Molecular biology of the blood-brain and the blood-cerebrospinal fluid barriers: similarities and differences

Efficient processing of information by the central nervous system (CNS) represents an important evolutionary advantage. Thus, homeostatic mechanisms have developed that provide appropriate circumstances for neuronal signaling, including a highly controlled and stable microenvironment. To provide such a milieu for neurons, extracellular fluids of the CNS are separated from the changeable environment of blood at three major interfaces: at the brain capillaries by the blood-brain barrier (BBB), which is localized at the level of the endothelial cells and separates brain interstitial fluid (ISF) from blood; at the epithelial layer of four choroid plexuses, the blood-cerebrospinal fluid (CSF) barrier (BCSFB), which separates CSF from the CP ISF, and at the arachnoid barrier. The two barriers that represent the largest interface between blood and brain extracellular fluids, the BBB and the BCSFB, prevent the free paracellular diffusion of polar molecules by complex morphological features, including tight junctions (TJs) that interconnect the endothelial and epithelial cells, respectively. The first part of this review focuses on the molecular biology of TJs and adherens junctions in the brain capillary endothelial cells and in the CP epithelial cells. However, normal function of the CNS depends on a constant supply of essential molecules, like glucose and amino acids from the blood, exchange of electrolytes between brain extracellular fluids and blood, as well as on efficient removal of metabolic waste products and excess neurotransmitters from the brain ISF. Therefore, a number of specific transport proteins are expressed in brain capillary endothelial cells and CP epithelial cells that provide transport of nutrients and ions into the CNS and removal of waste products and ions from the CSF. The second part of this review concentrates on the molecular biology of various solute carrier (SLC) transport proteins at those two barriers and underlines differences in their expression between the two barriers. Also, many blood-borne molecules and xenobiotics can diffuse into brain ISF and then into neuronal membranes due to their physicochemical properties. Entry of these compounds could be detrimental for neural transmission and signalling. Thus, BBB and BCSFB express transport proteins that actively restrict entry of lipophilic and amphipathic substances from blood and/or remove those molecules from the brain extracellular fluids. The third part of this review concentrates on the molecular biology of ATP-binding cassette (ABC)-transporters and those SLC transporters that are involved in efflux transport of xenobiotics, their expression at the BBB and BCSFB and differences in expression in the two major blood-brain interfaces. In addition, transport and diffusion of ions by the BBB and CP epithelium are involved in the formation of fluid, the ISF and CSF, respectively, so the last part of this review discusses molecular biology of ion transporters/exchangers and ion channels in the brain endothelial and CP epithelial cells.

Although there are several similar features between the blood-brain barrier (BBB) and the blood-cerebrospinal fluid barrier (BCSFB), it should be kept in mind that the cellular basis of these two structures as well as their primary functions differ: BBB is located in brain capillaries and, thus, it is an endothelial structure with its main role to protect the brain from physiological fluctuations in plasma concentrations of various solutes and from blood-borne substances that could interfere with neurotransmission, but at the same time to provide mechanisms for exchange of nutrients, metabolic waste products, signaling molecules and ions between the blood and the brain ISF. In contrast to this, the BCSFB is created by a layer of a modified cuboidal epithelium, the CP, that secretes cerebrospinal fluid (CSF) and this process could be considered as main function of this epithelium. The differences in principal function are related to differences in morphology and molecular biology. Brain capillaries express complex morphology that provide the restrictive characteristics of the endothelial layer with regard to diffusion of solutes; this is an essential feature to protect the brain from unwanted solutes from blood with tight junctions (Tjs) that interconnect adjacent endothelial cells and occlude the paracellular spaces. In addition, BECs show low pinocytotic activity and the endothelium is further secluded by a layer of astrocytic end feet and pericytes on the brain side that place additional restrictions on permeability. Thus, the BBB in vivo provides high resistance to movement of ions, with transendothelial electrical resistance (TEER) being in the range of 1500 Ω cm 2 (pial vessels), which is quite high when compared to TEER of 3-33 Ω cm 2 in other tissues [ 1 ]. The total capillary surface area in the brain is about 100-150 cm 2 g -1 [ 2 ], which when estimated for the whole brain approximates 20 m 2 [ 3 ], suggesting that the BBB can be considered as a large and thin membrane, providing ideal conditions for exchange processes between blood and brain interstitial fluid (ISF). When considering the total area available for exchange, it should be noted that brain capillaries are perfused all the time, but they shift to high blood flow with an increase in cerebral blood flow (CBF), or to low blood flow with a decrease in CBF [ 4 ]. Choroid plexuses are villous structures floating in the CSF and attached to the ventricular ependyma by a stalk. The ependyma is continuous with the epithelial layer of the CP which is composed of a single layer of cells filled with mitochondria and joined together by tight TJs (Figure 1 ) [ 5 ]. The TEER offered by these TJs cannot be measured in vivo in most animals. However, in vitro measurements using the single-sided fourth ventricle CP of the bull frog maintained in an Ussing chamber suggested values of about 150 Ω cm 2 [ 6 ], much less than the resistance of the BBB. The low value of TEER would suggest that the CPs fall into the class of leaky epithelia, similar to some segments of the kidney and gut, which form an isotonic fluid and do not generate steep transepithelial concentration gradients across the tissues [ 7 ]. These leaky epithelia can secrete large volumes of fluid but use relatively little energy for this process. CP epithelial cells posses a dense apical coat of microvilli, while kinocilia are rarely found; in contrast to this, the apical surface of ventricular ependymal cells demonstrates a large number of kinocilia [ 8 ], with rare microvilli of variable size. Between the lateral walls of the CP epithelial cells are complex interdigitations particularly apparent close to the blood side of the tissue laying on a basal lamina that demarcate the inner stroma of a highly vascularized connective tissue; these interdigitations expand the surface area of the CP [ 9 ]. Figure 1 Morphology of choroid plexus epithelium (CPE) in situ and in primary culture . A. Ultrastructure: CP from lateral ventricle of an adult Sprague-Dawley rat. Apical membrane (CSF-facing) shows numerous microvilli (Mv) and many intracellular mitochondria (M). J refers to the tight junction welding two cells at their apical poles. C: centriole. G and ER: Golgi apparatus and endoplasmic reticulum. Nucleus (Nu) is oval and has a nucleolus. Arrowheads point to basal lamina at the plasma face of the epithelial cell; the basal lamina separates the CPE above from the interstitial fluid below. Basal labyrinth (BL) is the intertwining of basolateral membranes of adjacent cells. Choroidal morphology resembles proximal tubule, consistent with both cell types rapidly turning over fluid. Scale bar = 2 μm, reproduced from [ 248 ] with permission. B. Phase-contrast micrographs of 8d-old sheep CPE cells cultured on laminin-coated filters shows a typical cobblestone arrangement of polygonal cells (scale bar 20 μm). C. Eight-day-old sheep CPE cells grown on laminin-coated filters were stained with primary antibodie

Adherens junctions (AJs) are specialized cell-cell junctions that are formed by cadherins and associated proteins into which actin filaments are inserted. Optimal function of cadherins requires association of their C terminus with catenins; cadherins bind directly to β-catenin and to p120 catenin, which can bind to α-catenin, a protein that in turn binds actin [ 48 ]. In endothelial cells, vascular endothelial (VE) cadherin is present [ 35 , 49 ]; however, a study has shown that barrier-forming endothelium (i.e. BECs) and barrier-forming epithelium (i.e. CPE) mainly expressed cadherin-10, while the expression of VE cadherin was scarce [ 50 ]. On the other hand, brain microvessels that do not have BBB properties (i.e. in the circumventricular organs and CP capillaries) expressed only VE-cadherin and did not express cadherin-10 [ 50 ]. Also, in the microvessels of glioblastoma multiforme tumors, which lose BBB properties, VE-cadherin was expressed instead of cadherin-10 [ 50 ]. These findings suggest that cadherin-10 has an important role in the development and maintenance of the BBB and the BCSFB. Cadherins regulate endothelial functions by direct activation of phosphoinositide 3-kinase, a signaling system that has a role in organization of the cytoskeleton and forms complexes with the vascular endothelial growth factor (VEGF) receptor 2. Thus, cadherin-mediated signaling is important for endothelial cell layer integrity and for the spatial organization of new vessels [ 51 ]. At least four catenins, β, α, χ and p120 are expressed at the BBB, with β-catenin linking the cadherin to α-catenin which binds the complex to the actin network of the cell skeleton [ 49 ]. However, a study has challenged this view, since it was unable to confirm actin binding to a preformed E-cadherin-β-catenin-α-catenin complex [ 52 ]. As mentioned above, CPE expresses cadherin-10 while CP capillaries express VE-cadherin [ 50 ]. Only two catenins, α and β, have been detected in the CP epithelium so far, with α-catenin binding to the actin network [ 52 ]. In summary, BECs and CP epithelium show many similarities in the organization of Ts and AJs; the main difference is that the CPE provides a barrier that offers lower TEER values and is less restrictive than the BBB. The molecular organization underlying that difference is probably related to expression of different claudins, since those proteins play an important role in barrier size-selectivity and selectivity to paracellular movement of ions.

Brain requires several essential amino acids (AA) for protein synthesis; although the rate-limiting step in brain uptake of circulating amino acids is BBB transport [ 94 ], under normal physiologic conditions the synthesis of brain proteins is not rate-limited by the availability of amino acids [ 95 ]. It was revealed that the influx of amino acids from blood-to-brain approximates the rates of amino acid incorporation into brain proteins [ 94 ]. Most essential AAs are neutral, with long or bulky chains and are substrates for some of the system-L amino acid transporters (LAT) [ 96 ]. It is believed that LAT1 (SLC7A5) is the main AA transporter at the BBB; immunohistochemical analyses have shown that the LAT1 was expressed in the BECs in rats in the luminal and abluminal membranes (Figure 3A ). It has been shown that, in fact, LAT1 activity is induced in Xenopus oocytes by cloned cDNA from mouse encoding 4F2 light chain (4F2lc), but its trafficking and insertion into the cell membrane depended largely on co-expression with 4F2 heavy chain (4F2hc) as 4F2lc-4F2hc covalent complex (which was also known as CD98 membrane antigen) [ 97 ]. This underlines the importance of 4F2 heavy chain in bringing and inserting LAT1/4F2lc into the plasma membrane. Human and rat 4F2hc when inserted into membranes alone induce so-called y+ L-like activity (sodium-independent transport for basic amino acids, and sodium-dependent transport for neutral amino acids). In contrast, transient transfection of rat 4F2hc in Chinese hamster ovary cells results in an increase in L-isoleucine transport with characteristics of system L [ 98 , 99 ]. Thus, it appears that 4F2hc is essential for proper function of LAT1, but this protein itself mediates amino-acid transport. In mouse BECs 4F2hc mRNA was the most abundant among all AA transporters mRNAs, as revealed by qPCR [ 98 ]. However, RT-PCR data and kinetic analysis of [ 3 H]-leucine uptake, revealed that LAT2 (SLC7A6), which has a lower affinity for this substrate, is also expressed in rat BECs in culture [ 100 ]. Kinetic analysis of amino acid transport by the brain provided data that could indicate that both LAT1 and LAT2 show affinity for small neutral AAs, alanine, serine and cysteine [ 101 ]. However, it should be noted that mouse BECs in primary cultures had significantly downregulated all mRNAs encoding AA transporters, as revealed by qPCR [ 47 ]. Some essential amino acid are cationic; these are transported from blood into brain by a Na + -dependent saturable carrier, system y + (SLC7A1) that is present at the luminal side of the BBB (Figure 3A ) and expression of y + in BECs exceeds 38-fold expression in the whole brain homogenate [ 102 ]. Beside LAT1, several other AA transporters are present at the abluminal, brain ISF-facing side of the BECs (Figure 3A ). System A (SLC38A2) (alanine preferring) was first characterized and previous kinetic studies showed that it actively transported small nonessential neutral amino acids [ 103 ]. At least four other Na + -dependent carriers exist at the abluminal membrane: system ASC (SLC1A5) alanine, serine, and cysteine preferring, [ 104 ], system Bo + (SLC7A3) for basic AAs [ 105 ], system N (SLC38A5) for nitrogen rich AA (glutamine, asparagine, and histidine) [ 106 ], and excitatory amino acid transporters (EAAT) (SLC1A1-3), that mediate transport of aspartate and glutamate [ 107 ]. Small AAs, alanine and serine are transported by two Na + -dependent transport systems that are located exclusively in the abluminal membrane [ 105 ]: the system A, which is probably the main route for Na + -dependent alanine transport with a Km of 0.6 mM and system ASC that also shows affinity for large neutral AA. The physiological importance of those two transport systems is unclear, but they may be related to AA efflux from the brain. The sodium-dependent system EAAT deserves attention because it permits a net removal of glutamate from the brain. Glutamate concentration in blood is 50-100 μM [ 107 , 108 ]; in whole brain homogenate it exceeds 10 mM, while in the brain ISF it is normally kept below 2 μM [ 109 ]. Glutamate can exert neurotoxicity if it accumulates in the brain ISF, because, through its action on metabotropic NMDA receptors, it could lead to Ca ++ overload, causing neuronal injury or death [ 110 ]. Glutamate is released during neurotransmission but is normally rapidly taken up by neurons and neighboring astrocytes. However, during cerebral ischemia and/or hypoxia, this AA accumulates in the brain ISF, especially in regions that are rich in glutaminergic neurons. There are three EAATs present at the abluminal side of the BECs, EAAT1 (SLC1A1), EAAT2 (SLC1A2), and EAAT3 (SLC1A3) [ 107 , 110 ] (Figure 3A ) and their action appears to be important to prevent excitotoxicity because they actively remove glutamate from the ISF into the BEC cytoplasm. At the luminal side of the BBB, glutamate is transported by facilitative glutamate transporter X G - [ 110

The delivery of peptides to the brain has important physiological and clinical implications, because in many neurodegenerative diseases it has been found that the application of various growth factors/neuroactive peptides may protect neurons and/or stimulate neuronal growth and repair and, thus, improve outcome for neurological disease. Peptide-based amyloid-β (Aβ)-aggregation inhibitors have been shown to decrease the deposition of Aβ in transgenic mouse models of Alzheimer's disease [ 131 ]. Also, nerve growth factor (NGF) showed the ability to reduce neuronal degeneration in animal models of Alzheimer's disease [ 132 ]. In vitro and in vivo data suggest that treatment with neurotrophic factors such as NGF, glial cell line-derived neurotrophic factor (GDNF), brain-derived neurotrophic factor (BDNF) and several neurotrophins (NTs) could induce survival of specific neuronal populations in Huntington's disease [ 133 ]. Treatment strategies aiming to regenerate existing dopaminergic neurons in Parkinson's disease by applying GDNF, BDNF, IGF and NT-4/5 have also been attempted [ 134 , 135 ]. However, blood-to-brain transfer of intact peptides remains controversial. Peptides cannot use AA transport systems for facilitative transport because of the existence of the peptide bond (for a review see [ 136 ]). Even dipeptides that contain LNAAs do not show measurable affinity for facilitative transport by LAT1 at the BBB [ 137 ]. However, there are specific transport systems that mediate transport of peptides. The peptide transporters that belong to the peptide transporter (PTR) family are solute carrier proteins (SLC15A) responsible for the membrane transport of di- and tripeptides [ 138 ]. Another peptide transporter family (PTS), that contain at least 9 members (PTS1-9) mediate transport of larger peptides (more than 3 AAs in chain) and in many tissues act primarily as an efflux pump, removing lipophilic peptides from cellular membranes. PTR family consists of four members, two peptide transporters PEPT1 and 2 (SLC15A1-2) and two histidine transporters that also transport dipeptides (PHT1 and 2, SLC15A3-4). PTRs couple substrate movement across the membranes to movement of protons down an inwardly-directed electrochemical proton gradient [ 138 ]. Early studies have shown that arginine vasopressin (AVP) [ 139 ], enkephalins [ 140 , 141 ], delta-sleep-inducing peptide (DSIP) [ 142 ] and luteinizing-hormone-releasing hormone (LHRH) had a measurable volume of distribution in the guinea pig brain after in situ perfusion but the rates of blood-brain transfer were 10 3 -10 4 fold lower than rates of carrier-mediated amino-acid transport. Tetrapeptide tyrosine melanocyte-stimulating inhibitory factor 1 (Tyr-MIF-1) was the first peptide shown to pass from blood to the brain by a saturable system [ 143 ]. Although there is no evidence so far that any of the PTR four members, that could mediate efflux transport of di- and tri-peptides are present in the BECs, these cells probably express at least some PTS members located at the abluminal side that mediate efflux transport of several small peptides from the brain ISF: enkephalins, Tyr-MIF-1, arginine vasopressin (AVP) and LHRH [ 144 ]. For example, pituitary adenylate cyclase-activating polypeptide (PACAP), which has neuroprotective effects against ischemia, can pass across the BBB, but its efflux, which is mediated by PTS-6, severely restricts its net entry into the brain ISF. However, when PTS-6 expression in BECs was inhibited by antisense targeting, brain accumulation of PACAP increased significantly [ 145 ], which indicates that the main role of this transporter is efflux transport. Thus, it appears that BBB transport system for peptides could be involved in impeding blood-to-brain ISF transfer of intact peptides. The brain delivery of peptides is further impeded by the existence of various enzymes in BECs that modify AA side chains or hydrolyze peptide bonds. These enzymes include γ-glutamyl transpeptidase, aromatic acid decarboxylase, dipeptidyl(amino)peptidase IV, and aminopeptidases A and N [ 146 ]. However, it has been shown that some neuropeptides, when present in capillaries, could be transferred to the brain ISF in intact form, like DSIP [ 147 ]. Larger peptides and proteins that have receptors present at the luminal side of BECs could use receptor-mediated transcytosis to pass across the BBB and that mechanism was revealed for insulin [ 148 ], transferrin [ 149 ], certain cytokines [ 150 ], leptin [ 151 , 152 ], immunoglobulin G [ 153 ], and insulin-like growth factor [ 154 ]. It seems that Aβ could also pass the BBB by receptor-mediated transcytosis. This peptide (MW ~4500 Da) is bound in plasma to several proteins, including albumin, apolipoprotein E (apoE), apolipoprotein J (apoJ), transthyretin (TTR), α2-macroglobulin (α2M) and low-density lipoprotein receptor related protein-1 (LRP1) [ 155 - 157 ]. There is evidence which suggests that influx of Aβ into the

There is evidence suggesting that there is a bulk flow of the brain ISF from brain capillaries towards the ventricular space and that ISF merges with the CSF; this flow takes place predominantly along perivascular spaces (for a review see [ 237 ]). This indicates that there is a constant production of a "new" ISF in the brain which contributes to total volume of the CSF. Although some of the ISF is probably generated from water produced by brain metabolism, fluid secretion by BECs appears to be an important source of this ISF [ 237 ], accounting for at least 30% of the ISF production [ 238 ]. This process is essential for maintaining correct fluid balance in the brain. Two membrane proteins that work simultaneously but at different membranes of the BECs are key regulators of net sodium and chloride transport across the BBB: Na + , K + -ATPases (ATP1 family) and the Na + , K + , 2Cl - cotransporter (SLC12 family). The Na + , K + -ATPase is localized on the abluminal membrane and provides a driving force for the net ion and water movement across the BBB [ 239 ] (Figure 5A ); 3 alpha- (ATP1A1-3) and 2 beta- (ATP1B1-2) subunit isoforms were found in rat BECs, which means that six structurally distinct Na + , K + -ATPase isoenzymes are likely to be expressed in brain microvessels [ 240 ]. As in other tissues, the activity of this enzyme is tightly associated with cell volume regulation [ 241 ]. The Na + -K + -2Cl - - cotransporter is located at the luminal side of BECs [ 242 ] and its activity is regulated by PKC signalling [ 243 ]. In addition, rat brain endothelial cells express Kv1 and Kir2 potassium channels; these are probably located on both the luminal and abluminal sides of the BECs [ 244 ] (Figure 5A ). Furthermore, BECs also express two ion exchangers that belong to the SLC family and are probably involved in intracellular pH regulation: chloride-bicarbonate (Cl - , HCO 3 - -) exchanger (SLC4A1) and sodium-hydrogen (Na + , H + -) exchanger (SLC4A6). Both isoforms of Na + , H + -exchanger (Nhe1 and Nhe2) are expressed on the luminal membrane, whereas chloride-bicarbonate exchanger(Ae1) is expressed at both luminal and abluminal membranes of the BECs (Figure 5A ) [ 245 ]. Net flux of K + at the BBB is critical for brain homeostasis, since changes in concentration affect resting membrane potential; however, net flux of this ion across the BBB is not well understood. Both Na + , K + -ATPase and the Na + -K + -2Cl - cotransporter bring this ion into the BECs and at least two potassium channels exist on both sides (Figure 5A ). A hypothesis suggested that there was a net K + efflux across the BBB, which contributed to low K + concentration in the brain ISF [ 246 ]. However, the CSF recovered from the brain, contains 2.5-3.0 mM K + and the CSF originates either from CP secretion or from the brain ISF, although the relative contribution of those two sources is a matter of debate. Thus, at least one of these two (CSF secreted by the CPS or brain ISF secreted by the BECs) have to be a source of K + in the CSF. CPs in fact mainly reabsorb K + from the CSF (see below). Thus, bearing in mind conservation of mass, a speculation could also be made that K + found in the CSF is, at least partially, K + that was secreted at the BBB. Figure 5 Distribution of ion transporters and channels in the BECs (A) and in the CP epithelium (B) . Only those transporters and channels that play a role in vectorial transport of Na + , Cl - , HCO 3 - and K + are shown. These include Na + , K + -ATPase, potassium channels Kir and Kv, chloride/bicarbonate channel Clir and members of the SLC: Na + , HCO 3 - cotransporters 1 and 2 (NBCn1 and 2), Cl - , HCO 3 - exchangers 1 and 2 (Ae 1 (in the BBB, figure A) and Ae2 (in the CP, figure B), Na + , H + exchanger 1 and 2 (Nhe1 and Nhe2), K + -Cl -- cotransporters 3 and 4 (KCC3 and 4, in the CP, figure B), Na + -K + -2Cl- cotransporter 1 (NKCC1, in the BBB, figure A) and electrogenic Na + , HCO 3 - exchanger (NBCe2, in the CP, figure B). In addition, localization of two carbonic anhydrase isoenzymes in the CP epithelium (figure B), CA2 and CA12 are shown, as well as localization of aquaporin 1 (AQP1). Symbol * indicates that NBCn1 was detected in CPE, but it probably does not play a role in vectorial transport of these two ions. Ion transport at the BBB also plays an important role in regulation of endothelial cell pH, especially bicarbonate transport driven by Cl - , HCO 3 - -exchanger and H + transport driven by Na + , H + -exchanger. It has been shown that following in vitro loading of BECs with small acid load, HCO 3 - influx was mainly responsible for the acid extrusion and it was mediated partially by Cl - dependent HCO 3 - transporters. However, after large acid loads, BECs removed acid almost solely by Na + , H + exchange, with the rate of its activity depending linearly on intracellular pH [ 247 ]. Following an alkaline load to BECs, the intracellular pH was restored by acid loading which

α 2M: α2-macroglobulin; AA: amino acid; AD: Alzheimer's disease; AJ: adherens junctions; ANP: atrial natriuretic peptide; AVP: arginine vasopressin; BCRP: breast cancer resistance protein; BCSFB: blood-cerebrospinal fluid barrier; BDNF: brain-derived neurotrophic factor; BEC: brain endothelial cells; CAR: androstene receptor; CBF: cerebral blood flow; CP: choroid plexus; CPE: choroid plexus epithelium; CSF cerebrospinal fluid; DEP: diesel exhaust particles; DSIP: delta-sleep inducing peptide; EAAT: excitatory amino acid transporter; GDNF: glial cell line-derived neurotrophic factor; HIF-1: hypoxia-inducible factor 1; HUGO: Human Genome Organization; ISF: interstitial fluid; JAM: junctional adhesion molecules; JNK c-Jun N-terminal kinase; LAT: system-L amino acid transporter; LRP1: low-density lipoprotein receptor related protein-1; MCT: monocarboxylate transporter; MRP: multidrug resistance-related protein; NFκB: nuclear factor-κB; NGF: nerve growth factor; OAT: organic anion transporter; OATP: organic anion transporting polypeptide; P-gp: P-glycoprotein; PKC: protein kinase C; PTR: peptide transporters; PXR: pregnane-X receptor; qPCR- real time PCR; RAGE: receptor for advanced glycation end products; RAP: receptor-associated protein; SLC: solute carriers; TBI: traumatic brain injury; TEER: transendothelial/transepithelial electrical resistance; TGF: transforming growth factor; TJ: tight junction; TTR: transthyretin; VEGF: vascular endothelial growth factor; ZO: zonulla occludens;

ZR: sole author. The author has read and approved the final version of the manuscript.