Glucose

The glucose transporter at the BBB is of special importance because glucose is the primary source of energy for the brain and is required for normal brain activity and function. Transportation across the BBB occurs via a glucose transporter. There are five members of the sodium-independent glucose transporters in the brain, including GLUT-1 (ECs), GLUT-3 (neurons),⁴³ and GLUT-5 (microglia).⁴⁴ Each transports 2-deoxyglucose, 3-O-methylglucose, mannose, galactose, and glucose across the membrane.⁴⁴ GLUT-1 is a 45- to 55-kD protein, depending on its glycosylation state. It is present in high concentration in the ECs of arterioles, venules, and capillaries and facilitates movement of d-glucose enantiomers from the peripheral circulation to the brain.⁴⁵ Although the BBB is small, glucose transport is the rate-limiting step of primary energy acquisition. GLUT-1 expression is three to four times higher on the abluminal membrane and is altered with processes such as diabetes, epilepsy, trauma, and tumors.

Amino Acids

Delivery of amino acids across the BBB is achieved by carrier-mediated transport across the abluminal and luminal membranes in both a concentration-dependent and stereospecific manner.⁴⁶ Factors affecting amino acid uptake include plasma concentration, affinity of the transport system for a particular amino acid, and competition among amino acids for a particular transport system. Amino acid transport is classified into four systems based on seminal experiments conducted on the BBB and amino acid transport.⁴⁷ The first group is the large, neutral-charged amino acid transporter (L type), which has a preference for leucine residues. These transporters are sodium independent, saturable, and stereospecific. The rate of transfer across the BBB is high because of the high requirement for this subtype of amino acids by the brain. The second group of transporters consists of small, neutral-charged amino acids (A type). These transporters are dependent on sodium for movement across the BBB. Independent transporters for acidic and basic amino acids make up the third transport system. The rate of transfer of basic amino acids is high because of a high requirement for this subtype by the brain. Finally, alanine, serine, cysteine, and threonine residues use a sodium-dependent transport system that shuttles small neutral amino acids across the BBB. Investigation of this transport system has suggested that its primary role is to transport amino acids out of the brain.⁴⁸

Glutamate is the most abundant amino acid in the CNS and is stored intracellularly against its concentration gradient. It is an excitatory neurotransmitter that serves a number of functions in the brain. It is involved in energy metabolism, molecular synthesis of glutathione and GABA, and breakdown and removal of ammonia. In addition to transport systems, the BBB also plays a role in glutamate metabolism.⁴⁹

Neurotransmitters

A number of neurotransmitters cross the BBB, including glutamate, aspartate, glycine, and GABA. In many instances, amino acid precursors cross the BBB before chemical modification into a functionally active protein. Such precursors include tryptophan, tyrosine, and histidine, which are modified to serotonin, dopamine and norepinephrine, and histamine, respectively. A classic example of the role of the BBB in synthesis is levodopa transport and dopamine synthesis. Levodopa is structurally similar to a neutral amino acid and enters ECs via a large neutral amino acid carrier. ECs contain dopa decarboxylase, which converts levodopa to dopamine. The amount of dopamine generated in the brain is limited by EC expression of dopa decarboxylase. Other enzymes regulated by BBB metabolism include monoamine oxidase A and B, catechol O-methyltransferase, and pseudocholinesterase.

Ions and Water

Adequate ion transport across the BBB is critical for brain function. Na⁺,K⁺-ATPase is a well-established ion exchange protein on brain ECs. This channel protein controls K⁺ levels by pumping Na⁺ out of cell and K⁺ into cell to preserve an electrochemical gradient across the membrane.

Nitric oxide (NO) is well known for its properties as a vasodilatory agent and also plays an important role in BBB signal regulation and autocrine activity.^50,51 It mediates the transport of ions and nutrients essential for brain function and is regulated by cyclic guanosine monophosphate and cyclic adenosine monophosphate.

All movement of ions across the BBB is also associated with the movement of water. Water passes through the plasma membranes via facilitated diffusion through water channels called aquaporins, by cotransport with organic or inorganic ions, and by diffusion across the lipid bilayer. The water channel expressed most in the CNS is aquaporin-4, principally by astrocytes. Immunolocalization studies with double staining for aquaporin-4 and glial fibrillary acidic protein (GFAP), which stains astrocytes, show strong colocalization at the level of the BBB.^52,53 Involvement of aquaporin-4 in brain edema and water homeostasis has been well established.

Lipoproteins

The mechanism underlying lipoprotein transport across the BBB has garnered considerable interest among researchers for potential exploitation of this channel for delivering drug therapy. Low-density lipoprotein (LDL) receptors are expressed on the BBB, and LDL is transported across the BBB by endocytosis. Scientists are now attempting to use this receptor as a vehicle for delivering therapeutic drugs into the brain.⁵⁴

Multidrug Resistance

Multidrug resistance (MDR) protein has also been intensely studied as a possible vehicle for drug delivery. Pgp is an efflux transporter protein found in ECs, astrocytes, and microglia.⁵⁵ It is expressed on the luminal surface of the endothelial membrane and glia⁵⁵ and prevents toxins from entering the brain.⁵⁶ Many drugs are substrates for Pgp, which limits their accumulation in the brain. Vinca alkaloids, anthracyclines, and taxanes are among the anticancer agents known to be transported by Pgp. Preclinical models have revealed that patients with deletion of Pgp have 100-fold increased sensitivity to chemotherapy agents and antiviral compounds in comparison to control subjects.^57–60 Pgp is also protective in disease states such as Alzheimer’s, in which there is decreased deposition of β amyloid.⁶¹ However, overexpression of Pgp is also found in patients with epilepsy, although it is unclear whether upregulation of the transporter is a pathologic process of epilepsy or secondary to resistance to antiepileptic drugs (AEDs).⁶² In vivo studies have shown depletion of Pgp in patients with Parkinson’s disease.⁶³

Leukocytes

It was an early notion that leukocytes are rare within the brain and that the architecture of brain microvessels maintains its immune-privileged status. There is now evidence that leukocytes traverse microvessels via a transcellular route. In addition, activated T lymphocytes can cross the endothelial wall in the normal state. Immunologic response to foreign pathogens, surveillance, and inflammation are regulated by leukocyte transportation across the BBB. Transportation occurs by means of integrins, intercellular adhesion molecules (ICAMs), and direct intercellular activity.^64,65 In HIV encephalitis, actin cytoskeletal proteins such as Rho guanosine triphosphatase (GTPase) facilitate leukocyte migration, and blocking these compounds maintains the integrity of the BBB against leukocyte transport.⁶⁶

Membrane-Associated Guanylate Kinase Homologue

This class of proteins forms multimers that anchor TJs to the cell membrane. There are three main adhesion proteins: zona occludin-1 (ZO-1), ZO-2, and ZO-3. The first TJ-associated protein identified was ZO-1, a 220-kD membrane-associated protein expressed on both EC and epithelial cell surfaces.⁶⁷ It serves as a scaffold for TJ formation by binding to the C-terminal of the cytoplasmic tail of occludin and the cytoskeletal protein spectrin.⁶⁸ The importance of ZO-1 in TJ stability is illustrated by research showing that barrier stringency is compromised when ZO-1 expression is reduced or dissociated from membrane proteins.⁶⁹ ZO-2 is a 160-kD protein with homologous regions to ZO-1 and has been identified as a ZO-1–associated protein.⁷⁰ It binds to a transcription factor and transmembrane domains of the TJ. During proliferation of the TJ, ZO-2 migrates to the nucleus.^71,72 Although ZO-1 and ZO-2 are seen in both brain and peripheral ECs, ZO-1 is continuous in brain but discontinuous in peripheral ECs.⁷³ ZO-3 is seen solely in epithelial cells, unlike ZO-1 and ZO-2, which are intricate components of ECs.⁷⁴

Occludin

Occludin is a 65-kD transmembrane protein located at TJ margins^19,75 and is associated with cytoskeletal signaling proteins, including ZO-1, and ZO-2.⁷⁰ Antibodies against occludin cluster around cell-cell contacts in brain ECs.⁷⁶ Occludin is composed of four transmembrane domains. The terminal ends of the protein face the cytoplasm, and two extracellular loops span the intracellular cleft.⁷⁷ The concentration of occludin in peripheral ECs is much lower than in brain ECs. Occludin is thought to be transcriptionally regulated, as measured by significant differences in mRNA levels between EC types.⁷³ Occludin contains numerous phosphorylation sites, which are directly related to substance permeability. The cytoplasmic C-terminal provides association among occludin, ZO-1, and ZO-2. Occludin dysfunction is seen in numerous diseases such as HIV.^66,78 Furthermore, downregulation of 55-kD occludin expression is seen in high-grade (III and IV) brain tumors.⁷⁹

Claudins

Claudins are a third group of membrane proteins. These proteins function to selectively permit entry of cations through TJs.⁸⁰ There is sequence conservation in the first and fourth transmembrane domains in the approximately 24 claudins identified in mammals.⁸¹ Both homophilic and heterophilic binding of extracellular domains facilitates close interaction of cell layers.⁸² In elucidating the roles of claudins and occludin in TJ formation, overexpression of claudin generated cell layers resembling TJs, which was not the case with occludin.⁸³ Based on staining experiments, claudins are arranged continuously along ECs, in contrast to occludins.⁸³ These results imply that claudins form the fundamental scaffold of TJs and occludins secure TJs. ECs express at least four claudin subtypes, claudin-1, claudin-3, claudin-5, and claudin-12. In vitro studies have revealed that claudin expression is significantly downregulated in anaplastic astrocytoma and GBM cell cultures.^84,85

Junctional Adhesion Molecules

Members of the immunoglobulin supergene family are localized within TJs. Junctional adhesion molecule-1 (JAM-1) is a 40-kD IgG protein that regulates cell membrane attachment between large-domain and smaller chains.^86,87 In animal models, JAMs regulate transendothelial cell migration, but their role in the BBB in vivo remains unclear.⁸⁶

Endothelial TJs can rapidly and reversibly alter their conformation to permit plasma constituents to pass. The mechanism of control of this system is the degree of tyrosine phosphorylation of ZO-1 or other contact proteins; with increased phosphorylation, there is decreased electrical resistance and TJ permeability increases.⁸⁸ Other mechanisms of control of cell-cell junctions are regulated by factors secreted by astroglial cells. These factors alter the microenvironment of the junction, thereby modifying permeability of the TJ.

Ion Signaling

Phosphorylation indirectly controls TJ permeability by acting on associated proteins.⁸⁹ Various pathologic states affect the function of phosphorylation on TJs. For example, there is significant phosphorylation of TJs in the presence of vascular endothelial growth factor (VEGF), which compromises the integrity of the BBB.⁹⁰ In contrast, in bacterial infections there is significant dephosphorylation of TJs, which increases BBB permeability.^91,92

Calcium

Calcium (Ca²⁺) acts as second messenger for regulation of BBB activity. Low levels of extracellular Ca²⁺ decrease TJ barrier stability,⁹³ possibly dispersing ZO-1 and occludin away from TJ sites.⁹⁴ Intracellular Ca²⁺ levels also affect BBB integrity by activating signaling cascades, regulating TJ transcription, and modulating the distribution of TJs.⁹⁵ Release of intracellular Ca²⁺ compromises the BBB by inducing phosphorylation of TJs.⁹⁶

G Protein

G proteins regulate scaffold proteins responsible for the cytoarchitecture of the TJ and membrane-associated proteins regulating TJ activity.⁹⁷ They also facilitate leukocyte transport into brain via Rho GTPases.⁶⁶ Protein kinase C (PKC) regulation also modulates BBB function. Control of PKC-α and PKC-ζ influences BBB permeability.⁹⁸ In HIV disease, the cell surface protein gp120 activates PKC. Inhibitors of gp120 activity block PKC activation and prevent breakdown of the BBB.⁹⁹ AF-6 is a Ras-binding protein that also binds to ZO-1 to stabilize the BBB. Overexpression of Ras results in disruption of ZO-1 and increased permeability of the BBB.¹⁰⁰

Importance of Adequate Central Nervous System Drug Penetration

Improved regulation of BBB permeability would result in increased efficacy in treating the vast majority of CNS disorders. Nevertheless, most research in this area has focused on a handful of specific diseases, including brain tumors, HIV disease, epilepsy, Parkinson’s disease, Alzheimer’s disease, and infections. Both primary and metastatic CNS neoplasms, the most common of which in adults are astrocytomas, oligodendrogliomas, and mixed oligoastrocytomas,¹¹⁹ are difficult to treat because most chemotherapeutic agents are unable to adequately traverse the BBB as a result of their low lipophilicity.^40,119 Partly as a consequence of poor drug delivery to the CNS, many tumors are currently treated by surgery and radiation therapy. However, difficulties often arise in surgery because precise boundaries of some tumors are difficult to locate. Other tumors may lie in inoperable locations or have already metastasized to multiple sites. Although radiotherapy is often beneficial as well, it may lead to secondary, more aggressive tumors.^120,121 Chemotherapy is the best option in these cases; however, the drugs must be able to enter the CNS to be effective.

AEDs are often excluded from the CNS by the BBB and are therefore clinically unusable despite demonstrating potent and selective in vitro action.¹²² Pgp has been implicated as an important transport protein for some of these AEDs,¹²³ including phenytoin¹²⁴ and carbamazepine.¹²⁵ However, other reports question the role of Pgp in AED extrusion.^126,127 Approximately 30% of epileptic patients do not respond to the common pharmaceutical treatments, thereby resulting in increased morbidity and mortality in these resistant patients.¹²⁸ In an immunohistochemical study, Volk and Loscher¹²⁹ demonstrated that rats resistant to AED therapy had higher levels of Pgp expression in brain capillary ECs than did control animals. Similar variations in Pgp in human epileptics would help explain the existence of patients with medically intractable epilepsy who do not respond to AEDs.

In the treatment of Parkinson’s disease, it is estimated that only 5% of orally administered levodopa reaches the circulation after first pass through the liver. Less than 1% of the oral dose enters the brain because of blockage at the BBB.¹³⁰ Recent studies are beginning to elucidate the potential mechanisms for exclusion of levodopa from the CNS. Levodopa enters the brain via the luminal neutral amino acid transporter L1 on ECs, but it may be excreted by the sodium-dependent large neutral amino acid (LNAA) transporter on the abluminal side of the membrane.^131,132 Other molecules of levodopa are degraded by the endothelial enzyme monoamine oxidase B in the periphery,¹³³ although drugs such as carbidopa are now available to inhibit this enzyme.

Medications for Alzheimer’s disease are excluded by the BBB as well.¹³⁴ Given the rising proportion of elderly individuals in whom this disease is diagnosed in the United States, effective treatment of Alzheimer’s disease will be critical. Certain modifications of the BBB may eventually be used to provide better treatment of this debilitating illness.

Infectious disease could also be treated more successfully by opening the BBB, which would allow CNS penetration of more effective antibacterial, antifungal, and antiviral agents for the treatment of meningitis, encephalitis, abscesses, and other infectious diseases of the CNS. Once bacteria are killed, the BBB must be permeable to the remnants of their cell walls, which could otherwise irritate sensitive nervous tissue.¹³⁵

HIV may harbor itself in the brain where antiviral drugs cannot penetrate effectively and then re-emerge later.^136,137 Several studies have shown that zidovudine (azidothymidine [AZT]) does not effectively penetrate the BBB in rodents,^138–140 although this does not appear to be the result of Pgp export.¹⁴¹ Strazielle and colleagues¹⁴² suggested that the organic anion transporters, a different class of protein pumps at the BBB, may be responsible for transport of AZT. Other anti-HIV medications are exported by similar means. The nucleoside reverse transcriptase inhibitor (NRTI) stavudine (d4T) may be removed from the BBB by the same transporter as AZT.¹⁴³ Another NRTI, lamivudine (3TC), does not significantly penetrate the CNS either.¹⁴⁰ Some protease inhibitors used for the treatment of HIV also appear to be susceptible to BBB exporters, primarily Pgp.¹⁴⁴ However, there are some anti-HIV drugs that do enter the CNS in significant amounts. Some non-NRTIs, including nevirapine, appear to cross the BBB,¹⁴⁵ whereas other anti-HIV drugs are believed to cross the BBB as a result of their CNS side effects.¹⁴⁶

Drug Modifications

Expanding on the initial experiments of Brodie and associates⁸ in the 1960s, recent studies have shown that the permeability of most molecules can be predicted by determining their octanol-water coefficients based on their respective nonpolar and polar solubilites.^147,148 Specifically, substances with the greatest ability to pass through the BBB generally have a log octanol-water coefficient between −0.5 and 6.0¹⁴⁹ and a molecular mass of less than 400 to 500 daltons and do not form hydrogen bonds with water.¹⁵⁰ With this understanding, many pharmaceutical researchers have conjugated their bioactive compounds to lipophilic moieties in the hope that they will become sufficiently lipid soluble to passively move through the BBB. Others have masked the hydrophilic groups of the compounds in an effort to increase lipophilicity.⁴² Conjugation by esters and disulfide bonds allows enzymes to cleave the lipids from the drug once it has passively entered the CNS, thus making the drug polar and trapping it inside. An example is heroin, an opiate that passes through the BBB 100 times more easily than morphine and is subsequently converted to morphine in the CNS.^141,151 Such prodrugs have proved to be useful, although more research is needed to evaluate their efficacy and safety.¹⁵² Additionally, cleavage to form the active drug may not occur at a sufficient rate and with the necessary accuracy to produce localized therapeutic concentrations of the drug.

Lipophilic conjugation has been used successfully for introduction of the chemotherapeutic agent chlorambucil into the CNS.^153,154 Kitagawa and coworkers¹⁵⁵ evaluated the conjugative properties of adamantine, a compound related to the drug memantine used for the treatment of Parkinson’s disease. By conjugating adamantine to [D-Ala2] leu-enkephalin, the opioid gained the ability to pass through the BBB into the CNS. In another experiment, Prokai-Tatrai and colleagues¹⁵⁶ successfully conjugated a different leu-enkephalin analogue to a lipophilic moiety for CNS administration. However, conjugation of drugs may not always be necessary. The simple reduction of hydrogen bonding potential by altering polar side groups has successfully increased permeability of the BBB to some small peptides.¹⁵⁷

Despite these successes, many difficulties have arisen in the search for successful methods to increase the lipophilicity of pharmacologic compounds. Modification often increases the mass of the drug, and even lipophilic drugs do not cross the BBB effectively when their mass has increased to greater than the 400- to 500-dalton threshold.⁴² Increased size of the drugs can affect transport as well. The limit for molecular area appears to be around 80 Å², and increases in size to greater than this seem to decrease BBB permeability dramatically.¹⁵⁸ In addition to the physical constraints on lipophilic conjugation, chemical constraints have been demonstrated as well. Conjugation or masking of hydrophilic side groups may make the drug biologically inactive.¹⁵² An increase in lipophilicity can also make the drug susceptible to transport by Pgp and other export proteins despite having little if any susceptibility before the modification. Alternatively, conjugation of some drugs that are already substrates for Pgp can have an added beneficial effect by successfully preventing their export through hindrance of their ability to attach to the Pgp binding site. A derivative of the chemotherapeutic agent paclitaxel has been successfully modified in this fashion while still maintaining cytotoxic action against cells of the breast cancer lineage.¹⁵⁹ Drug modification can alter pharmacokinetic parameters as well. Conjugation may decrease the solubility, plasma protein binding, and liver and reticuloendothelial uptake, thus altering the bioavailability of the drug.¹⁵² New side effects may also be due to increases in drug uptake into other organs as a result of lipophilic modification and potentially damage the more sensitive organs.

Another strategy of drug modification for bypassing the BBB is conjugation to bioactive molecules, either those that are normally transported into the CNS by specific transport proteins or some that enter the CNS via receptor-mediated endocytosis by cerebrovascular endothelium. Friden and associates¹⁶⁰ successfully conjugated nerve growth factor to an antibody for the transferrin receptor in the rat. Binding of the antibody to the receptor stimulates receptor-mediated endocytosis and provides transcellular passage through the endothelium. Other researchers have successfully conjugated drugs to insulin fragments or antibodies to insulin factor to permit transfer through the BBB.^46,161 Even large molecules, such as the enzyme β-galactosidase, have been successfully transported into the CNS via similar methods of bioactive conjugation.¹⁶²

Exporter Protein Modulation

The most well characterized export protein of the BBB, Pgp, was first described in hamsters in 1976 by Juliano and Ling.¹⁶³ As a member of the APT-binding cassette family of transport proteins, Pgp serves to protect the CNS by pumping xenobiotic compounds out of the brain and spinal cord into the vasculature.¹⁶⁴ In general, substrates for Pgp are lipophilic, planar molecules that are either neutral or cationic.¹⁶⁵ Pgp may have developed to remove hydrophobic substances that partitioned into the lipid core of the plasma membrane^166,167 and therefore may be activated by sensing disruptions of the lipid bilayer.¹⁶⁵ Unfortunately, the broad specificity of Pgp, although beneficial in preventing penetration of neurotoxins into the brain and spinal cord, also hinders therapeutic drugs from reaching their targets, thereby creating great difficulty for those researching CNS pharmaceutical design.

The specific location of Pgp in the BBB has recently been a source of debate. Studies have shown that Pgp is expressed in both cerebrovascular ECs and astrocytes in the human brain.^168,169 Many studies suggest that the primary localization of Pgp is on the luminal EC membrane, where it serves to pump compounds directly into the lumen of the microvasculature.^23,170–173 However, some evidence points to localization of Pgp on astrocyte foot processes as well.^168,169 Although Pgp is important in the case of toxin exposure, animal experiments suggest that it may not have any necessary function during normal homeostasis. Mice deficient in the export protein have no changes relative to the wild type unless exposed to drugs that are normally pumped out of the cerebrovascular endothelium by Pgp.¹⁶⁴ In a related experiment, dogs with increased susceptibility to neurotoxicity with the administration of ivermectin were found to be deficient in Pgp as a result of a deletion mutation in the mdr1 gene.¹⁷⁴

In the treatment of Parkinson’s disease, levodopa is often administered along with carbidopa to increase levels of dopamine in the brain. Carbidopa does not cross the BBB but rather serves to inhibit the peripheral decarboxylases that would convert levodopa to dopamine and thus prevent its entry into the CNS. Similarly, inhibitors of Pgp have been developed to be coadministered with CNS-acting medications for the treatment of a variety of conditions.

Despite its high lipophilicity, cyclosporine was found to ineffectively penetrate the CNS as a result of interactions with Pgp. Studies have shown that cyclosporine inhibits Pgp, but it is not generally used for this purpose because of its immunosuppressive effects.^175–177 The cyclosporine analogue PSC 833 was developed later and maintains the Pgp inhibitory action of its parent drug without the resulting immunosuppression.^178,179 An array of other Pgp inhibitors have since been developed^180–182 and used successfully with a variety of medications.¹⁶⁵ For example, Pgp inhibitors have been shown to enhance delivery of the chemotherapeutic drugs paclitaxel^183,184 and docetaxel to the CNS in animals.¹⁸⁴ Recently, some natural products have been demonstrated to have Pgp inhibitory characteristics as well, including psoralen found in grapefruit,^185,186 ginsenoside Rg₃ from red ginseng,¹⁸⁷ and piperine from black pepper.¹⁸⁸ Clearly a variety of compounds have effects on Pgp, and as knowledge of these substances increases, researchers should be able to develop more effective compounds.

The major drawback of Pgp inhibitors is that historically, many have been toxic in concentrations sufficient for inhibition. However, the toxicologic profiles of the inhibitors have improved in recent years. First-generation inhibitors such as cyclosporine and verapamil were very toxic, whereas second-generation inhibitors such as valspodar and biricodar had improved tolerability.¹⁸⁹ Second-generation inhibitors were later found to have unpredictable interactions when coadministered with chemotherapeutic agents. The recently developed third-generation inhibitors, including tariquidar, zosuquidar, and laniquidar, are very specific for Pgp and have less interaction with coadministered drugs. These new Pgp inhibitors are currently undergoing clinical trials.¹⁹⁰

Opening of Endothelial Tight Junctions—Osmotic Disruption

Although most vascular ECs contain no TJs, epithelial barriers throughout the body possess these intercellular structures to ensure that the luminal contents remain separate from underlying tissue. TJs between ECs of the cerebral microvasculature seem to be able to separate, effectively open up the BBB, and allow the paracellular passage of water and other molecules. Although the actual mechanism for this type of nonspecific BBB opening remains to be elucidated, a variety of substances appear to alter the permeability of the BBB in this manner. The most common of these techniques used today is osmotic disruption, which involves the injection of hypertonic nonmetabolizable solutes such as mannitol directly into either the internal carotid arteries or the vertebral arteries. The location of injection is important because the permeability of the BBB will be affected only in the vasculature distal to the site of solute injection.¹⁹¹ CNS drug delivery via this method of BBB opening has been shown to be increased by up to 100-fold.¹⁰

The predominant hypothesis for the mechanism of osmotic BBB disruption is that shrinking of ECs occurs as a result of hyperosmolarity of the vasculature. This reduction in cell volume causes the TJs to separate and thereby open up the paracellular space for molecular movement.^192,193 However, newer studies suggest that this model may be an oversimplification. Farkas and coworkers¹⁹⁴ have provided evidence of phosphorylation of the multifunctional protein β-catenin during osmotic BBB opening, thus suggesting a more active cell response than passive shrinkage. A calcium-mediated contraction mechanism involving circumferential bundles of actin that interact with a variety of proteins in ECs has been proposed as well.^195,196 Exposure to hypertonic substances appears to increase intracellular calcium in cultured cerebrovascular ECs, which may trigger pathways leading to cell shrinkage itself.¹⁹⁷

Several studies have demonstrated the benefits of osmotic BBB disruption for the delivery of chemotherapeutic agents to the CNS to treat both primary and metastatic malignancies. Neoplasia represents a unique challenge for administration of pharmaceuticals. The BBB is often more permeable at the core of a tumor, but the brain tissue adjacent to the tumor maintains a high level of BBB integrity.^10,198 Consequently, significantly less drug reaches the periphery and most goes to the center of the tumor.¹⁹⁹ This “sink effect” causes very low levels of drug in the brain tumor periphery, in essence sparing tumor cells and contributing to neoplastic recurrence.¹⁰ Additionally, the surrounding normal brain tissue may contain metastatic seeds that are hidden behind a relatively normal BBB, thus further confounding effective treatment.¹³³ Evidence also suggests that although the BBB is leaky in the center of tumors, it does not disappear completely in either primary or metastatic lesions.^200,201 In one experiment in mice, Pgp still played a role in the microvasculature of brain tumors despite their increased permeability in comparison to normal brain tissue.²⁰² The use of osmotic BBB disruption in conjunction with chemotherapy helps maintain a relatively constant level of chemotherapeutic agents throughout the entire region around the tumor, thereby preventing the “sink effect” and exposing any smaller metastatic lesions that would otherwise be inaccessible to the drugs. Osmotic BBB disruption also allows prolonged exposure of tumor to higher localized concentrations of drugs.^203,204

The first phase I trial of osmotic BBB disruption began in 1979 and used mannitol (25%, 1.37 mol/L) infused via a catheter into the internal carotid artery at 4 to 8 mL/sec for 30 seconds. A transient rise in intracranial pressure occurred in this study, but no clinical sequelae were observed. In a study by Dahlborg and colleagues,²⁰⁵ 645 osmotic BBB disruptions in conjunction with chemotherapy were performed on 34 patients, and more than 80% showed a partial response and 62% showed a complete response. Kraemer and coworkers²⁰⁶ found increased survival rates in patients with primary CNS lymphomas who underwent osmotic BBB disruption for delivery of chemotherapeutic drugs. They also identified a positive correlation between patient survival and the number of BBB disruptions. In another study, the estimated 5-year survival rate for patients with non–AIDS-related primary CNS lymphoma treated by osmotic BBB disruption and chemotherapy was 42% with a median survival time of 40.7 months.²⁰⁷ Such techniques have also been used successfully in pediatric patients.²⁰⁸

Despite its increased use in recent years, osmotic BBB disruption is far from perfect. Early hints of potential difficulties with the procedure came from studies in rats in which seizure-like events occurred during BBB modification.²⁰⁹ Transient increases in intracranial pressure are common during osmotic disruption. In the aforementioned study by Dahlborg and associates,²⁰⁵ which included 34 patients, one episode of tonsillar herniation occurred with no neurological sequelae, seizures developed in 4% of patients, and sepsis or granulocytopenic fever developed in 3%. In a study by McAllister and coworkers²⁰⁷ on primary CNS lymphoma, seizures occurred in 6% to 8% of those who underwent osmotic BBB opening. Four of the 74 patients in this study died within 30 days of the procedure, all attributed to infection, with three deaths occurring before the administration of granulocyte colony-stimulating factor. In one case study, expressive aphasia developed in a patient after osmotic BBB disruption but later resolved.²¹⁰ Vasospasm is another problem associated with osmotic BBB disruption²¹¹ that can lead to ischemic stroke. Higher relative permeability of the BBB to viruses and proteins also occurs during osmotic BBB disruption.^212,213 Exposure of nervous tissue to protein, specifically serum albumin, may cause astrocyte activation and seizures,²¹⁴ although others argue that osmotic BBB disruption does not allow extravasation of serum albumin.²¹⁵ Finally, osmotic BBB disruption is a costly and invasive procedure that requires highly trained practitioners to coordinate the treatment effectively.

Opening of Endothelial Tight Junctions—Parasympathetic Stimulation

Although the sympathetic division of the autonomic nervous system is responsible for maintaining and altering vasomotor tone in the body and thus peripheral blood flow, it has little if any effect on cerebral blood flow.²¹⁶ This discovery prompted researchers to evaluate alternative pathways for cerebrovascular control. Interestingly, sensory innervation to the circle of Willis travels via the nasociliary nerve, a branch of the ophthalmic nerve, after leaving the cranial cavity through the ethmoidal foramen.²¹⁷ Previous studies have shown that substance P–containing pain fibers may also dilate blood vessels via separate branches ending in motor terminals.²¹⁸

Parasympathetic input to the cerebral vasculature was evaluated by Suzuki and associates.^219,220 In this study, the nasociliary nerve was cut to interrupt sensory stimulation, and then the parasympathetic fibers entering the ethmoidal foramen were stimulated, which resulted in a 17% increase in cerebral blood flow. Sectioning the fibers did not affect blood flow, so parasympathetic stimulation also did not contribute to resting levels of vasomotor tone. Moreover, anticholinergic agents such as atropine and scopolamine did not attenuate the increase in cerebral blood flow, thus suggesting that the active neurotransmitter was not acetylcholine. The neurotransmitter mediating this effect was later shown to be NO.^221–223 In another study, stimulation of the rat sphenopalatine ganglion (SPG) increased cerebral blood flow by up to 50% in the ipsilateral parietal cortex.^224–227

Studies by Mayhan^228,229 showed that both NO donors and histamine increase permeability of the BBB. Yarnitsky and coworkers²³⁰ subsequently hypothesized that stimulation of parasympathetic fibers may increase the permeability of the BBB because parasympathetic fibers release NO. They evaluated the permeability of the BBB by using fluorescein isothiocyanate–labeled dextran in rat parietal cortex exposed by craniotomy. Evans blue–labeled albumin and two chemotherapeutic agents, anti-HER2 monoclonal antibody used for the treatment of breast cancer and etoposide, were used in closed-cranium experiments to evaluate permeability of the BBB as well. All these agents entered through the BBB in significant amounts throughout the ipsilateral brain after stimulation of the postganglionic parasympathetic fibers in the SPG. Some of the contralateral brain was affected as well. Smaller molecules penetrated the BBB more easily than larger ones, thus suggesting that the mechanism of this effect works by opening TJs rather than vesicular transport. No injurious effect on the brain, as measured by nicotinamide adenine dinucleotide (NAD)/reduced NAD (NADH) balance, was found, and no significant brain edema occurred. In another experiment by Yarnitsky and colleagues,²³¹ SPG stimulation in dogs showed similar effects on the BBB and was also observed to increase BBB permeability of the optic nerve, an extension of the CNS. Unlike the experiments in rodents, however, the effect in dogs and domestic pigs was entirely ipsilateral and only the anterior circulation was affected because the posterior vasculature is innervated by branches from the otic ganglion in these animals. Such findings suggest that SPG stimulation in humans would have similar effects on the ipsilateral anterior cerebral circulation.

Convection-Enhanced Delivery

Chemotherapy is infused into the brain tumor under constant pressure to deliver drug by bulk flow through catheters placed into the tumor bed.²³² Image guidance is used to optimize the accuracy of drug delivery. There are numerous variables that determine whether effective doses of drugs have been delivered to the tumor: the rate of infusion and total volume infused, the molecular weight of the drug agent, the affinity for and concentration in and around the target receptor, and finally, the density of brain tissue.²³³ Technical challenges of convection-enhanced delivery include injury to the brain during insertion of the catheter, backflow of infusate, drug infused into the EC space of the CNS, drug efflux by transport mechanisms related to the BBB, premature metabolism of the infusate, and finally, interstitial pressure gradients preventing adequate concentration of drug to remain in the tumor bed.²³⁴

Local Delivery of Polymer-Infused Chemotherapy

Incorporation of chemotherapy into biodegradable polymers was first developed in 1991²³⁵ and approved for used in 1996 in patients with malignant gliomas. A craniotomy must be performed to implant the drug wafers into the tumor resection cavity. There is variability in the amount of total drug released into the resection cavity, as well as nonuniform release of carmustine into a tumor. This approach relies on passive diffusion of drugs into the brain, unlike convection-enhanced delivery, which forces fluid drugs through the brain.

Targeted Toxin Therapy

This drug delivery system relies on a fusion protein consisting of a protein with affinity for the IL-4 receptor, which is upregulated on tumor cells, and a bacterial toxin that causes apoptosis by inhibiting translation.²³⁶

Trojan Horse Liposome

An appropriately named therapy, immunoliposomes carry a gene for protein correction gene therapy or small hairpin RNA expressed in plasma to inhibit protein synthesis by RNA interference (RNAi). The molecular Trojan Horse liposome is conjugated to a polyethylene glycol moiety, which increases the circulating half-life of the fusion construct. Polyethylene glycol is a peptidomimetic that binds to BBB transport receptors such as insulin or transferrin to enter the brain parenchyma. Preclinical studies have shown this technique to be a means of gene therapy for treating Parkinson’s disease.²³⁷ The most widely characterized receptor-mediated transcytosis system for targeting of drugs to the brain is the transferrin receptor, which mediates cellular uptake of iron bound to transferrin. Drugs targeting the transferrin receptor can be developed either by using an endogenous ligand, transferrin, or by using an antibody directed against the transferrin receptor. The insulin receptor has also been used for targeted delivery of drugs to the brain.

Focused Ultrasound Disruption of the Blood-Brain Barrier

This method uses ultrasound to create lesions in the BBB to increase its permeability. It is monitored closely by MRI thermometry.²³⁸ Vascular penetration was generated without measurable tissue injury by using intravenous contrast dye to visualize disruption of the BBB.²³⁹ Imaging revealed expanded TJs and increased migration across ECs.²⁴⁰ Preclinical studies with doxorubicin showed increased drug in sites targeted by ultrasound and retained binding of antibodies attached to a chemotherapeutic agent for their ligand.²⁴¹

S-100β: A Peripheral Marker of Blood-Brain Barrier Damage

Opening of the BBB provides molecules normally present in blood open passage into the CNS. However, this opening, unless involving specific transporters, does not work in just one direction. Proteins normally present in blood are free to diffuse into the CNS, and in turn, proteins normally present in high concentration in the CNS are free to diffuse down concentration gradients into blood. These peripheral BBB markers can be detected in blood to evaluate the permeability characteristics of the BBB at any given time. In a recent review article, Marchi and associates²⁴⁶ discussed the ideal properties of a peripheral marker of BBB disruption. Such proteins should have low or undetectable plasma levels in normal subjects, be normally present in CSF, and have a higher normal concentration in CSF than in plasma. Additionally, the CSF concentration of the protein should increase in response to insults. The protein should be normally blocked by the BBB and exhibit flux across the BBB during barrier disruption. Several proteins, including S-100β, neuron-specific enolase, and GFAP, have been evaluated for this purpose, but only S-100β meets the characteristics of having very low plasma levels with a concentration less than that found in CSF in normal subjects.

Moore discovered the S-100 protein family in 1965 by isolating a fraction of subcellular material containing proteins from bovine brain.²⁴⁷ Moore named the fraction S-100 because its contents were soluble in 100% saturated ammonium sulfate at neutral pH. The S-100 family contains more than 15 different calcium-binding proteins, including S-100β. All of them contain EF-hand calcium-binding domains and are approximately 10,000 daltons in size,²⁴⁷ significantly larger than the 400- to 500-dalton limit for passage through the BBB. Of the proteins in this family, S-100β is unique in its predominant location in the CNS^248–251 and is specifically located primarily in both astrocyte end-foot processes and Schwann cells.^252–254 Although S-100β is found in several other body tissues, its concentrations in these peripheral locations are significantly lower than in the CNS.^248,249 The primary role of S-100β remains to be elucidated, but it interacts with a number of cytoplasmic proteins with calcium-dependent actions and thus exerts a variety of influences on cells. It may exert its effects via a cyclic adenosine monophosphate–related mechanism.²⁵⁵

Plasma levels of S-100β are normally a third of those found in CSF and are nearly undetectable.²⁵⁶ Several diseases cause an elevation in plasma levels of S-100β, which can be detected and used for both diagnostic and prognostic purposes, as well as for evaluation of disease progression. Plasma S-100β levels increase with cerebral ischemia, with peak levels occurring approximately 3 days after infarction.^257–260 These levels have served as a useful marker of both infarct size and long-term clinical outcome.^{257–259,261,262} Traumatic brain injury has also been shown to increase S-100β levels in plasma,^263,264 with a positive correlation between the extent of damage after head injury and elevation in plasma S-100β.²⁶⁵ Ingebrigtsen and coworkers^266,267 found a negative predictive value of 0.99 for detecting intracranial pathology via serum S-100β levels versus CT studies. The highest S-100β levels in one study of traumatic brain injury were observed in samples taken approximately 2.5 hours after trauma, which is a considerably shorter period than that required for the maximal peak in plasma concentrations during ischemic stroke.²⁶⁸ A positive correlation was also found between early rises in blood plasma levels, up to 5 hours after trauma, and unfavorable outcome.^{262,265,267–270}

Plasma levels of S-100β have also been elevated in patients with hemorrhagic shock,²⁶³ aneurysmal subarachnoid hemorrhage,²⁷¹ hypoxia secondary to cardiac arrest,²⁷² and brain damage subsequent to attempted cardiopulmonary resuscitation.²⁷³ In all these cases, the elevation in S-100β plasma concentration is thought to result from damage to nervous tissue, including both neurons and glial cells. Additionally, depleted ATP levels in the brain during ischemic, hypoxic, and traumatic injury lead to increased adenosine levels,^274–276 which may activate A₁ adenosine receptors and cause release of S-100β from astrocytes. In experiments on cultured astrocytes, Ciccarelli and colleagues²⁷⁷ observed that adenosine receptor agonists cause the released S-100β to be nearly 160% that of controls. Plasma S-100β has also been elevated during cardiothoracic surgery,^278–280 although these findings remain debatable.²⁸¹ Currently, plasma levels of S-100β are predominantly used clinically for the monitoring of melanoma because these malignant cells express the protein. High plasma levels of S-100β correlate with disease progression.²⁸²

With the exception of melanoma, all the aforementioned disease processes involve some degree of brain damage. Diseases of the CNS that are associated with opening of the BBB do not necessarily cause brain damage or increased adenosine levels, so S-100β plasma levels must rise in response to BBB disruption without coexisting nervous tissue damage. To evaluate this, plasma S-100β levels were determined in patients with primary CNS lymphoma treated by chemotherapy, with intra-arterial infusion of mannitol used for osmotic BBB disruption. Plasma S-100β concentrations increased with mannitol administration. Intra-arterial administration of methotrexate once again caused plasma levels to rise, although patients receiving methotrexate without BBB disruption had no change in plasma S-100β. This elevation in peripheral S-100β levels occurred nearly immediately after BBB disruption, which excludes protein synthesis as a source for the increased S-100β.²⁸³ Intra-arterial infusion of methotrexate with BBB disruption does not appear to damage nervous tissue,²⁸⁴ thus suggesting that the elevated plasma concentrations reflect release of baseline levels of S-100β protein in the CNS and not release as a result of damaged brain tissue. Further studies have also shown that peripheral S-100β levels rise even in the absence of brain damage.^246,285

Three hypotheses have been proposed for how the rises in peripheral S-100β occur. CNS concentrations of the protein may increase first because of neuronal damage, with plasma levels rising after subsequent opening of the BBB. Alternatively, the BBB may open first, with subsequent neuronal damage elevating plasma levels. Finally, the rise in peripheral S-100β levels may be due to release of the normal amount of the protein in the CNS after opening of the BBB. This last hypothesis of plasma S-100β elevation suggests that it is a useful specific marker of BBB permeability. To evaluate the diagnostic utility of S-100β levels, Kanner and associates²⁸⁵ conducted a prospective study by determining S-100β levels in 51 patients undergoing MRI with and without gadolinium contrast enhancement for diagnostic and volumetric purposes. Normal MRI findings were seen in six patients from this group, and their plasma S-100β levels were found to be close to normal as well. In two patients treated for trigeminal neuralgia, one had a normal MRI study, whereas the other showed lacunar infarcts and related white matter changes. Plasma S-100β levels were elevated in this second patient, probably because of ischemia from the infarcts. All the remaining patients had MRI studies that demonstrated gadolinium enhancement and had significantly elevated basal S-100β levels. No significant differences in S-100β levels were found in patients with metastatic tumors of various origins. Moreover, there was no correlation between tumor size and rise in plasma S-100β concentrations. Primary CNS lymphoma was the only tumor evaluated that did not cause an elevation in S-100β. Immunocytochemical studies showed that this tumor had a uniform cytology devoid of normal astrocytic markers, including GFAP and S-100β, thus explaining the absence of plasma S-100β elevations in these patients.

Based on these studies on the use of peripheral detection of S-100β, Janigro and Marchi published methods to determine plasma levels of S-100β for use in the diagnosis of new conditions, determining the prognosis and progression of various disorders, and providing insight into the permeability characteristics of the BBB for proper drug administration.²⁸⁶ Research is currently under way to ensure that peripheral S-100β levels correlate well with gadolinium enhancement on MRI. If so, plasma S-100β could be used to screen patients for MRI for the diagnosis of brain tumors. Because BBB disruption can occur transiently for a variety of reasons, elevated plasma S-100β would not necessarily mean that a lesion was present. However, a normal plasma level would suggest that MRI would show no gadolinium enhancement, thus making MRI unnecessary. A simple and inexpensive blood test could be used regularly in place of expensive and time-consuming MRI studies.

Peripheral detection of S-100β also allows unique drug delivery opportunities. Design of a point-of-care system for evaluation of S-100β plasma levels would provide rapid determination of BBB permeability status and allow physicians to administer drugs during periods of some CNS diseases when the BBB is compromised. Moreover, some chronic disorders such as multiple sclerosis may result in alterations in permeability of the BBB that could be detected by measuring plasma S-100β and used for drug administration. Many more potential uses of S-100β measurement may develop as further research is performed on this peripheral marker of BBB disruption.

Blood-Brain Barrier and Neurological Disorders: Epilepsy

A seizure is a paroxysmal event caused by abnormal, excessive, hypersynchronous discharges from an aggregate of CNS neurons. This abnormal electrical activity can be reflected by various clinical/behavioral manifestations ranging from dramatic convulsive activity to experiential phenomena not readily discernible by an observer. Although a variety of factors influence the incidence of seizures, approximately 5% to 10% of the population will have at least one seizure during their lifetime, with the highest incidence occurring in early childhood and late adulthood.

Patients with epilepsy have seizures intermittently, and depending on the underlying cause, many patients are completely seizure free for months or even years. This sporadic appearance of seizures implies that precipitating factors induce seizures in patients with epilepsy. Numerous groups have described a number of vascular/blood-related factors that may be tipping the fragile epileptic brain toward seizures.^{214,287–294} Seizures are a result of a shift in the normal balance of excitation and inhibition within the CNS. Given the numerous properties that control neuronal activity, it is not surprising that there are many different ways to perturb homeostasis and therefore precipitate seizures. One of the main determinants of neuronal firing rate and synchronicity is the extracellular potassium level. Potassium controls glial and neuronal resting potential, repolarization, ion channel conductance, cerebral blood flow, and Na⁺/K⁺ pump activity. The complexity of CNS potassium homeostasis underscores its importance in the mammalian brain and also the role of the BBB. The process involves different cell types (neurons, glia, and ECs), several extracellular mechanisms (spatial buffering, cerebral blood flow), and strictly controlled segregation of potassium concentrations between blood (4.0 to 5.0 mM) and brain parenchyma (2.5 to 3.0 mM). Other molecular elements that may either participate in seizure onset or decrease the seizure threshold are brain levels of albumin, antibodies, or drugs.

Systemic pathologies causing BBB failure may also include hypertension, stroke, blood hyperosmolarity, or systemically mediated inflammatory processes (secondary to the production of TNF-α, IL-1β, IL-6, histamine, arachidonic acid, or reactive oxygen species). Furthermore, the normal brain is capable of seizing under the appropriate circumstances if the BBB is transiently breached. Finally, opening of the BBB lowers the threshold for seizures. All these underlying endogenous factors may influence the seizure threshold. Whether these factors are the same in a chronically epileptic patient as those involved in the generation of acute seizures (e.g., in “normal” brain) remains unknown.

Seizures and epilepsy are commonly observed in conjunction with stroke, traumatic brain injury, and CNS infections—all conditions known to compromise BBB function. It remains debatable whether the compromised integrity of the BBB is a factor involved in the etiology of epilepsy or secondary to such pathologies. The etiologic role that the BBB plays in seizures is supported by the fact that BBB disruption after acute head trauma is a well-known pathologic finding in both animal and human studies of S-100β.^294–297 BBB disruption may persist for weeks to years after the injury and may colocalize with abnormal electroencephalographic (EEG) activity. The increased interest in osmotic opening of the BBB as a viable mechanism of increasing drug delivery to the brain provides an opportunity to explore the connection between BBB disruption and seizures in a controlled clinical environment. The marked increase in BBB permeability to intravascular substances (10- to 100-fold for small molecules) after this osmotic disruption procedure is due to both increased diffusion and bulk fluid flow across the TJs. The permeability effect is largely reversed within minutes.²⁹⁸ In rodents, loss of BBB integrity by intra-arterial administration of hyperosmotic mannitol has been shown to rapidly lead to EEG changes consistent with epileptic seizures (spike/wave complexes interspersed with decreased EEG voltage that persist for several hours after the BBB disruption event).^290,299

Given these findings, it is not surprising that seizures are the primary complication of osmotic BBB disruption. Indeed, seizures occur in a relatively large number of these patients (13% to 55%). This high incidence was initially attributed to meglumine iothalamate, a known epileptogenic agent used as a contrast agent for CT.³⁰⁰ However, seizures associated with BBB disruption continued to occur (albeit with decreased frequency) when the disruption was monitored by radionuclide scanning rather than CT. Current research is focused on attempting to establish a correlation between the level of BBB disruption and the probability of a seizure occurring.^290,301