SCIENCE WATCH J Am Soc Nephrol 11: 1542–1547, 2000

Gene Therapy for Lysosomal Storage Disorders with Neuropathology

YIANNIS A. IOANNOU Departments of Human Genetics, Gene Therapy and Molecular Medicine, The Mount Sinai School of Medicine, New York, New York.

Lysosomes are acidic cellular organelles that function as terminal veloping enzyme or gene therapy protocols. Thus, replacing 1 to degradative compartments (1). Biochemically, these organelles 5% of the circulating enzyme in these disorders may not be are rich in lysosomal , negative for mannose-6- sufficient to revert the disease phenotype to that seen in the milder phosphate receptors (1), and acidified through the actions of a forms. v-type proton ATPase (2). More than 40 acidic hydrolases reside in the lumen of this organelle and catalyze the stepwise degrada- Lysosomal Storage Diseases as Therapeutic tion of complex carbohydrate, protein, and lipid substrates (3). Models Lysosomal storage diseases (LSD) are a group of approxi- For more than three decades, these disorders have been mately three dozen heterogeneous human disorders characterized models for the development of therapeutic endeavors for in- by the accumulation of undigested macromolecules within the herited metabolic diseases. In the early 1970s, it was shown lysosomes (Table 1), resulting in an increase in the size and that low levels of the proper normal enzyme could correct the number of these organelles (4). In 1965, Hers (5) developed the metabolic defect in cultured fibroblasts from affected patients concept of LSD to explain the relationship between ␣-glucosidase with various types of LSD (8–10). This observation fascinated and Pompe disease. Initially, it was ambiguous whether LSD investigators, who have since pursued enzyme replacement as arose by lack of degradation or by increased synthesis of the a means of treating these disorders. For example, clinical trials accumulated substrates; however, experimentally induced lysoso- of enzyme replacement in patients with type 1 Gaucher disease mal storage of compounds such as sucrose and dextran seemed to have confirmed the effectiveness of this approach in reversing suggest the former alternative (6). LSD are generally classified by the reticuloendothelial cell pathology in affected individuals the accumulated substrate (7), and they include , (11). These long-term clinical trials provide dramatic evidence glycoproteinoses, mucolipidoses, mucopolysaccharidoses, and of the clinical effectiveness of enzyme replacement for a non- others. It is now known that the LSD result from a deficiency of neuronopathic lysosomal disease. However, enzyme replace- a specific lysosomal enzyme or protein; however, it is still unclear ment trials in patients with the neuronopathic forms of Gaucher how storage of accumulated substrates relates to the pathology disease (types 2 and 3) have been unsuccessful, attesting to the seen in many of these disorders. inability of an intravenously administered enzyme to cross the With a few exceptions, these disorders lead to a severe neuro- blood–brain barrier (BBB) (12). degenerative phenotype (Table 1), which further complicates po- Bone marrow transplantation (BMT) studies (13) yielded tential treatment modalities, including gene therapy. In some analogous results. Most BMT recipients do not experience diseases (e.g., Gaucher, Niemann-Pick, Tay-Sachs, and Schin- neurologic or intellectual improvement, and most have a pro- dler), milder or late-onset subtypes have been identified, owing to longed clinical course before experiencing a neurologic de- the presence of residual enzymatic activity. These subtypes and mise. These studies reflect the present limitations of BMT for variants demonstrate that even low levels (1 to 5% of normal) of the treatment of LSD with significant neurologic involvement enzyme activity can alter a severe neurodegenerative disease and emphasize the need to develop novel strategies to treat course to a milder, often non-neurologic phenotype. It should be LSD whose primary site of pathology is the central nervous noted, however, that these low residual activities are present in all system (CNS). cells or at least in those that require these enzymes. This is an Clearly, the LSD have been invaluable paradigms for the important point that must be taken into consideration when de- development of novel therapeutic strategies such as enzyme replacement therapy (ERT) and will continue to serve as ex- cellent model systems in which to develop and evaluate new Received April 17, 2000. Accepted May 1, 2000. disease treatments. Correspondence to Dr. Yiannis A. Ioannou, Department of Human Genetics, Gene Therapy, and Molecular Medicine, One Gustave L. Levy Place, Box Lessons from Enzyme Replacement in Animal 1498, New York NY 10029-6574. Phone: 212-659-6720; Fax: 212-348-3605; E-mail: [email protected] Models 1046-6673/1108-1542 ERT for several LSD has been reported using a variety of Journal of the American Society of Nephrology animal models. For example, MPS type VII mice receiving Copyright © 2000 by the American Society of Nephrology single or multiple injections of ␤-glucuronidase exhibit attrib- J Am Soc Nephrol 11: 1542–1547, 2000 Lysosomes and Lysosomal Storage Diseases 1543

Table 1. The lysosomal storage disordersa

Glycogen storage type II ␣-Glucosidase Mucopolysaccharidoses (MPS) I Hurler syndrome ␣-L-Iduronidase II Hunter syndrome Iduronate sulfatase III Sanfilippo syndrome Heparan N-sulfatase IV Morquio syndrome Galactose 6-sulfatase VI Maroteaux-Lamy syndrome Arylsulfatase B VII Sly syndrome ␤-Glucuronidase II and III Phosphotransferase ␣-N-Acetylgalactosaminidase degradation ␣- and ␤- carbohydrate-deficient glycoprotein syndrome Wolman and cholesterol ester storage disease Acid lipase Ceramidase Niemann-Pick disease type A Sphingomyelinase type B Sphingomyelinase Niemann-Pick C1 NPC1 Niemann-Pick C2 ?? Gaucher disease types I, II, and III ␤-Glucosidase ␣-Galactosidase A Multiple sulfatase deficiency Sulfatases GM1 and Morquio B disease ␤-Galactosidase Protective protein GM2 gangliosidosis (Tay-Sachs and Sandhoff diseases) Cystinosis Cysteine transporter Sialic acid storage disease Sialic acid transporter Pyknodysostosis Cathepsin K Metachromatic Galactose-3-sulfatase Galactosialidosis , ␤-galactosidase, protective protein Neuronal ceroid lipofuscinosis (infantile) Palmitoyl protein thioesterase Neuronal ceroid lipofuscinosis (late infantile) Carboxypeptidase Cobalamin deficiency type F Cobalamin transporter

a Diseases shown in boldface exhibit neurological pathology. utable substrate reduction in and spleen, whereas reduc- tients. For example, some patients with glycogenosis type II tion in heart and kidney is seen only after multiple injections administered human or Aspergillus niger ␣-glucosidase show (14). MPS type I dogs receiving multiple injections of ␣-L- decreases in glycogen accumulation in liver (19). Also, purified iduronidase show substantial decreases in storage in liver, human acid ␤-glucosidase is capable of reducing storage in liver, spleen, and kidney, but none in heart (15,16). At low doses, red blood cells, lymphocytes, and platelets in patients with Gau- only liver Kupffer cells are cleared, whereas at high doses, both cher disease (20). And most importantly, urinary con- Kupffer cells and hepatocytes are cleared (16). MPS type VI centration is significantly decreased in a patient with Sandhoff cats administered multiple injections of N-acetylgalac- disease following infusion of plasma, suggesting a catabolism of tosamine-4-sulfatase show reversal of lysosomal accumulation renal by active (21). Finally, as men- in liver and heart (17), but heart depletion is dependent on dose tioned above, the most successful example of ERT is that for (18). Gaucher disease type I (22). The administration of recombinant Early studies of ERT in humans with LSD also provide evi- ␤-glucosidase in this disease has been an effective mode of dence of catabolism of accumulated substrates in tissues of pa- treatment for almost 10 years (reviewed in reference (23). 1544 Journal of the American Society of Nephrology J Am Soc Nephrol 11: 1542–1547, 2000

The results of these studies, albeit mixed, provide important ery of chemotherapeutic agents to combat brain tumors (28), clues regarding the proper dose and mode of delivery of a delivery of gene packages has not been attempted until now. therapeutic protein. Clearly, replacing a deficient lysosomal We have made a number of assumptions to devise a scheme enzyme requires a large dose of recombinant protein; this has for effective delivery of genes to the CNS. First, polycationic major implications for gene replacement endeavors. polypeptides that can complex and compact the plasmid DNA are available. Second, these compacted DNA packages can be targeted via specific cell ligands for plasma membrane binding Gene Therapy for LSD and initial internalization. Third, endocytosed DNA packages Because ERT is clearly limited to disorders without CNS can be designed to bypass the endosomal/lysosomal system. involvement, other modalities for treatment must be explored Fourth, nuclear localization signals in these packages will for diseases with neural involvement. One of the most prom- ensure efficient nuclear localization. We propose that all of the ising approaches involves gene delivery to the brain. However, above prerequisites can be met by the use of in vivo selected the brain presents unique physical and logistical barriers to peptide sequences fused to the small, DNA-compacting protein most, if not all, gene therapy approaches. Also, it should be protamine. The following discussion describes our approach to emphasized that the neuropathology in LSD is spread through- formulating these DNA packages, encoding the Aequorea vic- out the entire brain and is not limited to specific areas, thus toria green fluorescence protein (GFP), which can cross the requiring global neural gene delivery for effective treatment. BBB and transduce neural cells. Thus, the development of novel strategies to deliver genes to neurons and other neural cells throughout the CNS is necessary DNA Compaction to achieve the continuous expression of the proper therapeutic An important consideration, for crossing the BBB in partic- enzyme and correction of the metabolic defect. The main ular and for any nonviral gene delivery vehicle in general, is obstacles to this goal are the difficulty in delivering genes to the size of the DNA complex. Recent work on poly-L-lysine the CNS, and, within the CNS, the limited accessibility of (PLL) as a condensing moiety for DNA has identified condi- neurons, the main site of neuropathology in the LSD, to the tions by which the size of the PLL-DNA complex can be therapeutic gene. To overcome these obstacles, novel “global controlled and reduced to approximately 20 nm, compatible neural delivery” and “neural cell targeting” strategies must be not only with endocytic uptake (limit approximately 100 nm), developed. but also for diffusion through various tissues in vivo (see Early efforts to solve the above-mentioned problems in- below). Furthermore, even smaller DNA particles can be pro- volved using viral vectors to deliver genes directly into the cured by using PLL of lower molecular weight (approximately CNS and were of limited success. For example, herpes virus 3.5 kD), which has the added advantage of reduced toxicity. It vectors have been used to express the lysosomal enzyme ␤-glu- has also been shown that derivatization of PLL with appropri- curonidase in the brains of mice following stereotactic injec- ate glycans improves cellular transfection and transduction, tion. However, very few cells were transduced and they re- presumably attributable to a more efficient uptake of the com- mained clustered near the injection tract (24). Furthermore, plex (29) and/or decreased toxicity. Finally, PLL is easily issues of cytotoxicity remain a problem with herpes virus derivatized biochemically and thus can provide extremely use- vectors. Adenovirus vectors have also been stereotactically ful experimental data relevant to the development of the con- introduced into the brain and have been used to express various jugate artificial vector that will eventually be obtained by marker proteins. However, analogous to the results obtained recombinant DNA techniques. with the herpes virus vectors, the percentage of cells that was Alternatively, salmon protamine (SP), a protein found nat- transduced in vivo was very low and the cells did not appear to urally complexed with DNA at high concentrations in sperma- migrate significantly from the site of injection (reviewed in tozoa, can be used. This small 4-kD arginine-rich cationic reference (25). In addition, questions remain regarding the protein can be engineered as a fusion protein to any cell- persistence of expression in various neural cells using viral specific peptide (see below), resulting in a cell-targeted chi- vectors. Alternatively, nonviral delivery systems have been meric protein that can bind to the expression construct DNA. used to express genes in the brains of animals, but these also To obtain complexes between pGL, a 5-kb plasmid encoding have had limited success (reviewed in reference (26). Thus, GFP driven by the cytomegalovirus promoter, and either of the little progress has been made regarding the in vivo delivery of DNA-compacting proteins, PLL (approximately 22.5 kD) or genes to the brain and the proper expression of these genes in SP (approximately 4 kD), we used a published procedure with neural cells. Moreover, efforts to deliver genes to the entire appropriate modifications (30). By tightly controlling the brain (i.e., global CNS delivery) have lagged behind, mostly DNA:protein ratio, reaction volume, timing, order of addition, because of the presence of the BBB and the failure of intrave- and salt concentration, we have consistently obtained appar- nously administered gene delivery vehicles to cross this bar- ently torroidal PLL-DNA and SP-DNA particles of approxi- rier. An approach with a high potential for achieving global mately 10 to 20 nm in diameter, as assessed by transmission neural gene delivery involves the transient disruption of the electron microscopy (TEM) (Figure 1). Preliminary TEM ev- BBB and injection of nonviral DNA delivery vehicles. Al- idence suggests that by using PLL 3.5 kD, the diameter of the though the transient disruption of the BBB has been carried out torroidal complexes can be further reduced to Ͻ10 nm, which successfully both in animals (27) and in humans for the deliv- should facilitate intercellular movement. J Am Soc Nephrol 11: 1542–1547, 2000 Lysosomes and Lysosomal Storage Diseases 1545

Figure 1. Electron microscopy of compacted DNA. Transmission electron micrographs of poly-L-lysine-DNA (A) and salmon protamine-DNA (B) complexes. Apparently torroidal particles 10 to 20 nm in diameter are obtained. Bar, 100 nm.

Cell-Specific Targeting and Endosomal Escape displayed peptides (Figure 2). The peptides identified by this The choice of targeting ligand depends on the cell type to be procedure (out of a pool of Ͼ109 particles) show a consensus targeted and would preferably be a peptide recognized by a sequence in 5 of 10 sequences. The proline at position 4 receptor (not necessarily a protein) that undergoes endocytosis. (Figure 2) suggests that these phage particles were endocytosed Although most endocytosed plasma membrane proteins/receptors and escaped endosomal degradation by an analogous mecha- enter the endosomal/lysosomal system, recent evidence suggests nism, probably via the same receptor system. This consensus, that an alternate nonendosomal uptake system is functional in hydrophobic-X-Pro-X-positively charged (Gln or Asn), may be most cells (31). Support for the existence of this uptake system has important in endocytosis and endosome escape. Furthermore, recently been generated through studies of the small DNA virus, analysis of peptides capable of binding to MHC class I recep- SV40, which apparently enters cells in an endosome-independent tors reveals a striking similarity between the isolated peptides manner involving noncoated, caveolae-like vesicles (32). In ad- and MHC binding peptides, which also contain proline and dition to SV40, several receptors are internalized via this system glycine in similar positions. (reviewed in reference (33), making it an attractive targeting Thus, in vivo screening of phage display libraries can lead to the pathway that can bypass the problems and limitations associated isolation of small peptides with novel and desirable cell targeting with entering clathrin-coated endosomes. and uptake characteristics. Such screening procedures can easily The recent availability of phage peptide display libraries be adapted to the isolation of peptides taken up by specific cell allows the rapid selection of small peptides with specific bind- types, such as neuronal cells, by an endosome-independent path- ing properties. Phage display libraries have been used in sev- way for delivery of DNA constructs to the nucleus. eral in vitro selection procedures to isolate ligands that bind to specific proteins (34) and recently in an in vivo system to Nuclear Localization isolate peptides that bind to specific mouse organs (35). Once the vector DNA complex is released into the cytosol of Using A431 cells, a commercially available, seven amino the targeted cells, the DNA must gain access to the cell nucleus. acid Escherichia coli phage display library was used to deter- mine whether small peptides that undergo internalization via an endosome-independent pathway could be isolated. Cells were plated in culture dishes and the phage library, representing about 5 ϫ 109 independent phage particles, was allowed to adsorb onto the cell monolayer. The rationale for this experi- mental setup was the following: Phage that were endocytosed via an endosome-dependent pathway would be partially or completely destroyed in the lysosomal system, whereas phage that were internalized by an endosome-independent pathway would be left intact and could be recovered by infecting fresh Escherichia coli. The cells were subsequently washed to re- move any phage particles that had not been internalized, lysed, and used to infect competent Escherichia coli cells to recover viable phage particles. Recovered phage were expanded en Figure 2. Peptides isolated by in vivo selection. A ClustalW alignment masse and used for another round of adsorption onto A431 of 10 peptides isolated by in vivo selection as described in the text. cells as above. This procedure was repeated three times to The conserved proline residues are boxed. A consensus sequence of reduce or eliminate background, and the phage recovered after hydrophobic-X-proline-charged amino acid can be clearly seen in the the third round were sequenced to determine the identity of the isolated peptide sequences. 1546 Journal of the American Society of Nephrology J Am Soc Nephrol 11: 1542–1547, 2000

Nuclear targeting can be achieved by incorporation of a nuclear localization signal into the DNA-containing complex. Such sig- nals have been well characterized for several mammalian and viral proteins and appear to be small, positively charged peptides of about 7 to 10 residues. Nuclear targeting is particularly impor- tant for cells that do not actively divide. However, use of PLL or protamine to formulate the DNA complex may obviate the need for a nuclear localization signal, because these polycations are already effectively targeted to the nucleus.

Global Neural Delivery A method to bypass the BBB is necessary to access the CNS in a global manner. The extensive vascularization of the brain pro- vides an attractive delivery route. However, the BBB prevents the passage of macromolecules from the circulation to the CNS parenchyma. An approach that can be used to deliver DNA packages to the brain is the transient opening of the BBB by the intracarotid injection of hyperosmolar solutions of mannitol or other agents (36). Electron microscopic observations have con- firmed that hyperosmolar solutions cause dehydration and shrink- age of vascular endothelial cells, thereby transiently relaxing the tight junctions and opening the intercellular spaces (reviewed in reference (37). This method provides the means to globally de- liver relatively large DNA packages (10 to 20 nm) to the brain. In humans, hyperosmolar BBB disruption has been applied to patients for the treatment of certain brain tumors (38). The therapeutic effectiveness of the procedure has been limited due to its cytotoxicity to normal cells (39), resulting from the high concentration of cytotoxic drugs delivered to the entire CNS parenchyma. These observations attest to the effectiveness of this strategy as a global CNS delivery system. Moreover, few complications have been reported in patients (40) or animals undergoing short-term experiments (41), thereby justifying its use to deliver nontoxic, therapeutic DNA vector constructs to the brain parenchyma. To test this CNS delivery approach, protein-DNA com- plexes were infused by intracarotid injection into rats after BBB disruption. Neural cell transduction by uptake of protein- DNA packages was judged by GFP expression in rats, which were sacrificed 5 d after treatment and examined by confocal microscopy. Numerous GFP-positive parenchymal cells were detected in the infused animals (Figure 3). In addition, positive neural cells can be seen expressing GFP, suggesting that the Figure 3. Confocal microscopy of rat brain sections. Micrograph of compacted DNA packages can be endocytosed by all neural rat brain section showing green fluorescence-positive cells (arrows) cells. Thus, the transient disruption of the BBB and infusion of due to expression of the green fluorescence protein (GFP) transgene in compacted DNA packages can accomplish the initial require- numerous parenchymal cells. (A) Phase contrast of GFP field shown ment of global neural delivery of therapeutic genes to the brain. in Panel B.

Conclusion In conclusion, ERT has not proved to be a viable mode for Acknowledgments treatment of LSD with neuropathology. Gene therapy ap- I acknowledge the contribution of many people in my laboratory, proaches using current viral vectors have also been ineffective first and foremost that of Dr. Mario Rattazzi, whose expertise in BBB due to the presence of the BBB. Based on our experimental disruption has made this work possible, and Ronald Gordon for data, delivery of nonviral DNA complexes to the brain after electron microscopy. In addition, I thank Compton Benjamin, Takashi transient disruption of the BBB appears to be the most prom- Tsuda, and Annette Enriquez for their hard work. Finally, I thank ising approach to global neural delivery of therapeutic genes. Fannie W. Chen for careful review of the manuscript. J Am Soc Nephrol 11: 1542–1547, 2000 Lysosomes and Lysosomal Storage Diseases 1547

References 21. Desnick RJ, Dawson G, Desnick SJ, Sweeley CC, Krivit W: 1. Kornfeld S: Trafficking of lysosomal enzymes. FASEB J 1: Diagnosis of glycosphingolipidoses by urinary-sediment analy- 462–468, 1987 sis. N Engl J Med 284: 739–744, 1971 2. Mellman I, Fuchs R, Helenius A: Acidification of the endocytic 22. Masek BJ, Sims KB, Bove CM, Korson MS, Short P, Norman and exocytic pathways. Ann Rev Biochem 55: 663–700, 1986 DK: Quality of life assessment in adults with type 1 Gaucher 3. Storrie B, Desjardins M: The biogenesis of lysosomes: Is it a kiss disease. Qual Life Res 8: 263–268, 1999 and run, continuous fusion and fission process? Bioessays 18: 895– 23. Brady RO: Gaucher’s disease: Past, present and future. Baillieres 903, 1996 Clin Haematol 10: 621–634, 1997 4. Gieselmann V: Lysosomal storage diseases. Biochim Biophys 24. Deshmane SL, Valyi-Nagy T, Block T: An HSV-1 containing the Acta 1270: 103–136, 1995 rat beta-glucuronidase cDNA inserted within the LAT gene is less 5. Hers HG: Inborn lysosomal diseases. Gastroenterology 48: 625– efficient than the parental strain at establishing a transcriptionally 633, 1965 active state during latency in neurons. Gene Ther 2: 209–217, 1995 6. Lloyd JB: Experimental support for the concept of lysosomal 25. Barkats M, Bilang-Bleuel A, Buc-Caron MH: Adenovirus in the storage disease. In: Lysosomes and Storage Diseases, edited by brain: Recent advances of gene therapy for neurodegenerative Hers HG, Van Hoof F, New York, Academic, 1973, pp 173–195 diseases. Prog Neurobiol 55: 333–341, 1998 7. Glew RH, Basu A, Prence EM, Remaley AT: Lysosomal storage 26. Weihl C, Macdonald RL, Stoodley M, Luders J, Lin G: Gene diseases. Lab Invest 53: 250–269, 1985 therapy for cerebrovascular disease. Neurosurgery 44: 239–253, 8. Brot FE, Glaser JH, Roozen KJ, Sly WS, Stahl PD: In vitro 1999 correction of deficient human fibroblasts by ␤-glucuronidase 27. Chi OZ, Lee DI, Liu X, Weiss HR: The effects of morphine on from different human sources. Biochem Biophys Res Commun blood-brain barrier disruption caused by intracarotid injection of 57: 1–8, 1974 hyperosmolar mannitol in rats. Anesth Analg 90: 603–608, 2000 9. Turner BM, Turner VS, Hirschhorn K: Metabolic correction of 28. Siegal T, Rubinstein R, Bokstein F: In vivo assessment of the fucosidosis fibroblasts by human alpha-L-fucosidase. J Cell window of barrier opening after osmotic blood-brain barrier Physiol 98: 225–235, 1979 disruption in humans. J Neurosurg 92: 599–605, 2000 10. Dawson G, Matalon R, Li YT: Correction of the enzymatic 29. Erbacher P, Roche AC, Monsigny M, Midoux P: The reduction defect in cultured fibroblasts from patients with Fabry’s disease: of positive charges of polylysine by partial gluconoylation in- Treatment with purified ␣-galactosidase from Ficin. Pediatr Res creases the transfection efficiency of polylysine/DNA com- 7: 694–698, 1973 plexes. Biochim Biophys Acta 1324: 27–36, 1997 11. Brady RO, Murray GJ, Barton NW: Modifying exogenous glu- 30. Perales JC, Ferkol T, Molas M, Hanson RW: An evaluation of cocerebrosidase for effective replacement therapy in Gaucher receptor-mediated gene transfer using synthetic DNA-ligand disease. J Inherit Metab Dis 17: 510–519, 1994 complexes. Eur J Biochem 226: 255–266, 1994 12. Zirzow GC, Sanchez OA, Murray GJ, Brady RO, Oldfield EH: 31. Anderson HA, Chen Y, Norkin LC: Bound simian virus 40 Delivery, distribution, and neuronal uptake of exogenous man- translocates to caveolin-enriched membrane domains, and its nose-terminal in the intact rat brain. Neuro- entry is inhibited by drugs that selectively disrupt caveolae. Mol chem Res 24: 301–305, 1999 Biol Cell 7: 1825–1834, 1996 13. Hoogerbrugge PM, Valerio D: Bone marrow transplantation and 32. Chen Y, Norkin LC: Extracellular simian virus 40 transmits a gene therapy for lysosomal storage diseases. Bone Marrow signal that promotes virus enclosure within caveolae. Exp Cell Transplant 21[Suppl 2]: S34–S36, 1998 Res 246: 83–90, 1999 14. Sands MS, Vogler C, Kyle JW: Enzyme replacement therapy for 33. Norkin LC: Simian virus 40 infection via MHC class I molecules murine type VII. J Clin Invest 93: 2324– and caveolae. Immunol Rev 168: 13–22, 1999 2331, 1994 34. Barbas CF: Recent advances in phage display. Curr Opin Bio- 15. Shull RM, Kakkis ED, McEntee MF, Kania SA, Jonas AJ, technol 4: 526–530, 1993 Neufeld EF: Enzyme replacement in a canine model of Hurler 35. Barry MA, Dower WJ, Johnston SA: Toward cell-targeting gene syndrome. Proc Natl Acad Sci USA 91: 12937–12941, 1994 therapy vectors: Selection of cell-binding peptides from random 16. Kakkis ED, McEntee MF, Schmidtchen A: Long-term and high- peptide-presenting phage libraries. Nat Med 2: 299–305, 1996 dose trials of enzyme replacement therapy in the canine model of 36. Rapoport SI: Blood–Brain Barrier in Physiology and Medicine, mucopolysaccharidosis I. Biochem Mol Med 58: 156–167, 1996 New York, Raven, 1976 17. Crawley AC, Brooks DA, Muller VJ: Enzyme replacement ther- 37. Kroll RA, Neuwelt EA: Outwitting the blood-brain barrier for apy in a feline model of Maroteaux-Lamy syndrome. J Clin therapeutic purposes: Osmotic opening and other means. Neuro- Invest 97: 1864–1873, 1996 surgery 42: 1083–1099, 1998 18. Crawley AC, Niedzielski KH, Isaac EL, Davey RCA, Byers S, 38. Neuwelt EA, Rapoport SI: Modification of the blood–brain Hopwood JJ: Enzyme replacement therapy from birth in a feline barrier in the chemotherapy of malignant brain tumors. Fed Proc model of mucopolysaccharidosis type VI. J Clin Invest 99: 651– 43: 214–219, 1984 662, 1997 39. Suhr ST, Gage FH: Gene therapy for neurologic disease. Arch 19. Huijing F, Waltuck BL, Whelan WJ: ␣-Glucosidase administra- Neurol 50: 1252–1268, 1993 tion: Experiences in two patients with glycogen storage disease 40. Gumerlock MK, Belshe BD, Madsen R, Watts C: Osmotic compared with animal experiments. In: Enzyme Therapy in Ge- blood-brain barrier disruption and chemotherapy in the treatment netic Diseases, edited by Desnick RJ, Bernlohr RW, Krivit W, of high grade malignant glioma: Patient series and literature Baltimore, Williams & Wilkins, 1973, pp 191–194 review. J Neurooncol 12: 33–46, 1992 20. Beutler E, Dale GL, Guinto E, Kuhl W: Enzyme replacement 41. Kajiwara K, Ito H, Fukumoto T: Lymphocyte infiltration into therapy in Gaucher’s disease: Preliminary clinical trial of a new normal rat brain after carotid infusions of hyperosmolar solu- enzyme preparation. Proc Natl Acad Sci USA 74: 4620–4623, 1977 tions. J Neuroimmunol 27: 233–240, 1990