excrete the metabolites or parent compounds (increased expression of efflux pumps). As one might expect, ramping up of this response profoundly affects drug absorption, distribution, metabolism and excretion in the periphery; recent findings indicate that this is also the case for the CNS. The classical mechanism by which these receptors affect gene expression is through translocation of receptor plus ligand to the nucleus, followed by binding to the promoter regions of target genes. Often the receptor finds a partner protein in the nucleus (e.g., RXR for the pregnane-X receptor, PXR and the constitutive androstane receptor, CAR and ARNT for the arylhydrocarbon receptor, AhR) and it is the heterodimer that actually binds to DNA and facilitates assembly of the transcription complex.

Available data indicate that a number of ligand-activated receptors are expressed in brain capillary endothelial cells and intact brain capillaries from humans, pigs and rodents (Fig. 2) (10, 20, 30). These include PXR, CAR, AhR, peroxisome proliferator-activated receptor α (PPARα), vitamin D receptor (VDR) and glucocorticoid receptor (GR). Ligands for these receptors include a number of highly prescribed drugs dietary constituents and nutraceuticals (PXR, CAR, PPARα, VDR and GR) and persistent environmental contaminants (AhR). In brain capillary endothelial cells and in isolated brain capillaries, activation of any of these receptors increases expression of P-glycoprotein and BCRP (10). To date, evidence for increased blood-brain barrier P-glycoprotein expression and reduced drug delivery to the brain after in vivo exposure to receptor ligands has been presented for PXR (31, 32), CAR (33), AhR (34), VDR (35) and GR (Smith, Boni and Miller, unpublished data). For these receptors, there is no evidence indicating that this is accomplished at the blood-brain barrier though any mechanism other than direct interaction of the ligand-bound receptor with promoter regions in the target transporter genes.

Figure 2.

Transcription factors and receptors that alter expression of P-glycoprotein and BCRP at the blood-brain barrier. The horizontal line separates receptors that appear to directly increase transporter expression (above the line) from those that signal to other transcription factors (below the line). For specific references, see text.

One problem in assessing mechanism of action of receptor ligands and receptors arises from the possibility that certain ligands could interact with more than one receptor. This may be the case for the synthetic glucocorticoid, dexamethasone, which is a ligand for both GR and PXR. Multiple studies have shown that exposing isolated brain capillaries and brain capillary endothelial cells capillaries to dexamethasone increases expression and transport activity of both P-glycoprotein and BCRP (32, 36–38). In those studies, it was assumed that the synthetic glucocorticoid acted through PXR and no attempt was made to consider the possibility that action was through the GR. We recently tested this assumption. We have found high expression of GR mRNA and protein in brain capillaries from rats and mice (Smith, Boni, Miller, unpublished data). In isolated brain capillaries, dexamethasone-induced increases in P-glycoprotein activity/expression could be blocked by the GR-antagonist, RU486. However, RU486 also blocked the PCN–induced increase in transporter activity, indicating that the drug also interacted with both PXR (Smith, Boni, Miller, unpublished data). To determine the site of dexamethasone action in brain capillaries, we examined capillaries from PXR-null mice (the global knockout of GR is an embryonic lethal). In those capillaries, dexamethasone induced the same increase in P-glycoprotein activity as in capillaries from wild-type mice. As expected, PCN increased transporter activity in capillaries from wild-type mice, but not in capillaries from PXR-null mice. RU486 blocked the dexamethasone effects in capillaries from wild-type and PXR-null mice as well as the PCN effects in capillaries from wild-type mice (Smith, Boni, Miller, unpublished data). Thus, at least for mouse, dexamethasone increases P-glycoprotein expression in brain capillaries by acting through GR, not PXR.

Increased Expression by Receptor Signaling

Studies in rat and mouse brain capillaries have disclosed more complicated signaling mechanisms underlying stress-induced increases in ABC transporter expression (Fig. 3). These involve extended signaling from the plasma membrane or cytoplasm to transcription factors not directly activated by ligands. The transcription factor, NF-κB, appears to be a downstream point of convergence of stress-induced signals (inflammation, epileptic seizures, oxidative stress) at the blood-brain barrier. Moreover, it targets multiple ABC transporters. Thus, NF-kB appears to be, a stress-driven, master regulator of ABC transporter expression at the blood-brain barrier. Below I provide two examples of receptor-driven signaling that activates NF-kB and increases ABC transporter expression at the blood-brain barrier.

Figure 3.

Receptor/sensor signaling to NF-kB increases P-glycoprotein expression at the blood-brain barrier. For specific references, see text.

Inflammation accompanies all CNS diseases. Systemic administration of proinflammatory cytokines or bacterial endotoxin can alter P-glycoprotein expression at the blood-brain barrier and thus drug access to the CNS. Effects on transporter expression in vitro and ion vivo can be complex, with the nature of the stimulus, the dose and the time of exposure affecting the response (39). In rat brain capillaries, tumor necrosis factor- α (TNF-α) exposure triggers complex, inside-out signaling (40). TNF-α exposure activates TNFR1 and induces release of big- endothlin-1 (ET-1), which is converted to ET-1 by extracellular ET converting enzyme. ET-1 binds to ET_BR, which activates iNOS, protein kinase C isoform β2 (PKCβ2) and NF-κB in sequence (Fig. 3) (40, 41). Note that the upstream elements of this TNF-α signaling pathway are identical to those that rapidly reduce P-glycoprotein transport activity (42). In fact, TNF-α actually turns on both responses, with the loss of P-glycoprotein activity being seen early and the overall increase in transporter expression and activity being seen late. The mechanistic explanation for this is that signaling branches downstream of iNOS, at the PKC isoforms. Activating PKCβ1 initiates the rapid loss of activity with no change in expression and activating PKCβ2 independently initiates the delayed increase in transporter expression.

A second example involves barrier responses to oxidative stress mediated by the molecular sensor, nuclear factor (erythroid-derived 2)-like 2 (Nrf2) (43). Previous studies showed that activation of Nrf2 by sulforaphane (SFN), an electrophile and a constituent of broccoli, was neuroprotective in rodent models of ischemic stroke, traumatic brain injury, spinal cord injury and subarachnoid hemorrhage (44). As a result, Nrf2 has been proposed as a therapeutic target. Given that SFN acting through Nrf2 increases expression of multiple ABC transporters in liver, we wondered whether this would also happen at the blood-brain barrier, making it more difficult to deliver drugs to the injured brain. In this regard, we recently found that rats dosed with SFN exhibit increased protein expression of P-glycoprotein, MRP2 and BCRP in brain capillaries and reduced delivery to the brain of a P-glycoprotein substrate (6). Given the large number of therapeutic drugs handled by these three efflux transporters, one would expect reduced CNS access and thus reduced drug efficacy when Nrf2 is activated for neuroprotection.

Experiments with isolated rat and mouse brain capillaries also showed that Nrf2 action requires elements of both metabolic and genomic signaling (6). In those capillaries, Nrf2 acts indirectly to increase expression of P-glycoprotein, MRP2 and BCRP, requiring functioning p53, p38 and NF-κB (Fig. 3) (6). Although Nrf2, p53, p38 and NF-κB work in concert to increase transporter expression, it is not yet clear how this happens. Additional support for p53 as a signaling element that activates NF-κB and increases ABC transporter expression comes from our unpublished studies in rat and mouse brain capillaries and in vivo in rats where radiation-induced, DNA damage increases P-glycoprotein and BCRP expression through activation of ataxia telangiectasia mutated (ATM), p53 and NF-κB (Fig. 3) (Cannon and Miller, unpublished data).

Decreased P-glycoprotein Activity

Work from this laboratory has defined two extended signaling pathways that rapidly (minutes) and reversibly reduce P-glycoprotein transport activity in isolated rat and mouse brain capillaries. Such signaling does not reduce transporter protein expression, nor does it alter the activity of other ABC transporters or increase tight junction permeability. Importantly, in vivo brain perfusion with P-glycoprotein substrates (verapamil, morphine, loperamide and paclitaxel) and a tight junction marker (sucrose) shows that activation of signaling increases drug delivery to the brain without altering sucrose permeability (41, 45, 46).

One signaling pathway that reduces P-glycoprotein transport activity involves vascular endothelial growth factor (VEGF) signaling through flk-1 and Src kinase (46). Such effects were demonstrated using isolated rat brain capillaries. Consistent with those in vitro findings, intracerebroventricular injection of low doses of VEGF in rats increases brain accumulation of the P-glycoprotein substrates, ³H-morphine and ³H-verapamil, but not the tight junction marker, ¹⁴C-sucrose. Systemic administration of a Src kinase inhibitor blocked the effects of VEGF effects on P-glycoprotein-mediated transport (46). At higher doses, VEGF has multiple effects on the blood-brain barrier, including disruption of tight junctions. It is unlikely that VEGF itself would be a useful tool in the clinic to modulate P-glycoprotein activity. However, once the more downstream elements of VEGF signaling to P-glycoprotein are identified, they could be targeted to modulate P-glycoprotein activity acutely and thus improve drug delivery to the brain.

The second pathway shows more promise for use in the clinic (Fig. 4). It is clearly not complete. Substantial information is missing on the intermediate steps between some of the upstream signals, on additional signaling following mammalian target of rapamycin (mTOR) activation and on the mechanism of transporter inactivation (see below). By several criteria, this pathway is complex. First, it consists of three signaling modules: proinflammatory (41, 42, 47), sphingolipid (45) and protein kinase (Fig. 4) (Cannon and Miller, unpublished data). Second, two steps in the pathway involve inside-out signaling (5, 42, 47). That is, they generate metabolites within cells that act on external facing receptors and thus require a specific means of efflux. TNF-α exposure of capillaries stimulates release of big-ET-1 from the cell, which is cleaved to active ET-1 by an extracellular endothelin converting enzyme before binding to the ET_B receptor (42). Sphingosine is then converted to sphingosine-1-phosphate (S1P), which is transported out of the endothelial cells by MRP1 so that it can activate S1PR1 (5). Third, the upstream portion of the pathway can also signal increased P-glycoprotein expression. The pathways branch at the PKC isoforms, with PKCβ1 signaling reduced activity and PKCβ2 signaling increased transporter expression (41).

Figure 4.

An extended signaling pathway that regulates basal P-glycoprotein activity at the blood-brain barrier. Activation of the pathway in vitro (isolated brain capillaries) causes rapid and reversible loss of transport activity. Activation of the pathway in vivo rapidly increases drug delivery to the brain. For specific references, see text.

Activating this pathway has the potential to transiently reduce P-glycoprotein activity at the blood-brain barrier and thus improve drug delivery to the brain. Assuming the same signaling pathway exists in the human blood-brain barrier, several important questions still need to be addressed experimentally. First, which part of the pathway should be targeted in the clinic? One might expect that going after downstream targets would increase specificity and reduce side effects. Second, can the blood-brain barrier pathway be targeted selectively without affecting transporter activity in peripheral tissues? Recent studies show that the same sphingolipid signaling pathway that regulates P-glycoprotein transport activity at the blood-brain barrier also does so in renal proximal tubules from a teleost fish, a longstanding comparative renal model (48). Thus, it is possible that similar signaling will affect P-glycoprotein transport activity in other barrier and excretory tissues and thus alter peripheral drug pharmacokinetics. Limiting effects to the blood-brain barrier could be accomplished by carotid artery infusion of drugs meant to modify signaling. Finally, it is not yet clear that we can safely achieve a high enough concentration of signaling modifier in blood to obtain a substantial effect on transport activity. A recent review article argued that this would not happen with systemically administered S1PR agonists (49). One way around this limitation might be carotid artery infusion of drugs that modify signaling. This could provide a first-pass, CNS effect before dilution of the in the systemic circulation. It remains to be seen whether this approach would be practical in the clinic.

Decreased BCRP Activity/Expression

Estradiol (E2) decreases BCRP expression in a number of tissues and tumor cell lines (50). Targeting the blood-brain barrier with estradiol could provide a way to improve the therapeutic profile in the CNS for drugs that are substrates for BCRP alone and (along with an appropriate P-glycoprotein modifier) for drugs that are BCRP and P-glycoprotein substrates. Brain capillaries from female and male rats and mice express both estrogen receptor α (ERα) and ERβ, with expression of the latter dominating at both the mRNA and proteins levels (51, 52). Exposing rat and mouse brain capillaries to subnanomolar to nanomolar concentrations of E2 rapidly (minutes) and reversibly reduces BCRP-mediated transport activity without altering protein expression (42, 51). The reduction in activity is blocked by an inhibitor of membrane trafficking, suggesting that loss of activity follows trafficking of the transporter away from the plasma membrane. Both pharmacological studies with specific agonists and antagonists and experiments with capillaries from receptor-null mice indicate that both ERα and ERβ are involved (Fig. 5). These findings point to a non-classical (non-genomic) mechanism of E2 action, perhaps involving the classical receptors acting in concert with a membrane-bound G-protein coupled estrogen receptor (Fig. 5) (53). If a non-classical mechanism of ER signaling indeed drives the reduction in BCRP activity, targeting it could provide an important strategy to improve delivery of certain drugs to the CNS.

Figure 5.

Signals driving the loss of BCRP transport activity and protein expression following E2 exposure. For specific references, see text.

Extending E2 exposure of capillaries to 6 h reduces BCRP mRNA and protein expression and transport activity (51, 54). These effects are mediated by ERβ, which signals though phosphatidylinositide 3-kinase (PI3-K), phosphatase and tensin homolog (PTEN), protein kinase B (Akt) and glycogen synthase kinase 3 beta (GSK-3β) (Fig. 5). ERα is not involved in long-term signaling, so ERα agonists have the potential to be used to transiently reduce BCRP activity. Importantly, dosing mice with E2 (0.1 mg/kg by i.p.injection) recapitulates the complex time course of changes in BCRP activity and expression seen in E2-exposed brain capillaries (51). That is, capillaries isolated from dosed mice show reduced BCRP transport activity at 1, 6 and 24 h after dosing. Transporter protein levels (western blots of capillary membranes) are not affected 1 h after dosing, but are substantially reduced 6 and 24 h after dosing. Note that mice given a single dose of E2 exhibit a transient spike in plasma E2 levels; after the spike plasma E2 rapidly returns to control levels. We do not know how rapidly transport expression and activity recover in these mice or whether they recover at all.

Finally, there is crosstalk between estrogen receptor signaling and the Wnt/β-catenin pathway (see, e.g., (55). Recent experiments with a mice and human brain capillary endothelial cell line suggest that GSK-3β, a constitutively active protein kinase, links E2 signaling, Wnt/β-catenin signaling and down regulation of BCRP expression. GSK-3β is a central player in the canonical Wnt/ β-catenin pathway, phosphorylating β-catenin and targeting it for degradation. Intracerebral microinjection of a GSK-3β inhibitor in mice activates β-catenin and increases expression and transport activity of multiple blood-brain barrier transporters, including BCRP and P-glycoprotein (56). Moreover, in human brain capillary endothelial cells, activation of β-catenin signaling increases P-glycoprotein expression/activity and inhibition of the pathway decreases P-glycoprotein expression/activity (57). These experiments suggest new signaling-based targets that may be used to reduce expression of both BCRP and P-glycoprotein.

Altered Transporter Expression in Disease

For many neurological diseases, e.g., Alzheimer’s disease, Parkinson’s disease, epilepsy and ALS, there is good evidence for changes in blood-brain barrier ABC transporter expression in the affected regions of patients’ brains and spinal cords (58–61). For most of these diseases, experiments with animal models show similar effects of disease progression on transporter expression, but (except for epilepsy, see below) mechanisms that drive increased transporter expression in patients and in animal models are poorly understood (14, 58, 62). Although not reported for patients, P-glycoprotein upregulation is seen in animal models of stroke. Multiple mechanisms appear to contribute to increased transporter expression, including those dependent on the liver-X receptor and Apolipoprotein E (63, 64).

Substantial information is available from animal models on the signals driving increased P-glycoprotein expression following epileptic seizures. About 30% of epileptic patients exhibit resistance to antiepileptic drugs (AEDs). Several mechanisms underlie this multidrug resistance, including seizure-induced upregulation of ABC transporters at the blood-brain barrier (59, 65, 66). Studies of the mechanism driving the increases in transporter expression focused primarily on P-glycoprotein. As with inflammation (above), NF-κB appears to drive the increase in expression, but following seizures, a completely different signaling sequence activates NF-κB (Fig. 3). In this case excess glutamate, released through the increased neuronal activity, binds to an ionotropic N-Methyl-D-aspartate (NMDA) receptor, presumably increasing Ca⁺⁺ entry and activating phospholipase A2, which produces arachidonic acid. This metabolite is converted to prostaglandin E2 (PGE2) by cyclooxygenase-1 (COX-2). PGE2 efflux from the cells is mediated by an MRP, likely MRP4. Extracellular PGE2 binds to the prostaglandin E2 receptor, which then signals NF-κB activation and increased P-glycoprotein expression (Fig. 3) (67–69). Consistent with this sequence, inducing seizures in rats increases P-glycoprotein expression by a NMDA receptor and COX-2 dependent mechanism and inhibiting elements of the signaling pathway improves efficacy of certain AEDs (59). A recent publication shows that a similar signaling pathway (NMDA receptor, COX-2) upregulates P-glycoprotein activity in capillaries from human brain (70).

CONCLUSIONS

Delivery of small molecule drugs designed to access CNS targets remains a problem in the clinic. High expression of certain ABC transporters at the blood-brain barrier contributes substantially to the problem. Over the past decade, we have begun to develop an understanding of the mechanisms that regulate the expression and activity of these transporters. On the one hand, we now know that transporter expression at the barrier can be increased by exposure to xenobiotics (toxicants and therapeutic drugs), dietary constituents and inflammatory and oxidative stress. Given the breadth of the list of factors that increase transporter expression at the barrier, it is hard to believe that a substantial portion of the human population is not already induced and thus multidrug resistant. One wonders whether such resistance could be reversed through a modified diet that could be given before CNS pharmacotherapy.

On the other hand, we now know that basal activities of P-glycoprotein and BCRP at the blood-brain barrier are acutely regulated and that signaling pathways involved contain multiple elements that could be manipulated with drugs already in use in the clinic. Transiently reducing transport activity through signaling could provide a window in time when that part of the barrier is lowered and drugs that are transporter substrates could enter the CNS unimpeded. It remains to be seen whether any of these signaling-based strategies can be translated to the clinic.