A breach in the blood-brain barrier

A fundamental problem affecting the efficacy of drugs is their ability to be delivered from the site of administration to the site of pathology. Nowhere is this issue a more confounding problem than for protein drugs designed to treat the neuropathology associated with many lysosomal storage diseases (LSDs) where the blood-brain barrier (BBB) stands in the way. Although an increasing number of lysosomal storage diseases are amenable to treatment by enzyme replacement therapy (ERT), those with neurological consequences have remained challenging to treat. The report by Vogler et al. (1) in this issue of PNAS provides an encouraging demonstration that a therapeutic lysosomal enzyme can be delivered across the BBB, thereby reversing storage in brain tissue. Overcoming the BBB can be seen as the latest in a series of drug-targeting challenges that have faced researchers attempting to treat LSDs by delivering therapeutic enzymes to the lysosome of afflicted cells.

Lysosomal Storage Diseases and Enzyme Replacement Therapy

Lysosomal storage diseases are a group of >40 genetic disorders with heterogeneous symptoms (2). Each results from the deficiency of a lysosomal hydrolase that functions in the lysosome to degrade macromolecules, primarily glycoconjugates such as sphingolipids, glycosaminoglycans (GAG, formerly mucopolysaccharides), glycogen, or glycoproteins. The pathology associated with these diseases depends on the nature of the accumulated glyconjugate and the tissues in which the storage material accumulates. Many of the lysosomal storage diseases affect brain tissue.

ERT has become the treatment modality of choice for an increasing number of LSDs. ERT is a therapeutic strategy to correct an enzyme deficiency by infusing the missing enzyme into the bloodstream. As the blood perfuses patient tissues, an enzyme is taken up by cells and transported to the lysosome, where the enzyme acts to eliminate the storage material.

ERT was established as a therapeutic paradigm by the work of Roscoe Brady and colleagues at the National Institutes of Health, whose efforts to develop an ERT for Gaucher disease culminated in Food and Drug Administration (FDA) approval of the first Gaucher ERT in 1991 (3, 4). Gaucher disease is caused by the deficiency of the lysosomal enzyme glucocerebrosidase. Type I Gaucher patients accumulate large quantities of glucocerebroside in the lysosomes of macrophage, particularly in the liver and spleen but also in bone and other tissues.

Targeting Lysosomes

A key element in the successful denouement of the Gaucher program was the development of a successful strategy for targeting the enzyme to Kupffer cells, liver resident macrophages that were the prime sites of storage of glucocerebroside. Initial studies revealed that the majority of infused glucocerebrosidases were delivered to hepatocytes. Brady's group devised a successful glyco-remodeling strategy to target the enzyme to a macrophage receptor with high affinity for mannose-terminated carbohydrates by using exoglycosidases to expose terminal mannose residues on the carbohydrate (5).

All ERT therapeutics subsequently developed have relied on interaction of the carbohydrate on the surface of the enzyme with a cell-surface receptor to deliver the enzyme to the lysosome. However, in contrast to Gaucher disease, most of the other lysosomal storage diseases also exhibit storage defects in nonreticuloendothelial cells that lack the mannose receptor. Delivery of enzymes to these cell types has taken advantage of a novel carbohydrate modification, mannose-6-phosphate (M6P) first identified by William Sly and colleagues as the carbohydrate moiety that conferred “high-uptake” status to lysosomal enzymes (6) and later understood to be instrumental in the intracellular trafficking of most enzymes destined for the lysosome.

M6P is a high-affinity ligand for two M6P receptors found in mammalian cells (7). The cation-independent M6P receptor (CI-MPR) is present on the surface of most cell types and is capable of internalizing M6P-containing ligands resulting in lysosomal delivery. The need for M6P in the delivery of enzymes to nonreticuloendothelial cells was explicitly demonstrated in a murine mucopolysaccharidosis (MPS) VII model by comparing the distribution of M6P-containing enzyme with that of enzyme-lacking M6P (8).

In the years that have followed the introduction of the Gaucher drug, ERT therapeutics for Fabry disease and MPS I (marketed by Genzyme and jointly by BioMarin and Genzyme, respectively) were approved by the FDA in 2003, and a MPS VI ERT (marketed by BioMarin) was approved by the FDA in May 2005. A Biologic License Application (BLA) was filed by Genzyme for its Pompe therapeutic in July 2005, and a positive Phase III clinical trial outcome reported earlier this year by Transkaryotic Therapies (acquired by Shire) may be the basis for a similar filing for a MPS II therapeutic. Thus, ERT-based therapies have become available for a number of LSDs.

Crossing the BBB

Although ERT has proven to be an effective strategy to treat nonneurological symptoms of LSDs, treatment of the neurologic consequences of many LSDs requires the delivery of therapeutic proteins across the BBB. The BBB has been an insurmountable barrier. The BBB maintains a homeostatic environment in the central nervous system (9). It is formed by the endothelial cells of the brain's microvasculature. Unlike endothelial cells in other tissues, adjacent brain capillary cells have tight junctions that block passage of most molecules. Although low molecular weight lipid-soluble materials may diffuse across the BBB, water-soluble materials, such as glucose, and amino acids do not pass through. Mediated transport mechanisms exist to transport glucose and essential amino acids across the BBB. Active efflux systems pump out compounds that accumulate in the brain.

Similarly, most proteins do not traverse the BBB. However, mechanisms permitting proteins and peptides to traverse the BBB do exist. Some proteins traverse the BBB through a receptor-mediated process of transcytosis in which proteins are endocytosed at the endothelial cell surface facing the bloodstream and then exocytosed on the post-capillary side of the endothelial cell. Because the microvasculature of the brain is so densely distributed (with a mean intercapillary distance of 40 μm), proteins transcytosed across the BBB can access the entire brain tissue. Attempts have been made to use proteins capable of being transported across the BBB as carriers. When such carrier proteins or peptides are joined to a therapeutic protein, they may ferry the attached protein cargo across the BBB (9-11).

Other routes into the brain are possible. Direct injection of protein drugs into the cerebrospinal fluid (CSF) could enable the protein to contact many surfaces of the brain tissue. Delivery of protein drugs by this route is challenged by the requirement that material diffuses across the ependymal cell layer to penetrate the underlying parenchyma. Large concentration gradients are reported to exist between the CSF and tissue 2 mm beyond the ependymal layer (9). High levels of therapeutic protein would need to be present in the CSF to overcome this steep concentration. For example, Kakkis et al. (12) have shown that intrathecal injection of up to 4 mg of recombinant α-l-iduronidase into the CSF in a canine model of MPS I generated a sufficient concentration gradient to permit the recombinant protein to penetrate at least 3 mm deep into the brain paryenchma with concomitant reversal of storage in all cell types assayed.

Nevertheless, given the invasive nature of intrathecal injection and the superior access of the brain microvasculature to the brain tissue, direct i.v. delivery of therapeutic proteins across the BBB would be highly advantageous. So the demonstration by Vogler et al. (1) that clinically significant quantities of enzyme infused into the bloodstream can be delivered across the BBB in a murine MPS VII model is welcome news.

The murine MPS VII model has proven to be an excellent model for assessment of various therapeutic strategies for the treatment of LSDs from gene therapy to ERT. MPS VII results from the deficiency of β-glucuronidase, which causes an accumulation of GAGs in many tissues. MPS VII mice share with human patients many of the clinical manifestations of the disease, including cognitive, hearing, visual, and skeletal defects, as well as a shortened lifespan. GAGs are eliminated from many tissues after infusion of β-glucuronidase into MPS VII mice, but no reduction in storage in the brain occurs unless treatment is initiated while the mice are <2 weeks of age (13). One explanation for this data might be that the neonatal BBB is somewhat leaky. However, recent experiments have shown otherwise.

Urayama et al. (14) demonstrated that M6P-containing but not M6P-lacking β-glucuronidase could transit the mouse neonatal BBB. Competitors such as M6P effectively blocked the transport across the BBB. M6P-containing enzyme was unable to cross the BBB of mature mice. In contrast, albumin was equally ineffective in traversing the BBB of neonatal or mature mice. These data suggest that the M6P-containing enzyme was transcytosed across the BBB by a receptor for which M6P is a high-affinity ligand. This receptor is most likely the CI-M6P receptor, which is down-regulated in postnatal mice.

In this issue of PNAS, Vogler et al. (1) now demonstrate that it is possible to deliver M6P-labeled β-glucuronidase across the BBB of mature mice by using high doses of enzyme. Treatment regimens that included 13 weekly doses of 4 mg/kg or 4 weekly doses of 20 mg/kg were most effective in reversing GAG storage in brain tissue. Reversal of storage was most pronounced in perivascular meninges, glial cells, and parietal neocortical neurons.

Significant reversal in storage was detected despite the fact that the level of β-glucuronidase achieved in the brains of the test animals did not exceed 2.5% of wild-type levels. This finding is in accord with the notion that there is a threshold level of enzyme, significantly below wild-type levels, that is capable of managing the flux of storage material that accumulates in particular cells (15). The number may be >2.5% of wild type because it is not known whether the infused enzyme was uniformly delivered to brain tissue.

Given the ingenious strategies that have been deployed in previous attempts to deliver such proteins across the BBB, it is ironic that this simple method has proven effective. The most likely explanation for the delivery of this enzyme across the BBB is, as the authors point out, that the higher levels of infused enzyme generate a sufficient blood concentration to enable saturation of BBB CI-MPR receptor for a longer period. But other explanations are possible. Proof that the enzyme is delivered in an M6P-dependent manner could be had by comparing high-dose infusions of M6P-containing and -lacking enzyme into MPS VII mice. If the process were found to be CI-MPR dependent, it would be useful to follow the distribution in the brain of enzyme tagged with other ligands that bind to the CI-MPR receptor such as IGF-II (16).

It is too soon to say whether the observed reversal of storage in brain tissue described by Vogler et al. (1) will be sufficient to correct the neurological deficits caused by the disease. If so, one can hope that similar protocols will ultimately produce results in the clinic. There is certainly precedent for giving patients doses as high as those described in this study. For example, the pivotal Phase III clinical trials that are the basis for the filing by Genzyme of a BLA for its Pompe drug used dosages of 20-40 mg/kg. The extreme rarity of MPS VII patients might make a clinical trial for MPS VII difficult, although immeasurably valuable for those afflicted with MPS VII. Perhaps a high dosage trial of recombinant α-l-iduronidase might be conducted in patients with Hurler syndrome, a severe form of MPS I with neurologic symptoms that accompany the accumulation of GAGs in brain cells such as neuronal, glial, and menenges cells.

If the results reported by Vogler et al. (1) are demonstrated to be applicable to other lysosomal storage disorders, it will finally be possible to report that the BBB has been breached.