4  Transporters
A large spectrum of transporters exist for different substrates and following the sequencing of the human genome it is believed that approximately 500–1200 genes encode drug transporters (Sakaeda et al., 2004). These carriers have been classified into two main groups, designated the ATP-binding cassette (ABC) transporters and the solute-carrier (SLC) superfamily (Loscher and Potschka, 2005).

4.1  ABC transporters
ABC transporters are the most extensively studied group, which are responsible for the cellular extrusion of a variety of molecules by using energy produced from ATP hydrolysis (Loscher and Potschka, 2005). P-Glycoprotein (P-gp) (also known as MDR-1 or ABCB1) is the most characterised ABC transporter. Other members of the superfamily include multi-drug resistance-associated proteins (MRPs) (also referred to as ABCC transporters) and breast cancer resistance protein (BCRP) (also known as ABCG2). Recent studies have demonstrated that the expression of these transporters at the BBB and on other cells such as lymphocytes, CD4+ T cells and microglia, play a crucial role in permitting the entry of anti-HIV drugs into cellular and anatomical reservoirs of HIV-1.

4.1.1  P-gp
P-gp, first discovered in 1976, is a 150–180 kDa membrane protein encoded by the multi-drug resistance gene 1 (MDR-1). It is widely expressed in tissues such as the liver, kidney and intestine, as well as the luminal membrane of the BBB (Cordon-Cardo et al., 1989; Jette et al., 1993, 1995; Soontornmalai et al., 2006) and the choroid plexus epithelium (Rao et al., 1999; Gazzin et al., 2008). At the BBB it appears to be important in protecting the brain from hydrophobic molecules and drugs. However, its function at the blood–CSF barrier is ambiguous and it may not have a major neuroprotective role at this site (Gazzin et al., 2008). Uncharged or weakly basic molecules are most efficiently transported by P-gp, but acidic compounds can also be transported. A large body of evidence suggests that PIs are substrates of P-gp and as a result the limited ability of these drugs to transverse the blood–brain barrier is attributed to the activity of this efflux transporter (Kim et al., 1998; Polli et al., 1999; Choo et al., 2000; Park and Sinko, 2005; Bachmeier et al., 2005; Eilers et al., 2008).
In comparison to PIs, the interaction of P-gp and NRTIs has been less extensively studied. Recently a study provided in vivo and in vitro evidence that the nucleoside reverse transcriptase inhibitor, abacavir ([(−)-(1S, 4R)-4-[2-amino-6 (cyclopropylamino)-H-purin-9-yl]-2-cyclopentene-1-methanol; 1592U89]) is a P-gp substrate (Shaik et al., 2007; Giri et al., 2008). In fact it is likely that P-gp is the dominant transporter limiting the CNS penetration of abacavir (Giri et al., 2008). Similarly, earlier studies showed that both HIV-infected T cell and monocytic cell lines had increased P-gp expression which accumulated significantly less zidovudine (azidodeoxythymidine (AZT)) in comparison to uninfected cells (Gollapudi and Gupta, 1990). Likewise, a decrease in AZT accumulation in P-gp-over-expressing CEM VBL100 cells with a corresponding decline in antiviral efficacy of the drug has been observed (Antonelli et al., 1992).
Using P-gp in Caco-2 cell lines, the NNRTIs nevirapine, efavirenz, and delavirdine were found not to be substrates of this transporter. However, all of these drugs were found to induce the expression and function of P-gp, with nevirapine being the more potent inducer compared with the other two NNRTIs (Stormer et al., 2002). Alternatively, a recent study demonstrated that the effect of P-gp on intracellular HIV-1 replication may be more clinically relevant than the efflux function of P-gp on PIs. The data suggested that high-cellular P-gp activity corresponds with a lower intracellular HIV-1 load in vivo (Sankatsing et al., 2007). Interestingly, Langford et al. (2004) showed that AIDS patients with HIV encephalitis (HIVE) have higher brain P-gp levels than HIVE-negative patients. However, despite studies showing an upregulation of P-gp in HIV-1 infected macrophages, CD4+ T lymphocytes and glial cells (Langford et al., 2004), the pump function of P-gp in HIV-1 infected patients is thought to be decreased (Sankatsing et al., 2004).
Recent experiments using primary culture of rat astrocytes have demonstrated that both the expression and the transport function of P-gp are downregulated following exposure to HIV viral envelope protein, gp120. Collectively, these crucial glial cells that harbour the virus within the CNS are thought to form a dynamic barrier behind the BBB to further impede the access of anti-HIV drugs to sites of infection within the CNS (Ronaldson and Bendayan, 2006). Furthermore, using intact, isolated rat brain capillaries, Hartz et al. (2004) revealed that subnanomolar to nanomolar concentrations of the hormone endothelin-1 (ET-1) rapidly and reversibly attenuated P-gp-mediated transport function over the short term (minutes). This effect was found to be due to the stimulation of the ETB receptor with subsequent activation of nitric oxide synthase and protein kinase C. The release of ET-1 has been apparent in a number of CNS disorders including HIVE (Hartz et al., 2004) and AIDS dementia complex however the effect of ET-1 on brain capillary permeability remains controversial, with some studies claiming that ET-1 significantly increases brain permeability and others suggesting no effect. This discrepancy can be attributed to the different durations of the experiments. An increase in permeability was observed over hours to days, raising the possibility that capillary permeability may remain unchanged during early ET-1 exposure (Hartz et al., 2004).
Inflammation is a central pathophysiological mechanism in the majority of CNS diseases and is reproduced experimentally by the injection of the bacterial endotoxin—lipopolysaccharide (LPS). Altered P-gp expression and corresponding changes in the disposition of several xenobiotics have been observed in the LPS model (Miller et al., 2008). Recent studies have demonstrated evidence in line with these findings. P-gp was downregulated via an unknown mechanism following the administration of LPS into rat intracranial ventricles. This subsequently caused an accumulation of the P-gp substrate, digoxin, within the brain (Goralski et al., 2003). Other studies have shown that the proinflammatory cytokine TNF-α causes a rapid and reversible loss of P-gp activity in rat brain capillaries.
The proposed mechanism suggested that short-term exposure to the cytokine caused TNF receptor 1 stimulation resulting in ET-1 release and consequent ETB receptor, nitric oxide synthase and protein kinase C activation. This pathway was thought to be activated by LPS to reduce P-gp transporter activity (Hartz et al., 2006). More recently, the same research group found that this initial rapid decrease in transport preceded a 2–3-h plateau at this reduced level of transporter activity, and was then followed by a rapid increase in both transporter activity and protein expression. Collectively, these findings demonstrate that chronic inflammation can tighten the BBB to CNS drugs which are P-gp substrates by upregulating P-gp expression (Bauer et al., 2007). An upregulation of P-gp in rat brain endothelium was also observed in an inflammatory pain model causing a decrease in the penetration of the P-gp substrate, morphine and consequent antinociception (Seelbach et al., 2007).
HIV-Tat, a protein thought to be responsible for the vascular abnormalities and neurotoxicity in HIV, also induces the expression of P-gp in brain endothelial cells which correlated with a functional upregulation of the transporter function of P-gp (Hayashi et al., 2005). A similar change in P-gp expression has been observed following chronic exposure of bovine brain microvessel endothelial cells to ritonavir. In fact, the HIV PI increased P-gp activity and expression in a concentration-dependent manner in this in vitro model of the BBB, raising the possibility that HAART could itself contribute to the brain as a HIV sanctuary site by the induction of drug transporters (Perloff et al., 2007). Collectively, these studies suggest that the selective inhibition of P-gp may facilitate the entry of PIs and certain NRTIs into viral sanctuaries and enhance the concentration of anti-HIV drugs in these sites to therapeutic levels.

4.1.2  MRPs
MRPs are ATP-driven efflux transporters. Those localised at the luminal membrane of the brain capillary endothelial cells (i.e. Mrp5) contribute to the non-permissive nature of the BBB (Dallas et al., 2006). However, evidence also exists for an abluminal (Mrp1) and possibly cytoplasmic (Mrp2) expression of specific isoforms (Soontornmalai et al., 2006). Currently, direct evidence exists only for the expression of MRP-1, MRP-2, MRP-4 and MRP-5 at the BBB (Dallas et al., 2006; Soontornmalai et al., 2006; Eilers et al., 2008) and of Mrp1 and Mrp3 (but not Mrp5) at the choroid plexus (Soontornmalai et al., 2006). Interestingly the choroid plexus, but not the BBB, is the main site of blood-facing, Mrp1-dependent cellular efflux from the rat brain (Gazzin et al., 2008).
MRPs are known to co-transport drugs with glucuronide, glutathione or sulphate. Amongst the nine members of this transporter family (MRP1–9), the first five (MRP-1, MRP-2, MRP-3, MRP-4 and MRP-5) are frequently associated with the efflux of therapeutic agents. MRP-1, MRP-2 and MRP-3, transport hydrophilic anionic compounds, large molecules and peptidomimetics (Dallas et al., 2006); however, both MRP-4 and MRP-5 transport small polar compounds such as nucleosides, cyclic nucleotides and nucleoside analogs (Schuetz et al., 1999; Wijnholds et al., 2000).
The HIV PIs saquinavir, ritonavir and lopanivir were found to be substrates of MRP-1 and MRP-2 and therefore these transporters may also contribute to the limited penetration of PIs into the brain (Bachmeier et al., 2005; Park and Sinko, 2005; Janneh et al., 2005, 2007; Eilers et al., 2008). A recent in vitro study demonstrated that MRP-1, MRP-2 and MRP-3 were also inhibited in a concentration-dependent manner by NNRTIs (delavirdine, efavirenz and nevirapine), NRTI (abacavir, emtricitabine and lamivudine) and the NtRTI, tenofovir. This inhibition was particularly apparent for delaviridine, efavirenz and emtricitabine (Weiss et al., 2007). In addition, MRP-4 and MRP-5 are thought to be low-affinity transporters of nucleoside-based antiretrovirals (Schuetz et al., 1999; Wijnholds et al., 2000) and both have been implicated in the rapid removal of zidovudine from human brain microvessel endothelial cells (Eilers et al., 2008). Previous studies have also suggested an association between MRP-4 over-expression and increased efflux of nucleoside-based anti-HIV drugs (Schuetz et al., 1999). A previous in vitro study revealed that ritonavir induces the expression of MRP-1 in a concentration-dependent manner, however this was noted in human colon in adenocarcinoma cells (Gimenez et al., 2004). MRP-1 is also thought to promote HIV replication (Speck et al., 2002).

4.1.3  BCRP
BCRP is a novel ABC-transporter which is localised in various tissues including cerebral endothelial cells (Eisenblatter et al., 2003). It is believed to have a similar tissue localisation to P-gp (Fig. 1). PIs such as ritonavir, saquinavir and nelfinavir have found to be effective inhibitors, but not substrates of this transporter (Gupta et al., 2004). Studies have also found that BCRP is a cellular factor involved in the resistance to NRTIs. For example, the accumulation and anti-HIV activity of AZT was significantly reduced by diminishing its metabolites, in cells overexpressing BCRP and this was reversed by fumitremorgin C, a BCRP inhibitor, suggesting that AZT is a substrate of BCRP (Wang and Baba, 2005; Wang et al., 2004).
More recently, a murine homolog of human BCRP was used to investigate the contribution of BCRP in the directional transport of abacavir and AZT. Their data provided evidence suggesting that both these drugs are substrates for Bcrp1 and further directional transport studies confirmed the role of Bcrp in the polarised transport of both abacavir and AZT in vitro (Pan et al., 2007). However, in vivo results showed that deletion of Bcrp1 has little influence on the brain penetration or overall disposition of AZT and only a moderate effect on abacavir (Giri et al., 2008). Overall, findings from these studies suggest that BCRP could contribute to drug–drug interactions observed in vivo following HAART. The role of this efflux protein during co-administration of anti-HIV drugs that are either substrates, inhibitors or both of BCRP, could be very valuable for the delivery of HAART to viral sanctuary sites.

4.2  The SLC superfamily
A large body of evidence demonstrates that the solute carrier (SLC) superfamily plays a crucial role in the efflux transport of organic compounds, organic anions particularly, across the BBB (Kusuhara and Sugiyama, 2005). Some members of this superfamily include organic anion-transporting polypeptide (OATPs), organic anion transporters (OATs) and organic cation transporters (OCTs). These ATP and sodium-independent polypeptides are expressed in a variety of tissues including the brain capillary endothelium and choroid plexus epithelial cells and regulate the movement of drugs through the brain barriers.

4.2.1  OATPs
All Oatp/OATPS are members of the SLC21 family and 36 Oatps/OATPS have so far been identified in humans, rats and mice (Hagenbuch and Meier, 2004). Several studies have shown that these transporters have a role in the efflux of organic compounds. The expression of OATP1A2 and OATP1C1 have been found in the BBB and the brain, respectively. In rats, Oatp1a4, Oatp1a5 and Oatp1c1 are expressed at the BBB and the blood–CSF barrier (Kusuhara and Sugiyama, 2005). OATP has been implicated in the removal of 2′3′-dideoxycytidine (ddC; zalcitabine) from the brain and CSF (Gibbs and Thomas, 2002). Oatp2 is found at the cerebral capillary endothelium and the choroid plexus epithelium, suggesting a role for these carriers in the transport of drugs between the blood and CNS compartments (Gao et al., 1999). An Oatp-2 like transporter has been implicated in the uptake of the NRTIs, 2′,3′-dideoxyinosine (ddI, didanosine) and (−)-2′-deoxy-3′-thiacytidine (3TC, lamivudine) into the choroid plexus (Gibbs and Thomas, 2002; Gibbs et al., 2003a,b). HIV may enter the CNS via the choroid plexus and therefore transporters present at this site could play an important role in CNS efficacy of certain drugs.

4.2.2  OATs
This family of transporters comprise OAT1, OAT2, OAT3, OAT4 and RST. UST1, UST3 and OAT5 are also considered as transporters of organic anions, however this has not yet been proven (Anzai et al., 2006). The recently identified urate transporter, URAT1 is very similar to RST in terms of amino acid sequence and tissue distribution. Additionally, URAT1 appears to be the human ortholog of murine RST and rat OAT1 (Eraly et al., 2004). Many drugs exist as organic anions at physiological pH levels and therefore OATs present must have a pivotal role in handling these compounds. Studies using Xenopus laevis oocytes showed that antiretroviral nucleoside drugs are substrates for OAT proteins despite the fact that they are not conventional organic anions (Strazielle et al., 2003).
Oat3 is present in the rat brain capillaries (Ohtsuki et al., 2002). There is no direct evidence for the expression of Oat1 at the BBB, however, it is suggested that the observed efflux of certain substances known to be Oat1 substrates from brain to blood indicates that the presence of Oat1 is likely (Sun et al., 2003). ddC and AZT are thought to be removed from the brain by a member of the OAT family (Takasawa et al., 1997; Gibbs and Thomas, 2002; Strazielle et al., 2003). OAT1 and 3 are expressed in the human, murine and rat choroid plexus (Sweet et al., 2002; Alebouyeh et al., 2003) and seem to be largely important in determining the availability of organic anions in the CSF (Eraly et al., 2004). These isoforms are likely candidates for the uptake of CSF-borne AZT and ddC (Gibbs and Thomas, 2002; Strazielle et al., 2003). However, the efflux of AZT was inhibited by specific inhibitors of both of these transporters, which makes it difficult to attribute a particular isoform to this effect.
Anthonypillai et al. (2006) investigated the distribution of a prodrug of the NRTI tenofovir, known as PMPA to the brain, CSF and choroid plexus. Interestingly, the entry of this pro-drug into the brain was negligible, but it could reach the CSF. The presence of a transporter at the level of the choroid plexus was indicated. Since hOAT1 and hOAT3 are high- and low-affinity transporters of PMPA, respectively (Cihlar et al., 2001; Izzedine et al., 2005), the involvement of these carriers was speculated however no detectable interaction was observed (Anthonypillai et al., 2006).

4.2.3  OCTs
The literature concerning OCT expression in the brain is at times contradictory, and this maybe the result of variation between species and also differing area and levels of expression of the transporters within the brain (Sweet et al., 2001). A study by Slitt et al. (2002) found that mRNA for all five organic cation transporters could be detected in the rat brain. Although this work did not evaluate whether the transporters were present at the brain barriers, additional studies have provided some evidence for the presence for individual organic cation transporters at the BBB and choroid plexus. It has been established that OCTN2 is found on the luminal face (and possibly the abluminal face too) of the cerebral capillary endothelial cells of humans, rats, pigs and cows (Kido et al., 2001). OCT2, OCT3, OCTN1 and OCTN2 (but not OCT1) are also expressed in the rat choroid plexus (Sweet et al., 2001). Human OCTs play a crucial role in the initial step involved in hepatic or renal excretion of many cationic compounds. Furthermore, previous studies have demonstrated the involvement of rat OCTs and an undefined organic cation-proton exchanger in the cellular uptake of AZT and 3TC, raising the possibility that human OCTs could also be involved in the uptake of anti-HIV drugs (Minuesa et al., 2008). Moreover, Minuesa et al. (2008) revealed that hOCTs were heterogeneously expressed in primary lymphocytes, monocytes, macrophages and dendritic cells. These transporters were also upregulated upon lymphocyte activation.
A recent study suggested that the PIs, nelfinavir, ritonavir, saquinavir and indinavir are possibly inhibitors of OCT1 and OCT2 (Jung et al., 2008). This supported evidence provided by Zhang et al. (2000) that some HIV PIs may be potent inhibitors of cationic drug uptake, but poor substrates for human OCT1 (Zhang et al., 2000). Furthermore, the NRTIs, lamivudine and zalcitabine are substrates of OCT1 and OCT2 (Jung et al., 2008). Moreover, when assessing the expression of OCTs in lymph nodes of HIV-infected patients in comparison to healthy controls, Jung and colleagues found that OCT1 and OCT2 were upregulated in the lymph nodes of HIV-infected patients, indicating that the accumulation of OCT substrates would be much higher in the lymph nodes of these patients. This could be due to the activation of the immune system and effects of cytokines induced by HIV infection (Jung et al., 2008). This study implies that a sound knowledge of OCTs is crucial in understanding both drug interactions that may occur with HAART and also the role they play in regulating drug transport.

4.3  Nucleoside transporters
There are two main types of nucleoside transporters, classified as low-affinity equilibrative (SLC29) and high-affinity concentrative (SLC28), which are known to transport certain NRTIs.

4.3.1  Equilibrative nucleoside transporters (ENTs)
There are four members of the SLC29 family. hENT1, the first characterised family member, shares similar substrate specificities with hENT2. These transporters play a pivotal role in the uptake of nucleoside and nucleobases across a membrane down their concentration gradient. Based on their sensitivity to nitrobenzythioinosine, ENTs are subdivided into equilibrative sensitive (es) and equilibrative insensitive (ei) (Baldwin et al., 2004; Thomas, 2004).
ENT1 protein, which encodes the es nucleoside transporter, is present on the human BBB. It is thought that this carrier is responsible for the transport of ddI across the guinea pig BBB. This was confirmed when the es transporter substrates, adenosine and thymidine, inhibited the brain uptake of [3H]ddI (Gibbs et al., 2003a). Interestingly, the es transporter is also active in human lymphocytes, raising the possibility that it could facilitate the transport of NRTIs into this HIV reservoir (Chan et al., 1993). ENT2, the ei transporter, transports purine and pyrimidine nucleosides but with a lower affinity that ENT1. However unlike ENT1, ENT2 is a carrier of AZT and can transport ddI and ddC much more efficiently than ENT1 (Baldwin et al., 2004).
Recently, Minuesa et al. (2008) found that human ENTs were the most highly expressed transporters in peripheral blood mononuclear cells and CD4+ T-cells and that the expression and activity of this transporter were increased by 100- and 30-fold, respectively, following stimulation of primary T lymphocytes. Determining the isoform responsible for anti-HIV drug transport across cell membranes would prove valuable in targeting these drugs to infected cells and in understanding drug–drug interactions and therapy failure (Minuesa et al., 2008).

4.3.2  Concentrative nucleoside transporters (CNTs)
There are three subtypes of sodium-dependent CNTs (CNT1–3) in the SLC28 family. CNT1 is a pyrimidine specific transporter whereas CNT2 is a purine-preferring transporter which also has the ability to transport uridine. CNT3 is a broad selective nucleoside transporter (Gray et al., 2004). These transporters have been identified in the liver, intestine, kidney and choroid plexus. Redzic et al. (2005) demonstrated the presence of rENT1, rENT2 and rCNT2 mRNA and protein in primary cultures of rat brain endothelial cells and rat choroid plexus epithelial cells. rCNT3 was present at the transcript level in choroid plexus epithelium. Furthermore, they showed that the concentrative transport of nucleosides was associated exclusively with the membrane that faces brain fluids in situ, suggesting a role of these barriers in extruding nucleosides from the brain (Redzic et al., 2005).
Using rat and human tissues, CNT and ENT cDNAs were cloned and their functional characterisation was determined using Xenopus oocytes. This study revealed that CNT1, CNT3 and ENT2 but not CNT2 or ENT1 transported AZT but with a much lower affinity than endogenous nucleosides. In addition, the influx of AZT into the CSF through choroid epithelial cells was found to be unaffected by endogenous nucleosides (Yao et al., 2001; Strazielle et al., 2003). CNT has also been implicated in the transport of ddI across the BBB (Li et al., 2001). However, the role of human CNT1 as a uptake transporter for ddI and stavudine (d4T) has remained controversial, as some researchers have described such a function for hCNT1 whereas others (Minuesa et al., 2008; Chishty et al., 2004; Chang et al., 2004; Cano-Soldado et al., 2004) have argued that hCNT1/2 (and even hENT1) have a low affinity for these drugs due to the absence of the 3′OH group in the ribose ring of the structure of nucleoside analogs. Conversely, hCNT3 appears to be an efficient transporter of AZT, ddC and ddI (Minuesa et al., 2008).