Blood-brain barrier (BBB) research has made enormous progress during the past two decades. Identification of tight junction proteins as a static barrier and the subsequent identification of P-glycoprotein (P-gp/MDR1/mdr1a), followed by various transporters and receptors, as a functional barrier provided evidence that the BBB has dual physiological functions in terms of brain homeostasis (Ohtsuki and Terasaki 2007), i.e., a barrier function of excluding circulating drugs and toxic agents from entering the CNS and a carrier function of supplying essential nutrients, hormones and drugs to the brain and eliminating metabolites from the brain. However, most of these findings were based on rodent data, and the functions and molecular mechanisms of the human BBB are still poorly understood. Indeed, mdr1a has been known to be expressed in rodent BBB and to function as the efflux pump to prevent entry of xenobiotics into the brain for two decades, but the question of whether or not MDR1 works at the human BBB was only resolved recently (Sasongko et al. 2005). From the viewpoints of clinical pharmacology and therapeutics, it is important to quantitatively elucidate the molecular basis of the transport functions at the human BBB to understand the brain distribution of drugs and endogenous compounds in humans, and to identify species differences.
Recent studies using imaging technologies have shown that human brain penetration of [18F]altanserin and [11C]GR205171, substrates of MDR1, was 4.5- and 8.6-fold greater than in rodents, respectively (Syvanen et al. 2009). Thus, MDR1 function at the human BBB might be less than that in rodents. However, this remains uncertain, because other several substrates did not show any species difference in brain penetration (Friden et al. 2009). Amino acid availability and tryptophan transport into the brain have been reported to affect cerebral protein and serotonin synthesis, respectively, in the brain (Fernstrom and Wurtman 1972; Lajtha 1974; Pratt 1976; Pardridge and Oldendorf 1977). The protein synthesis rate and serotonin concentration in human brain have been reported to be several-fold smaller than those in rodent brain (Hawkins et al. 1989; Young et al. 1994; Irifune et al. 1997), which might imply decreased influx rates of amino acids through the human BBB. System L, corresponding to L-type amino acid transporter 1 (LAT1) and 4F2 heavy chain (4F2hc), is involved in amino acid influx transport and also mediates brain uptake of drugs such as levodopa and melphalan (Ohtsuki and Terasaki 2007). Therefore, interspecies differences in BBB permeability of drugs via system L would be expected to result in species differences of drug efficacy.
Imaging technologies such as positron emission tomography (PET) and single photon emission computed tomography (SPECT) are the only functional analytical methods able to quantitatively evaluate BBB permeability of compounds in humans. However, availability of suitable ligands is still limited to only a few transporters (Kusuhara and Sugiyama 2009), and it is difficult to evaluate accurately the activities of individual transporters, because substrate and inhibitor specificities overlap among transporters, such as MDR1 and breast cancer resistance protein (BCRP), or multidrug resistance-associated protein 4 (MRP4) and organic anion transporter 3 (OAT3) (Kusuhara et al. 1999; Uchida et al. 2007; Enokizono et al. 2008; Ose et al. 2009). Comprehensive quantification of mRNA expression has been examined, but mRNA expression levels are not necessarily correlated to transporter function (Lescale-Matys et al. 1993; Molina-Arcas et al. 2005).
To overcome these problems, we developed a liquid chromatography–tandem mass spectrometry (LC-MS/MS)-based absolute quantification method using in-silico peptide selection criteria, and employed it to determine the protein expression levels of 34 transporters in mouse brain microvessels (Kamiie et al. 2008). In this method, the protein expression of a target molecule is selectively quantified by using a specific peptide probe. The in-silico selection criteria make it possible to select an appropriate peptide probe specific for each target molecule, thereby affording highly sensitive and accurate quantification. The protein levels have been shown to correlate to the activities of various functional membrane proteins, such as Na+/glucose co-transporter 1, cytochrome P450 2D6 (CYP2D6) and β secretase (Dyer et al. 1997; Fukumoto et al. 2002; Langenfeld et al. 2009). Therefore, clarifying protein expression of transporters or receptors in human brain microvessels by means of this technique is a rational strategy to understand the physiological roles of transport systems present at the human BBB.
The purpose of the present study was to determine comprehensively the protein expression levels of transporters and receptors in human brain microvessels, to clarify the role of the human BBB in drug distribution and homeostasis in the brain. We examined protein expression levels of 114 molecules, including ATP-binding cassette (ABC) and solute carrier (SLC) transporters and receptors, and compared the results with the mouse data to elucidate species differences in the BBB transport systems.
- Top of page
- Materials and methods
- Conflict of interest
- Supporting Information
The present study is the first to determine comprehensively the quantitative protein expression profiles of membrane transporters and receptors in human isolated brain microvessels, and to evaluate species differences in the protein expression profiles between human and mouse BBBs. Among drug transporters, BCRP showed the most abundant protein expression in humans, and its expression level was greater in humans than in mice, whereas the expression levels of MDR1/mdr1a and MRP4 were smaller in humans than in mice.
BCRP is involved in the brain-to-blood efflux of various xenobiotics at the BBB (Breedveld et al. 2005; Enokizono et al. 2008). Functional and expressional analysis has shown that, in mouse, BCRP has a less significant influence on drug efflux from brain as compared with MDR1 (Enokizono et al. 2008; Kamiie et al. 2008). However, the relative importance of BCRP and MDR1 in human BBB has remained unknown, because no functional study with PET or SPECT has been conducted for BCRP in the human BBB. At the mRNA level, BCRP is expressed at a higher level than MDR1 in isolated human brain microvessels, while a human BBB model cell line (hCMEC/D3) expressed MDR1 more abundantly than BCRP in vitro (Dauchy et al. 2008, 2009; Carl et al. 2010). Our results show that the protein expression amount of BCRP is 1.34-fold greater than that of MDR1 in isolated human brain microvessels (Table 3). In contrast, bcrp expression was only 31.3% of mdr1a expression in mouse brain microvessels (Kamiie et al. 2008). These results suggest that the functional role of BCRP relative to MDR1 is greater in the human BBB than in the mouse BBB.
As shown in Table 5, BCRP is considered to have 1.85-fold greater transport function in human BBB than in mouse BBB, on the assumption that the intrinsic transport activity (transport rate per protein) of BCRP is the same in humans and mice. Various BCRP substrates, including imatinib, gefitinib, erlotinib, mitoxantrone and topotecan (Kusuhara and Sugiyama 2009; Urquhart and Kim 2009), have been used for the treatment of glioblastoma, but their therapeutic benefit is minimal (Wen and Kesari 2008). In contrast, temozolomide is effective and is generally used as a first-line drug for glioma patients (Wen and Kesari 2008). Temozolomide is not transported well by human BCRP (de Vries 2009). Therefore, we speculate that the high efflux activity of BCRP at the human BBB is one of the reasons for the limited pharmacological activity of these anti-cancer agents, other than temozolomide, against glioblastoma.
In contrast to BCRP, the protein expression of MDR1 in humans was significantly smaller (2.33-fold) than in mice (Table 5). PET analysis showed that the brain-to-plasma ratios of MDR1 substrates [11C]GR205171 and [18F]altanserin were 8.6- and 4.5-fold greater in humans compared to rodents, respectively (Syvanen et al. 2009), suggesting that MDR1 is functionally less active in human BBB than in rodent BBB. The decreased protein expression amount presumably accounts at least in part for the decreased function of MDR1 in human BBB. Decreased intrinsic transport may also play a role, although the intrinsic transport activity is unknown because the amount of MDR1/mdr1a protein has never been quantified before. These considerations suggest that further studies on species differences in intrinsic transport activity by using our quantitative targeted absolute proteomics method are likely to be fruitful.
MRP4 and OAT3 are well-characterized organic anion transporters at the rodent BBB, and, in vivo experiments with gene knockout mice have shown that mrp4 and oat3 play roles in drug efflux from brain to blood (Uchida et al. 2007; Ose et al. 2009). The present study shows that protein expression of MRP4 is significantly smaller (8.1-fold) in humans than in mice, and that of OAT3 is at least 5.7-fold smaller (Table 5), raising the possibility that the efflux activities of MRP4 and OAT3 are low in human BBB compared with mouse BBB. Homovanillic acid, a major metabolite of dopamine, was reported to be eliminated from the brain by OAT3 (Mori et al. 2003). The homovanillic acid concentration in brain striatum is higher in humans (3.24–7.20 μg/g brain) than in mice (0.31–1.45 μg/g brain) (Sharman 1966; Adolfsson et al. 1979; Irifune et al. 1997). This could support the view that the lower efflux activity of OAT3 in humans is due at least in part to the lower protein expression of OAT3 in humans. Furthermore, Ro64-0802, an active form of the anti-influenza virus agent oseltamivir, undergoes active efflux mediated by OAT3 and MRP4 at the BBB (Ose et al. 2009). Recently, abnormal behavior has been reported in teenagers or younger people prescribed oseltamivir, although rodent study showed no specific CNS or behavioural effects after administration of doses corresponding to at least 100 times the clinical dose (Toovey et al. 2008). Ro64-0802 is suspected to be one of the major causes of the CNS adverse effects in humans, because it is 30 times more potent than oseltamivir (Izumi et al. 2007). Hence, a possible explanation for oseltamivir toxicity in humans may be that low expression of MRP4 and OAT3 in the human BBB results in reduced efflux of Ro64-0802 from the brain, leading to greater accumulation in the brain, which in turn induces adverse effects on the CNS.
As was the case for OAT and MRP family members, except for MRP4, no protein expression was observed for all the OATP subtypes (Table 4), which transport a variety of amphipathic drugs including hydroxymethylglutaryl-CoA reductase inhibitors, angiotensin-receptor blockers, fluoroquinolones, steroids, β-blockers and synthetic peptides deltorphin II and D-penicillamine (2,5)-enkephalin (Urquhart and Kim 2009). Although OATP-A was detected in the brain capillaries of normal human brain cortex in an immunohistochemical study (Gao et al. 2000), the present study showed that the amount is no more than 0.695 fmol/μg protein. OATP-B was reported to be localized at the capillaries of human gliomas (Bronger et al. 2005), but we found that the amount of OATP-B in isolated human brain microvessels was as low as that of OATP-A. Oatp-2 is a possible mouse homolog of OATP-A and OATP-B, but we found that the amounts of OATP-A and OATP-B in humans were at least 3.0- and 6.3-fold less than that of oatp2 in mice, respectively (Table 5). These findings would suggest that OATP-mediated transport across human BBB is less active than that across mouse BBB, as would also be the case for OAT and MRP family members. Functional studies with in vivo imaging technologies such as PET and SPECT should also be informative to investigate the in vivo functions of OATP, OAT and MRP subtypes at the human BBB.
Interestingly, ABCA8 protein was detected in human brain microvessels in the amount of 1.21 fmol/μg protein, a level similar to those of mrp4, oat3 and oatp-2 in mouse brain microvessels (Table 5). An in vitro experiment using the Xenopus oocyte expression system indicated that ABCA8 transports organic anions, including estradiol-17β-D-glucuronide, taurocholate, ochratoxinA and digoxin (Tsuruoka et al. 2002). These compounds are also substrates of rodent mrp4, oat3 and oatp-2 (Kusuhara et al. 1999; Ohtsuki and Terasaki 2007). Hence, ABCA8 could play a similar role to mouse mrp4, oat3 and oatp-2 as an organic anion transporter if the intrinsic transport activity is close to those of mouse mrp4, oat3 and oatp-2.
In addition to drug transporters, transporters of endogenous compounds were also analyzed for species differences in protein expression amounts (Table 5). Among the glucose transporters, GLUT3 showed a remarkable species difference, with the protein amounts being at least 7.25-fold greater in human than in mouse isolated brain microvessels. By contrast, the amount of GLUT1 in human was not significantly different from that in mouse. At the functional level, PET studies found that the maximal velocity of glucose transport at the human BBB (0.4–2.0 μmol/min/g brain) was not significantly different from that at the rodent BBB (1.42 μmol/min/g brain) (Pardridge 1983; Gruetter et al. 1996). Therefore, glucose supply into the brain is likely to be mainly mediated by GLUT1 at the BBB.
System L, corresponding to LAT1 and 4F2hc, supplies large neutral amino acids including leucine, tryptophan, tyrosine and phenylalanine to the brain. The expression amounts of LAT1 and 4F2hc were both significantly lower (fivefold) in human than mouse isolated brain microvessels (Table 5). This result is consistent with the decreased maximum rate of phenylalanine uptake by isolated brain microvessels of humans compared with rodents, although there were large variations of Vmax and Km in human (Choi and Pardridge 1986). Furthermore, the cerebral protein synthesis rate in human brain (0.345–0.614 nmol/min/g) has been estimated to be lower than that in rodent brain (3.38 nmol/min/g) by PET using L-[1-11C]leucine and a three-compartment model (Hawkins et al. 1989), and the brain concentration of serotonin generated from tryptophan has been shown to be lower in humans than that in mice (e.g. 20 ng/g vs. 679 ng/g brain in the frontal cortex) (Young et al. 1994; Irifune et al. 1997). Cerebral protein and serotonin synthesis are likely affected by amino acid availability in the brain and tryptophan transport into the brain, respectively (Fernstrom and Wurtman 1972; Lajtha 1974; Pratt 1976; Pardridge and Oldendorf 1977). Therefore, the supply of amino acids via system L at the human BBB could be smaller than that at the mouse BBB, in accordance with the measured protein expression amounts of LAT1 and 4F2hc. No significant difference in an influx rate of amino acids at the BBB has been observed between humans and rodents using L-[1-11C]tyrosine, [11C]aminocyclohexanecarboxylate and L-[U-14C]phenylalanine (Koeppe et al. 1990; Wiesel et al. 1991; Nakamichi et al. 1993). However, system L is estimated to be over 90% saturated by plasma concentrations of amino acids under physiological conditions (del Amo et al. 2008), suggesting that the influx rate of PET probes would have been affected by the plasma concentrations of endogenous system L substrates, including amino acids. Therefore, the species differences in system L function at the BBB remain unclear and further studies are necessary to clarify in vivo system L function at the BBB.
Like LAT1 and 4F2hc, MCT1, which supplies lactate and ketone bodies to the brain, is also associated with the chaperone protein CD147. It has been reported that the protein levels of MCT1 are regulated by CD147, but the mRNA levels are not (Philp et al. 2003), suggesting that the mRNA and protein levels of MCT1 are not always correlated. Indeed, mRNA expression of MCT1 in brain decreased by fivefold from the postnatal day 15 to 60 (Pellerin et al. 1998), whereas the protein expression decreased by 25-fold from the postnatal day 17 to adult in the brain capillaries (Leino et al. 1999). Also, a ketogenic diet up-regulated the mRNA expression of MCT1 by threefold in rat hippocampus compared with the normal diet (Noh et al. 2004), whereas the protein expression was up-regulated by eightfold in the brain capillaries (Leino et al. 2001). Therefore, the quantitative analysis of protein levels, rather than in mRNA levels, may be preferable for understanding the role of MCT1 in human development and the effect of a ketogenic diet.
The peptide probes of 92 target molecules were not detected in the human brain microvessels, suggesting that the protein expression amounts of these 92 molecules were under the limits of quantification listed in Table 4. However, we cannot completely exclude the possibility that the expressions of some proteins were overlooked or underestimated, because the following two possibilities cannot be ruled out. Firstly, there might be unreported modifications or mutations involving the targeted peptides in the 92 molecules, which would have resulted in failure to detect the targeted peptides, although we checked that no relevant gene mutation or post-translational modification in the target peptide was registered in the UniProtKB database. Secondly, some molecules may have been insufficiently solubilized and digested with trypsin. We have validated the efficient solubilization and digestion of glut1 in mouse brain microvessels and human MDR1 in MDR1-over-expressed cells by comparing quantification values obtained with the present method and with binding assays, including immunoblotting (Kamiie et al. 2008). Furthermore, no bands over 20 kDa were detected by sodium dodecyl sulfate–polyacrylamide gel electrophoresis after trypsin digestion. The results suggest that the solubilization and trypsin digestion proceeded efficiently, but do not necessarily indicate complete solubilization and digestion of all molecules.
Immunohistochemical analyses have shown that MRP1, MRP5, OATP-A, OATP-B and RLIP76 are localized in human brain microvessels (Nies et al. 2004; Awasthi et al. 2005; Bronger et al. 2005), but these were not detected in the LC-MS/MS. As noted above, there are several possible reasons for this difference, and it is not necessarily the case that the protein levels were under the limit of quantification of our method. However, OATP-A and OATP-B were immunohistochemically detected in the microvessels of brain tumors. Therefore, it is possible that protein expression of OATP-A and OATP-B might be induced in brain tumors. Furthermore, OATP-A has been detected in monkey brain microvessels (0.724 fmol/μg protein) by using the peptide at the same position, although one amino acid is different in the monkey OATP-A peptide (Ito et al. 2011). This finding suggests that OATP-A protein in human microvessels would have been efficiently digested by trypsin, but that the protein level was indeed under the limit of quantification.
In the present study, brain microvessels were isolated as described previously (Dauchy et al. 2008) with minor modifications. In the method validation by Dauchy et al. (2008), the mRNA expression levels of neuron and astrocyte markers, but not pericyte markers, were shown to be significantly reduced in the isolated fraction compared with those in cerebral cortex. Therefore, neurons and astrocytes are likely to have been efficiently removed from the isolated fraction used in the present study. As regards pericytes, our isolated microvessels did not show detectable protein expression of MRP1 and PGT (Table 4) which were reported to be highly expressed in pericytes compared with brain capillary endothelial cells (Berezowski et al. 2004; Kis et al. 2006). This suggests that the extent of pericyte contamination is limited.
Generally, inter-individual differences in expression amounts were small, although EAAT1 showed a 4.24-fold difference at maximum (Table 3). ABCA8, sulfonylurea receptor 1 (SUR1), LAT1 and RFC were not detected in some donors, but their limits of quantification were close to the expression amounts found in other donors (< twofold difference), so it possible that these four transporters also do not show large inter-individual differences in their levels. According to a recent LC-MS/MS-based absolute quantification analysis, the protein expression amounts of cytochrome P450 in human liver microsomes were remarkably different among individuals, for example, over 20-fold differences for CYP1A2, 2A6, 2C19 and 3A4 (Kawakami et al. 2011). Additionally, protein expression of MRP2 showed a sixfold variation among individuals in the liver membrane fraction (Li et al. 2009). Compared with these differences in liver, the inter-individual differences in protein expression amounts of transporters and receptors in isolated human brain microvessels are quite small.
In conclusion, the present study has comprehensively elucidated the absolute protein expression profile of transporters and receptors in isolated human brain microvessels. Comparison of the results with mouse data revealed clear species differences between humans and mice. Furthermore, these differences are likely consistent with reported species differences in physiological phenomena and pharmacokinetics in the brain. These quantitative protein expression profiles would be important to increase our understanding of drug penetration into human brain, homeostatic mechanisms of endogenous compounds in human brain, and their species differences. As intrinsic transport activities, as well as protein expression amounts, affect the functions of transporters and receptors, further study to clarify the intrinsic transport activities is necessary to fully understand the brain distribution of drugs and endogenous compounds in humans.