Effective Gene Therapy for an Inherited CNS Disease in a Large Animal Model Charles H. Vite, DVM, PhD,1,2 Joseph C. McGowan, PhD,3 Sumit N. Niogi, BS,1 Marco A. Passini, PhD,1,2 Kenneth J. Drobatz, DVM, MSCE1 Mark E. Haskins, VMD, PhD,1 and John H. Wolfe, VMD, PhD1,2 Genetic diseases affecting the brain typically have widespread lesions that require global correction. Lysosomal storage diseases are good candidates for central nervous system gene therapy, because active enzyme from genetically corrected cells can be secreted and taken up by surrounding diseased cells, and only small amounts of enzyme (<5% of normal) are required to reverse storage lesions. Injection of gene transfer vectors into multiple sites in the mouse brain has been shown to mediate widespread reversal of storage lesions in several disease models. To study a brain closer in size to the human brain, we evaluated the extent of storage correction mediated by a limited number of adeno-associated virus vector injections in the cat model of human ␣-mannosidosis. The treated cats showed remarkable improvements in clinical neurological signs and in brain myelination assessed by quantitative magnetic resonance imaging. Postmortem examination showed that storage lesions were greatly reduced throughout the brain, even though gene transfer was limited to the areas surrounding the injection tracks. The data demonstrate that widespread improvement of neuropathology in a large mammalian brain can be achieved using multiple injection sites during one operation and suggest that this could be an effective treatment for the central nervous system component of human lysosomal enzyme deficiencies. Ann Neurol 2005;57:355–364 A large number of single-gene disorders affect the central nervous system (CNS).1 Many of these diseases are caused by deficiencies in metabolic pathways, which affect cells throughout the brain. Somatic gene transfer has the potential to permanently correct the underlying metabolic deficiency by transferring a normal copy of a defective gene into a patient’s own cells. However, gene transfer is generally restricted to areas near injection sites. Although global correction of genetic metabolic disease by gene therapy can be achieved under certain circumstances in the 400mg mouse brain,2,3 it would be necessary to achieve a scale more than 1,000fold greater to treat the human infant brain. We hypothesized that by developing effective methods in the 30g cat brain, we would show the potential for therapeutic efficacy likely to be encountered in the human brain. The cat model offers a number of advantages for investigating translational approaches to somatic gene therapy of the CNS. First, the feline nervous system has been well characterized functionally, anatomically, and physiologically, and the physical organization of the cat brain is more similar to the human brain than is the murine brain.4,5 Second, many naturally occurring inherited metabolic disorders affect the cat brain.2,3,6,7 Third, the background genetic heterogeneity among relatively noninbred cats is similar to the genetic diversity of affected human populations, which is unlike the relative genetic homogeneity of mouse models. Fourth, adequate numbers of affected individuals and healthy siblings can be produced for matched treatment and control groups to achieve statistical power.8 –12 Fifth, because of improved resolution of brain structures, the large size of the cat brain permits the study of regional correction of neuropathology using noninvasive imaging methods.13–16 We evaluated gene therapy in brains of cats affected with ␣-mannosidosis (AMD), a lysosomal storage disease caused by a deficiency of lysosomal ␣-mannosidase (LAMAN) activity. Human AMD is characterized by predominantly progressive neurological dysfunction and skeletal abnormalities.17–20 The disease in the cat is caused by a 4bp deletion in the feline ␣-mannosidase gene (fMANB) and is characterized by clinical signs, neuropathology, and biochemical abnormalities similar to those seen in human patients.9,21–29 AMD cats were From the 1School of Veterinary Medicine, University of Pennsylvania; 2Children’s Hospital of Philadelphia, Philadelphia, PA; and 3 US Naval Academy, Annapolis, MD. This article includes supplementary materials available via the Internet at http://www.interscience.wiley.com/jpages/0364-5134/suppmat Received Aug 24, 2004, and in revised form Nov 30. Accepted for publication Dec 1, 2004. Current address for S. Niogi: School of Medicine, Cornell University, New York, NY 10021. Current address for Dr Passini: Genzyme Corporation, Framingham, MA 01701. Published online Feb 24, 2005, in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/ana.20392 Address correspondence to Dr Wolfe, 503 Abramson Research Center, Children’s Hospital of Philadelphia, 3516 Civic Center Blvd., Philadelphia, PA 19104-4318. E-mail: email@example.com © 2005 American Neurological Association Published by Wiley-Liss, Inc., through Wiley Subscription Services 355 injected intracerebrally at multiple sites with an adenoassociated virus (AAV) packaged in a serotype 1 vector (AAV1) expressing a wild-type copy of fMANB complementary DNA (cDNA). The treated cats expressed LAMAN activity throughout the brain, showed widespread correction of neuropathology, and demonstrated significantly diminished effects of the clinical neurological syndrome. Materials and Methods Animals Cats were raised under National Institutes of Health and US Department of Agriculture guidelines, and all studies were approved by the Institutional Animal Care and Use Committee. Affected cats were bred from carrier cats, and all were genotyped for the mutant allele.21 Physical and neurological examinations were performed weekly from birth up until the animals were killed. Vector Production The AAV vector genome contained AAV2 inverted terminal repeats (ITRs) flanking a transcription unit containing fMANB under the control of a cytomegalovirus promoter and bovine growth hormone poly A sequences. The vector plasmid was packaged into an AAV1 capsid, which had a titer of 6.4 ⫻ 1012 genome equivalents per milliliter. Intraparenchymal Injection of Viral Vectors At 8 weeks old, six cats homozygous for the mutant allele were anesthetized with intravenous propofol (up to 6mg/kg; Abbott Laboratories, Chicago, IL), endotracheally intubated, maintained under anesthesia with isoflurane (IsoVet; Schering-Plough, Omaha, NE), and placed in a nonferrous stereotaxic head holder designed by the investigators. A midline incision was made over the dorsum of the calvaria, and holes (1mm in diameter) were drilled in each side of the calvaria over the brain regions of interest, which included the cortex of the frontal lobe, the head of the caudate nucleus, the thalamus, the corona radiata of the frontal lobe, the centrum semiovale, the internal capsule, and the cerebellum. A 25l Hamilton syringe with a 26-gauge needle was used to administer the AAV1-fMANB vector. The needle was placed into each drill hole and lowered to the base of the brain; 2l vector was injected, and 2 minutes was allowed for vector diffusion. The needle was withdrawn vertically at 0.25cm distances, and the injection procedure was repeated until the needle was no longer in the brain. A total of 125l vector was administered per cerebral hemisphere of each cat. A total of 25l vector was administered per cerebellar hemisphere. In all cats, both cerebral hemispheres and both cerebellar hemispheres received the AAV1-fMANB vector. Magnetic Resonance Imaging Magnetic resonance imaging (MRI) was performed on anesthetized cats (normal: n ⫽ 3; AMD: n ⫽ 3; treated: n ⫽ 3). Cats were imaged at 16 weeks old using a dedicated research 1.5Tesla MRI scanner (Signa; General Electric Corporation, Milwaukee, WI). The imaging session consisted of a sagittal spin-echo short repetition time (TR), short TE localizer, ax- 356 Annals of Neurology Vol 57 No 3 March 2005 ial fast spin-echo imaging with long TR/TE, and magnetization transfer imaging using previously described methods.9 Data were transferred to a Silicon Graphics workstation (Silicon Graphics, Mountain View CA) and examined using Interactive Data Language–based software (IDL; Research Systems, Boulder, CO) written by one of the authors (J.C.M.). The magnetization transfer ratio (MTR) was calculated as a quantitative index of magnetization transfer effect and was used as a surrogate marker of myelination.9,30 Five regions in the white matter (cerebellar white matter, corona radiata, internal capsule, corpus callosum, and centrum semiovale) and five regions in the gray matter (cerebellar gray matter, lateral geniculate nucleus, hippocampus, gray matter of the gyrus lateralis, and caudate nucleus) were examined in both the left and right side of the brain. Ten regions of interest were averaged in each brain region on each side of the brain. Preparation of Brain Tissue Immediately before death, cats were anesthetized and 0.5ml heparin (1,000U/ml) was administered intravenously. They were killed with an intravenous overdose of barbiturates, in accordance with the American Veterinary Medical Association guidelines, followed by intracardiac perfusion with 250ml of 0.9% cold saline, and then 750ml of 10% neutral buffered formalin. The cerebrum and rostral brainstem were sectioned (⬇3mm in thickness) in a dorsal plane. Sequential sections were alternately frozen in dry ice for biochemical analysis; fixed in 10% formalin for 24 hours, and then paraffin embedded for histology; and fixed in 10% formalin for 24 hours, and then embedded in ornithine carbamoyltransferase and frozen for in situ hybridization. The cerebellum and caudal brainstem were sectioned transversely into slices approximately 3mm in thickness and were processed in the same manner as the forebrain. Paraffin-embedded sections were stained with either hematoxylin and eosin or Luxol fast blue. Enzyme Biochemistry for Lysosomal ␣-Mannosidase Activity in Brain LAMAN enzymatic activity was measured on formalin-fixed brain tissue homogenized in Tris buffer. Samples were centrifuged, the supernatant was collected, and enzyme activity was determined on the substrate 4-methylumbelliferyl-␣-Dmannopyranosidase at 37°C for 1 hour.21,26 Protein determination was done using a modified Lowry method. Activity was expressed as nanomoles of 4-methylumbelliferone released per milligram total protein per hour in a dorsal plane section of tissue covering the entire head circumference. In Situ Hybridization for Feline ␣-Mannosidase cDNA Messenger RNA Nonradioactive in situ hybridization was performed on frozen tissue sections to detect fMANB messenger RNA. Digoxigenin-labeled antisense (700bp) and sense (1,115bp) complementary RNA probes were generated, and the probe length was verified on a formaldehyde gel. The in situ hybridization protocol and colorimetric staining using an alkaline phosphatase-mediated nitroblue tetrazolium reduction/ bromochloroindolyl phosphate reaction was performed as Table. Severity of Neurological Dysfunction in AMD and Treated Cats 8 Weeks AMD (n) Treated (n) Whole-body tremor Mild (6) Intention tremor 12 Weeks 18 Weeks AMD (n) Treated (n) AMD (n) Mild (6) Moderate (6) Moderate (6) Severe (6) Mild (6) Mild (6) Mild (6) Moderate (6) Moderate (6) Severe (6) Truncal ataxia Mild (6) Mild (6) Moderate (6) Moderate (6) Severe (6) Loss of balance Absent (6) Absent (6) Absent (6) Absent (6) Absent (5) Mild (1) Absent (6) Severe (6) Nystagmus at rest Absent (4) Mild (2) Absent (6) Euthanasia necessary 0/6 0/6 0/6 0/6 6/6 Mild (3) Moderate (3) Mild (3) Moderate (3) Absent (3) Mild (3) Absent (3) Mild (3) 0/6 Severe (6) Treated (n) All normal cats had no neurological dysfunction at any age. AMD ⫽ ␣-mannosidosis. described previously.31,32 Sections from both in situ hybridization and enzyme histochemistry reactions were mounted on 100% glycerol and photographed using a light microscope. Statistical Methods The Kruskal–Wallis test was used to compare whether a significant difference was present among normal, AMD, and treated cats. If the Kruskal–Wallis test was significant ( p ⬍ 0.05), then the Wilcoxon signed rank test was used for pairwise comparison of AMD-normal, AMD-treated, and normal-treated groups. A p value less than 0.05 was considered significant. Results Treated Cats Show Marked Improvement in Neurological Signs Cats affected with AMD had a progressive age at onset of specific signs of nervous system dysfunction. In six AMD cats, fine whole-body tremors were first seen at a mean age at onset of 5.0 ⫾ 0.0 weeks. Intention tremors and truncal ataxia developed at 7.3 ⫾ 1.5 and 7.3 ⫾ 2.5 weeks, respectively. Coarse whole-body tremors (12.0 ⫾ 3.6 weeks) and loss of balance (13.7 ⫾ 4.2 weeks) had a more variable age at onset of 7 to 16 weeks of age, with all cats showing these signs of disease by 16 weeks of age. Nystagmus at rest developed at 16 ⫾ 1.0 weeks of age. The cats’ conditions continued to deteriorate, and they were killed at between 18 and 20 weeks of age. Six 8-week-old AMD cats were injected with the AAV1-fMANB vector. These treated cats were injected intraparenchymally with a total of 125l vector into six tracks in each cerebral hemisphere, plus 25l vector into one track in each cerebellar hemisphere, for a total of approximately 2.0 ⫻ 1012 genome equivalents per brain. All cats recovered well and were eating within 12 hours of surgery. The six cats in the treatment group had disease progression similar to age-matched untreated cats until 12 to 14 weeks of age (Table), when the neurological disease stabilized or improved in every treated cat. Three cats did not experience development of loss of balance or a resting nystagmus, and none experienced development of coarse whole-body tremors (see the Table). From 12 to 18 weeks of age, the intention tremor and truncal ataxia improved in three cats, and the wholebody tremor improved in all treated cats (see the Table). By 18 weeks of age, all of the treated cats had only mild clinical signs, and their condition was markedly improved compared with untreated aged-matched AMD cats. The treated cats could be distinguished from normal cats only by a mild truncal ataxia, an intermittent fine head tremor, and a short-strided gait. All of the treated cats required euthanasia by 18 weeks of age. A video comparing a 16-week-old affected cat with an age-matched treated cat may be viewed (see supplemental figure). Magnetic Resonance Imaging Abnormalities Improve in Treated Cats T2-weighted MRI of the brains of 16-week-old cats with AMD showed the cerebral hemisphere and cerebellar white matter to be isointense to gray matter (Fig 1). In comparison, normal cats showed the white matter to be hypointense to gray matter. Abnormal white matter signal intensity in AMD cats has been previously shown to be caused by decreased myelination.9 Cats treated with AAV1-fMANB showed the signal intensity of the cerebral white matter to be hypointense to gray matter (see Fig 1). In the cerebellum, treated cats showed less of an improvement in signal intensity, with most regions remaining isointense to gray matter. The MTR is a normalized value of signal intensity that allows quantitative comparison of the magnetizaVite et al: Gene Therapy for ␣-Mannosidosis 357 Fig 1. Normalization of magnetic resonance imaging in adeno-associated virus serotype 1 vector-feline ␣-mannosidase gene (AAV1fMANB)–treated cats. Transverse images of the cerebrum at the level of the thalamus in 16-week-old cats are shown. In normal cats, white matter is hypointense to gray matter, as can be seen in the corona radiata (arrowhead). In contrast, in cats with AMD, white matter is isointense to gray matter. In affected cats treated at 8 weeks of age, white matter signal intensity is more similar to that of normal cats. Magnetization transfer ratio values (mean and standard deviation ) of white matter tracts at 16 weeks of age in normal, AMD, and AAV1-fMANB–treated cats are shown. AMD ⫽ ␣-mannosidosis. tion transfer effect, and it can be used as an antemortem quantitative marker of myelination.30 At 8 and 16 weeks of age, AMD cats had an MTR significantly less than age-matched normal cats, indicating the presence of significant myelination abnormalities.9 Remarkably, when treated cats were examined at 16 weeks of age, their MTR values were statistically the same as normal cats ( p ⬍ 0.05) in the corona radiata, internal capsule, and corpus callosum (see Fig 1). Comparison of treated cats with aged-matched AMD cats showed significant improvement ( p ⬍ 0.05) in the corona radiata, internal capsule, corpus callosum, and centrum semiovale (see Fig 1). No significant change was detected in the cerebellar white matter. In the gray matter, there are no statistically significant differences in the MTR between normal and AMD cats9; thus, no difference was expected in the treated cats, which was the case (data not shown). Postmortem Analysis Four of the treated cats were killed at 18 weeks of age to evaluate brain pathology. One treated cat was evaluated weekly until 30 weeks of age, and one cat was evaluated until 56 weeks of age. The cat that was killed at 30 weeks of age had begun to show neurological 358 Annals of Neurology Vol 57 No 3 March 2005 abnormalities similar to those seen in untreated 18week-old cats, except that no whole-body tremors developed; thus, disease progression was significantly delayed. The cat that was killed at 56 weeks of age had no deterioration in clinical status after 18 weeks of age. At 56 weeks of age, this cat showed no intention tremor and had only a mild, intermittent, fine, wholebody tremor; an occasional loss of balance, and a shortstrided gait. The brains were removed, and the location of the needle tracks were identified grossly by small indentations on each brain’s surface and histologically by the presence of hemosiderin (Fig 2A, B). In situ hybridization demonstrated fMANB messenger RNA to be present up to 2.5mm from the needle track, and evidence of transduction was found in the gray and the white matter (see Fig 2B). The maximum radius from the needle track in which in situ–positive cells were present was measured and averaged for the cerebral cortex, caudate nucleus, thalamus, hippocampus, and internal capsule. No significant differences in the radius of transduction were found among these brain regions, and the average of all five brain regions showed a transduction radius of 1.40 ⫾ 0.20mm. The enzymatic activity of LAMAN was measured on 1-mm- Fig 2. Increased ␣-mannosidase gene (MANB) messenger RNA and lysosomal ␣-mannosidase (LAMAN) activity in treated cats. (A) A dorsoventral view of the cat brain shows the injection sites of adeno-associated virus serotype 1 vector-feline ␣-mannosidase gene (AAV1-fMANB). Transduction of the following specific brain regions was assessed: the frontal cortex and corona radiata at site 1; the caudate nucleus at site 2; the thalamus at sites 3 and 4; the centrum semiovale and internal capsule at site 5; the hippocampus at site 6; and the cerebellum at site 7. (B) In situ hybridization with the antisense probe demonstrated MANB messenger RNA in gray and white matter up to 2.5mm from the injection site. The sense probe, used as a control, showed little to no evidence of hybridization (original magnification, ⫻50; bar ⫽ 1mm). (C) In treated cats (n ⫽ 3), LAMAN activity in a dorsal plane section of brain covering the entire head circumference was 4% of that found in normal brain (n ⫽ 3) and 7% of that found in the brain of heterozygote cats (n ⫽ 3). Vite et al: Gene Therapy for ␣-Mannosidosis 359 Fig 3. Resolution of neuronal storage seen on hematoxylin and eosin–stained tissue. ␣-Mannosidosis (AMD) cats showed marked neuronal swelling caused by lysosomal storage throughout the brain. In contrast, treated cats showed complete resolution of cellular swelling and storage several millimeters away from the injection site and decreased cellular storage in the occipital cortex, which was 15mm caudal to the nearest injection track. Original magnification, ⫻400. Bar ⫽ 0.1mm. (insets) Original magnification, ⫻1,000. thick slabs of formalin-fixed frozen brain tissue cut in a dorsal plane covering the entire brain circumference. LAMAN activity was measured as 4% of that found in normal brain (see Fig 2C). Histological analysis of tissue sections stained with hematoxylin and eosin showed complete resolution of storage in neurons, glia, and endothelial cells up to 4.5mm from the injection track (Fig 3). Complete correction of storage was defined as no evidence of cytoplasmic vacuoles in any cell in a ⫻40 field of view. The radius of the region of brain surrounding the needle track that showed complete resolution of lysosomal storage was measured and averaged for the cerebral cortex, caudate nucleus, thalamus, hippocampus, and internal capsule. No significant differences in the radius of complete correction were found among these brain regions, and the average of these regions showed a radius of complete storage correction of 2.12 ⫾ 0.84mm. No significant difference between the radius of transduction and the radius of complete correction of storage was found. Evidence of lysosomal storage increased as the distance from the needle track increased. However, no regions of the treated cat brain showed lysosomal storage as severe (as determined by evaluation of the number of cells affected and the foaminess of the cytoplasm) as that found in untreated AMD cats. Even in regions of the brain distant from the injection tracks, such as the occipital cortex, cells were not as swollen as those seen 360 Annals of Neurology Vol 57 No 3 March 2005 in untreated AMD cats (see Fig 3). Resolution of storage also could be seen in cells of the choroid plexus, ependyma, and meninges (Fig 4). Abnormalities of myelination also improved in treated cats. Histological examination of the brains of 18-week-old AMD cats confirmed myelin deficiency, identified by Luxol fast blue stain intensity, distention, and splitting of myelin sheaths (Fig 5). In contrast, the treated cats showed Luxol fast blue staining intensity of white matter tracts that was more similar to that seen in normal cats, as well as less splitting of myelin sheaths (see Fig 5). Finally, the brain of the 56-week-old treated cat was examined. In situ hybridization showed fMANB messenger RNA was being produced a year after injection of the vector. Lysosomal storage was absent to mild, and myelination appeared normal throughout the brain (data not shown). Discussion We tested whether an AAV vector could deliver sufficient amounts of MANB complementary DNA into neurons and glia to engineer cells in situ to serve as small “enzyme pumps” to release enzyme and crosscorrect neighboring untransduced cells.33 Injection of AAV1-fMANB into the brains of cats with AMD resulted in transduction of gray and white matter and increased LAMAN expression in the brain. Complete resolution of lysosomal storage was seen in the areas Fig 4. Resolution of cell swelling in nonneuronal cells of the brain. Treated cats showed resolution of cellular swelling and storage in the choroid plexus and in ependymal, meningeal, and endothelial cells. Original magnification, ⫻400. Bar ⫽ 0.1mm. (insets) Original magnification, ⫻1,000. AMD ⫽ ␣-mannosidosis. overlapping the areas of transduction. However, even in the untransduced cells of the caudal forebrain, cellular storage was much less than in the same brain regions of untreated AMD cats. Two mechanisms may account for the improvement in cellular storage at large distances from the transduction sites. First, only small quantities of LAMAN diffusing from the injection site are probably sufficient to improve storage in distant brain regions. Evidence that low levels (⬍5%) of normal enzyme can mediate dramatic decreases in lysosomal storage has been demonstrated in several mouse models of lysosomal storage diseases.2,3,34 Second, an alternate method of enzyme distribution may occur, such as axonal transport or circulation through the cerebrospinal fluid. Axonal transport of the lysosomal enzyme ␤-glucuronidase was demonstrated in the mucopolysaccharidosis VII mouse brain by finding cells positive for enzyme activity by a histochemical reaction stain and negative for gene expression by in situ hybridization.35,36 Unfortunately, an equally sensitive histochemical stain for LAMAN enzymatic activity does not exist; therefore, similar studies could not be performed in the feline AMD model. Methods to monitor treatment noninvasively will be critical to the success of gene therapy of the CNS. The large size of the cat brain allows for the evaluation of regional correction of neuropathology using MRI. We previously used MRI to identify and quantify significant myelin loss in the AMD cat brain at 8 and 16 weeks of age compared with normal age-matched cats.9 In normal cats, there is a significant increase in MTR in the white matter between 8 and 16 weeks of age corresponding to the time course of myelin maturation. No increase in MTR was seen in AMD cats during this same period, thus AMD prevents the maturation from proceeding normally.9 In contrast, the AAV1-fMANB–treated cats had MTR values of the forebrain that were statistically the same as values for normal cats at 16 weeks of age. The improved myelination seen in treated cats shows that myelination actually increased in treated cats beyond the results at 8 weeks of age. Thus, treatment of AMD cats during the period of myelin maturation can resolve some myelination abnormalities, even though extensive disease is already present at the time of therapy. Using the MTR as a surrogate measure of myelination, we noted greater improvement in the forebrain than in the cerebellum, but gene transduction was also less extensive in the cerebellum (see Fig 2). Evidence of improved cerebellar myelination in treated cats was seen by histological staining, suggesting that MTR is not as sensitive as histological analysis. Vite et al: Gene Therapy for ␣-Mannosidosis 361 Fig 5. Effect of adeno-associated virus serotype 1 vector-feline ␣-mannosidase gene (AAV1-fMANB) on white matter pathology. Luxol fast blue (LB)– and hematoxylin and eosin (H&E)–stained tissue were compared among normal, ␣-mannosidosis (AMD), and treated 18-week-old cats. AMD cats showed decreased LB staining intensity, increased spacing between myelin sheaths, and marked oligodendrocyte swelling. In contrast, treated cats showed LB stain intensity more similar to that seen in normal cats, and complete resolution of cellular swelling and storage several millimeters from the injection site. Original magnification, ⫻25 (LB); ⫻400 (H&E). Bar ⫽ 1mm (LB); 0.1mm (H&E). (insets) Original magnification, ⫻1,000. The treated cats showed remarkable improvements in neurological dysfunction, which were first observed 4 to 6 weeks after surgery (12–14 weeks of age), which is consistent with the increase of AAV vector expression in the mouse brain.35,37 Between 14 and 18 weeks of age, no worsening of clinical signs were seen, whereas the conditions of untreated cats deteriorated significantly during this period. Most importantly, all of the treated cats showed improvement in clinical signs rather than just stabilization, indicating that some signs of disease can be reversed by treatment. In all cases, the treated cats were doing well at 18 weeks of age and euthanasia was performed only to examine brain pathology. Two cats were examined until 30 and 56 weeks of age, which is longer than any untreated AMD cat had survived in our and other colonies.9,24,26,29,38 The resolution of clinical signs corresponded to the improvement in histopathological disease. Continuous action tremors can be caused by neuronal storage and decreased myelin,38,39 both of which improved in treated cats. Signs of cerebellar disease, such as loss of balance, intention tremors, nystagmus, and truncal 362 Annals of Neurology Vol 57 No 3 March 2005 ataxia, also improved in the treated cats, which showed a corresponding reduction in Purkinje cell storage and histological improvement in cerebellar white matter myelination. Resolution of some but not other signs of neurological dysfunction may be attributed to either differences in therapeutic effect among various parts of the nervous system or the age of cats at the time of surgery. A likely explanation for the absence of coarse tremors in the treated cats is that treatment was started before the age at which this tremor developed in untreated cats, thus preventing its onset. In contrast, gene transfer improved, but did not fully resolve, the truncal ataxia and head tremor. This may be because of the incomplete correction of storage in all cells of the brain or, alternatively, because of an inability to fully reverse disease that was present before the initiation of therapy. Some of the neuropathology may not be responsive to normal enzyme at the age we initiated treatment, because AMD cats treated with bone marrow transplantation at 8 to 12 weeks of age continue to exhibit mild cerebellar signs, despite significant improvements in overall neurological disease.9 This study shows that somatic gene transfer to the CNS can deliver functioning enzyme directly to deficient brain cells without the high morbidity and mortality rates associated with bone marrow transplantation.40,41 The global improvement in neuropathology was achieved by a single operation on a brain that is much closer in size to the human brain than is the mouse brain and that has an extensively sulcated cortex, suggesting that this may be a clinically feasible approach for human treatment. The cat and infant brain have, at their widest dorsal plane width, an area of 12.5 and 140 cm2, respectively. Significant correction of storage was seen in the occipital cortex of treated cats, approximately 15mm from the nearest injection track. Assuming that this correction was caused by LAMAN diffusing from the nearest injection sites, it is hypothesized that three injection tracks in each cerebral hemisphere may be sufficient to mediate significant storage correction throughout the entire cat forebrain. Assuming similar transduction and enzyme secretion in the human infant brain, approximately 20 to 30 injections into each cerebral hemisphere would be needed for correction. However, the use of stronger promoters to increase the total amount of LAMAN secreted from transduced cells could further reduce the number of injections necessary to treat the human brain. This study was supported by the NIH (National Institute of Neurological Disorders and Stroke, K08-NS02032, NS-38690, C.H.V.; National Institute of Diabetes and Digestive and Kidney Diseases, DK-07748, M.A.P.; DK-63973, J.H.W.; National Center for Research Resources, RR-02512, M.E.H.). We thank P. O’Donnell, J. Zweigel, K. Cullen, T. O’Malley, and the Animal Models Core of the Center for Comparative Medical Genetics (RR-02512) School of Veterinary Medicine for assistance with the animal procedures; A. Polesky, B. Chambers, and J. Burns for assistance in sectioning and processing tissue; D. Aleman for assistance in video editing; H. Poptani and S. Magnitsky for assistance in image analysis; T. Berg for providing the fMANB complementary DNA; and A. Salvetti and L. Wang for assistance in viral vector production (DK-47747). References 1. Scriver CR, Beaudet AL, Sly WS, Valle D. eds. The metabolic and molecular bases of inherited disease. 8th ed. New York: McGraw Hill, 2001. 2. Ellinwood NM, Vite CH, Haskins ME. Gene therapy for lysosomal storage diseases: the lessons and promise of animal models. J Gene Med 2004;6:481–506. 3. Watson DJ, Wolfe JH. Lentiviral vectors for gene transfer to the central nervous system: applications in lysosomal storage disease animal models. In: Machida C, ed. Viral vectors for gene therapy: methods and protocols. Totowa, NJ: Humana Press, 2002:383– 403. 4. Berman AL. The brain stem of the cat: a cytoarchitectonic atlas with stereotaxic coordinates. Madison, WI: University of Wisconsin Press, 1968. 5. Berman AL, Jones EG. The thalamus and basal telencephalon of the cat: a cytoarchitectonic atlas with stereotaxic coordinates. Madison, WI: University of Wisconsin Press, 1982. 6. Summers BA, Cumming JF, de Lahunta A. Veterinary neuropathology. St Louis: Mosby, 1995. 7. Jolly RD, Walkley SU. Lysosomal storage diseases of animals: an essay in comparative pathology. Vet Pathol 1997;34: 527–548. 8. Wolfe JH, Sands MS, Harel N, et al. Gene transfer of low levels of beta-glucuronidase corrects hepatic lysosomal storage in a large animal model of mucopolysaccharidosis VII. Mol Ther 2000;2:552–561. 9. Vite CH, McGowan JC, Braund KG, et al. Histopathology, electrodiagnostic testing, and magnetic resonance imaging show significant peripheral and central nervous system myelin abnormalities in the cat model of alpha-mannosidosis. J Neuropathol Exp Neurol 2001;60:817– 828. 10. Ponder KP, Melniczek JR, Xu L, et al. Therapeutic neonatal hepatic gene therapy in mucopolysaccharidosis VII dogs. Proc Natl Acad Sci USA 2002;99:13102–13107. 11. Xu L, Haskins ME, Melniczek JR, et al. Transduction of hepatocytes after neonatal delivery of a Moloney murine leukemia virus based retroviral vector results in long-term expression of beta-glucuronidase in mucopolysaccharidosis VII dogs. Mol Ther 2002;5:141–153. 12. Vite CH, Passini MA, Haskins ME, Wolfe JH. AAV vectormediated gene transfer for the cat brain. Gene Ther 2001;10: 874 – 881. 13. Davis D, Ulatowski J, Eleff S, et al. Rapid monitoring of changes in water diffusion coefficients during reversible ischemia in cat and rat brain. Magn Reson Med 1994;231: 454 – 460. 14. Power C, Moench T, Peeling J, et al. Feline immunodeficiency virus causes increased glutamate levels and neuronal loss in brain. Neuroscience 1997;77:1175–1185. 15. Heiss WD, Graf R, Weinhard K. Relevance of experimental ischemia in cats for stroke management: a comparative reevaluation. Cerebrovasc Dis 2001;11:73– 81. 16. Podell M, Hadjiconstantinou M, Smith MA, Neff NH. Proton magnetic resonance imaging and spectroscopy identify metabolic changes in the striatum in the MPTP feline model of parkinsonism. Exp Neurol 2003;170:159 –166. 17. Kjellman B, Gamstorp I, Brun, et al. Mannosidosis: a clinical and histopathologic study. J Pediatr 1969;75:366 –373. 18. Sung JH, Hayano M, Desnick RJ. Mannosidosis: pathology of the nervous system. J Neuropathol Exp Neurol 1977;36: 807– 820. 19. Michalski J-C, Klein A. Glycoprotein lysosomal storage disorders: alpha-mannosidosis and beta-mannosidosis, fucosidosis, and alpha-N-acetylgalactosaminidase deficiency. Biochim Biophys Acta 1999;1455:69 – 84. 20. Thomas GH. Disorders of glycoprotein degradation and structure: alpha-mannosidosis, beta-mannosidosis, fucosidosis, and sialidosis. In: Scriver CR, Beaudet AL, Sly WS, Valle D, eds. Metabolic basis of inherited disease. 8th ed. New York: McGraw-Hill, 2001:3507–3533. 21. Berg, T, Tollersrud OK, Walkley SU, et al. Purification of feline lysosomal alpha-mannosidase, determination of its cDNA sequence and identification of a mutation causing alphamannosidosis in Persian cats. Biochem J 1997;328:863– 870. 22. Sun H, Wolfe JH. Recent progress in lysosomal alphamannosidase and its deficiency. Exp Mol Med 2001;33:1–7. 23. Burditt LJ, Chotal K, Kirani S, Nagen PG. Biochemical studies on a case of feline alpha-mannosidosis. Biochem J 1980;189: 467– 473. 24. Vandevelde M, Fankhauser R, Bichsel P, et al. Hereditary neurovisceral mannosidosis associated with alpha-mannosidase deficiency in a family of Persian cats. Acta Neuropathol 1982;58: 64 – 68. Vite et al: Gene Therapy for ␣-Mannosidosis 363 25. Blakemore WF. A case of mannosidosis in the cat: clinical and histopathological findings. J Small Anim Pract 1986;27: 447– 455. 26. Jezyk PF, Haskins ME, Newman LR. Alpha-mannosidosis in a Persian cat. J Am Vet Med Assoc 1986;189:1483–1485. 27. Maenhout T, Kint JA, Dacremont G, Ducatelle R. Mannosidosis in a litter of Persian cats. Vet Rec 1988;122:351–354. 28. Cummings JF, Wood PA, de Lahunta A, et al. The clinical and pathologic heterogeneity of feline alpha-mannosidosis. J Vet Intern Med 1988;2:163–170. 29. Alroy J, Schunk KL, Ranghavan SS, et al. Alpha-mannosidase deficiency in Persian cats: a model of human alphamannosidosis. In: Salvayre R, Douste-Blazy L, Gatt S, eds. Lipid storage disorders. New York: Plenum Publishing, 1988: 649 – 659. 30. Dousett V, Franconi JM, Balderrama J, et al. Use of magnetization transfer contrast to improve 3D MR angiography. Neuoradiology 1995;36:188 –192. 31. Passini MA, Watson DJ, Vite CH, et al. Intraventricular brain injection of AAV1 in neonatal mice results in complementary patterns of neuronal transduction to AAV2 and total long-term correction of storage lesions in the brains of beta-glucuronidasedeficient mice. J Virol 2003;77:7034 –7040. 32. Barthel LK, Raymond P. In situ hybridization studies of retinal neurons. Meth Enzymol 2000;316:579 –590. 33. Taylor RM, Wolfe JH. Decreased lysosomal storage in the adult MPS VII mouse brain in the vicinity of grafts of retroviral vector-corrected fibroblasts secreting high levels of betaglucuronidase. Nat Med 1997;3:771–774. 364 Annals of Neurology Vol 57 No 3 March 2005 34. Snyder EY, Wolfe JH. Central nervous system cell transplantation: a novel therapy for storage diseases? Curr Opin Neurol 1996;9:126 –136. 35. Passini MA, Lee EB, Heuer GG, Wolfe JH. Distribution of a lysosomal enzyme in the adult brain by axonal transport and by cells of the rostral migratory stream. J Neurosci 2002;22: 6436 – 6446. 36. Hennig AK, Levy B, Ogilvie JM, et al. Intravitreal gene therapy reduces lysosomal storage in specific areas of the CNS in mucopolysaccharidosis VII mice. J Neurosci 2003;23:3302– 3307. 37. Skorupa AF, Fisher KJ, Wilson JM, et al. Sustained production of beta-glucuronidase from localized sites after AAV vector gene transfer results in widespread distribution of enzyme and reversal of lysosomal storage lesions in a large volume of brain in mucopolysaccharidosis VII mice. Exp Neurol 1999;160:17–27. 38. Duncan ID. Abnormalities of myelination of the central nervous system associated with congenital tremor. J Vet Intern Med 1987;1:10 –23. 39. Cuddon PA. Tremor syndromes. Progress in Veterinary Neurology 1990;1:3,285–299. 40. Walkley SU, Thrall MA, Dobrenis K, et al. Bone marrow transplantation corrects the enzyme defect in neurons of the central nervous system in a lysosomal storage disease. Proc Natl Acad Sci USA 1994;91:2970 –2974. 41. Wall DA, Grange DK, Goulding P, Daines M, et al. Bone marrow transplantation for the treatment of alpha-mannosidosis. J Pediatr 1992;133:282–285.