Gene therapy for amyotrophic lateral sclerosis and other motor neuron diseases.

There are several incurable diseases of motor neuron degeneration, including amyotrophic lateral sclerosis (ALS), primary lateral sclerosis, hereditary spastic hemiplegia, spinal muscular atrophy, and bulbospinal atrophy. Advances in gene transfer techniques coupled with new insights into molecular pathology have opened promising avenues for gene therapy aimed at halting disease progression. Nonviral preparations and recombinant adenoviruses, adeno-associated viruses, herpesviruses, and lentiviruses may ultimately transduce sufficient numbers of cerebral, brainstem, and spinal cord neurons for therapeutic applications. This could be accomplished by direct injection, transduction of lower motor neurons via retrograde transport after intramuscular injection, or cell-based therapies. Studies using transgenic mice expressing mutant superoxide dismutase 1 (SOD1), a model for one form of ALS, established that several proteins were neuroprotective, including calbindin, bcl-2, and growth factors. These same molecules promoted neuronal survival in other injury models, suggesting general applicability to all forms of ALS. Potentially correctable genetic lesions have also been identified for hereditary spastic hemiplegia, bulbospinal atrophy, and spinal muscular atrophy. Finally, it may be possible to repopulate lost corticospinal and lower motor neurons by transplanting stem cells or stimulating native progenitor populations. The challenge ahead is to translate these basic science breakthroughs into workable clinical practice.

Quantitative analysis of converging spinal and cuneate mossy fibre afferent projections to the rat cerebellar anterior lobe.

The convergence/divergence of mossy fibre afferent projections to the cerebellar anterior lobe from a single lumbar segment, from adjacent or widely separated lower thoracic and lumbar segments, and finally from the lower thoracic-upper lumbar spinal cord and the brainstem cuneate nuclei was quantitatively analysed in adult rats. Spinal and cuneate mossy fibre terminals were differentially labelled with biotinylated dextran amine and cholera toxin subunit B, immunohistochemically identified in the same histological sections, and their spatial distributions quantitatively plotted in computer reconstructions of the unfolded anterior lobe cortex. Afferent convergence was quantified by calculating the number of biotinylated dextran amine-labelled terminals that radially overlapped with cholera toxin-labelled terminals at points on the unfolded cortical map that represented theoretical Purkinje cells. Spino- and cuneocerebellar mossy fibre terminals are organized in patches that are oriented in parasagittally-oriented stripes or transversely oriented bands. Afferent convergence was greatest following biotinylated dextran amine and cholera toxin injections in the same or adjacent spinal lumbar segments (60 and 52%, respectively). When biotinylated dextran amine and cholera toxin were injected in a single segment differentially labelled terminals appeared randomly intermingled in common patches. There was a trend for terminals labelled from adjacent lumbar segments to be more segregated in the patches. Segmentally separated biotinylated dextran amine and cholera toxin spinal cord injections (four lumbar segments) resulted in clearly segregated (80%) biotinylated dextran amine from cholera toxin-labelled terminal patches or patches with distinct divergence of the differentially labelled terminals in the patch. Cuneocerebellar terminals labelled with biotinylated dextran amine were located in patches, stripes, and bands spatially segregated from terminal patches, stripes, and bands of cholera toxin-labelled spinal afferents except at their immediate borders where some radial overlap occurred (9-22%). These anatomical findings for a fractured somatotopy of spinal and cuneate inputs to the cerebellar anterior lobe complement neurophysiological findings for a very similar pattern of organization of cutaneous inputs to the posterior lobe, and are discussed in light of potential mechanisms for anterior lobe processing of somatosensory information.

Lower thoracic upper lumbar spinocerebellar projections in rats: a complex topography revealed in computer reconstructions of the unfolded anterior lobe.

The topography of wheatgerm agglutinin-horseradish peroxidase/horseradish peroxidase-labeled mossy fiber terminals of lower thoracic-upper lumbar (T12-L3) spinal projections to the cerebellar anterior lobe was quantitatively analysed in adult rats. Computer-based image analysis mapped the orthogonal (parallel to the surface) distribution of labeled terminals in two-dimensional reconstructions of the unfoled anterior lobe cortex. The radial (perpendicular to the surface) distribution of terminals within the granule cell layer was mapped by computing whether the terminals were in either the outer- or inner-halves of this layer. The number of labeled terminals in each lobule was calculated. In the anterior lobe, lower thoracic-upper lumbar spinocerebellar projections terminate primarily in lobules II (mean 27.14%), III (mean 38.68%), and IV (mean 19.31%). Different-sized bilateral injections restricted to L1 were used to study the organization of intrasegmental spinocerebellar projections. Small injections into L1 labeled a limited number of terminals which were located either in clusters or were spatially isolated. Intermediate-sized intrasegmental injections resulted in additional clusters of labeled terminals. Many of the terminal clusters were spatially related and formed larger irregularly shaped patches. Large intrasegmental injections labeled terminal clusters and patches that were discontinuous but aligned parallel to the longitudinal (transverse) axis of lobules II-IV. Injections including segments rostral and caudal to L1 were used to study the topography of intersegmental lower thoracic-upper lumbar spinocerebellar projections. Multisegmental injections increased the number of labeled terminal clusters and patches which obscured the pattern of segmental input, but there was still a transversely oriented pattern of termination. Distinct transversely aligned terminal free areas remained apparent. Lower thoracic-upper lumbar spinocerebellar projections terminated in both the outer- and inner-halves of the granule cell layer, but overall were more numerous in the outer-half of this layer. In serially spaced sagittal sections, however, the majority of terminals alternated between the outer- and inner-halves of the granule cell layer. Outer- and inner-terminals were not spatially segregated in their orthogonal distribution. These results indicate lower thoracic-upper lumbar spinocerebellar projections have a complex three-dimensional topography in the anterior lobe. These findings are discussed in relation to previous findings for a sagittally oriented topography for lower thoracic-upper lumbar spinocerebellar projections and in the context of how cerebellar somatosensory afferent input may be organized.

Transduction of murine cerebellar neurons with recombinant FIV and AAV5 vectors.

Our data demonstrate that vectors derived from recombinant feline immunodeficiency virus (rFIV) and adeno-associated virus type 5 (rAAV5) transduce cerebellar cells following direct injection into the cerebellar lobules of mice. Both recombinant viruses mediated gene transfer predominantly to neurons, with up to 2500 and 1500 Purkinje cells transduced for rAAV5 or rFIV-based vectors, respectively. The vectors also transduced stellate, basket and Golgi neurons, with occasional transduction of granule cells and deep cerebellar nuclei. rAAV5 also spread outside the cerebellum to the inferior colliculus and ventricular epithelium, while rFIV demonstrated the ability to undergo retrograde transport to the physically close lateral vestibular nuclei. Thus, AAV5 and FIV-based vectors show promise for targeting neurons affected in the hereditary spinocerebellar ataxias. These vectors could be important tools for unraveling the pathophysiology of these disorders, or in testing factors which may promote neuronal survival.

Differential labeling of converging afferent pathways using biotinylated dextran amine and cholera toxin subunit B.

We report a new technique for 2-tracer anterograde labeling that permits unequivocal identification of the differentially labeled projections in the same section. One pathway is labeled with biotinylated dextran amine and is visualized as a black to dark gray diaminobenzidine (DAB)-cobalt precipitate by an avidin-biotinylated peroxidase reaction. The other pathway is labeled with cholera toxin subunit B and is visualized as a reddish-brown reaction product using DAB without cobalt as the substrate for peroxidase immunohistochemistry. To maintain serial order, sections can be processed mounted on slides without any loss of sensitivity for either tracer.

Chronic NMDA receptor blockade or muscimol inhibition of cerebellar cortical neuronal activity alters the development of spinocerebellar afferent topography.

The requirement for cerebellar cortical neuronal activity in the development of spinocerebellar afferent topography was investigated in neonatal rats. In adult rats lower thoracic-upper lumbar spinocerebellar projections are localized to sharply circumscribed patches in the granule cell layer of the cerebellar anterior lobe. In transverse sections these patches appear as sagittally oriented stripes. This pattern develops postnatally as many spinal axons which initially project between the incipient stripes are eliminated thereby sharpening the stripe boundaries. We attempted to alter cerebellar cortical neuronal activity in neonatal animals to study the effects of these changes on the development of spinocerebellar stripes. In some experiments glutaminergic excitatory synaptic transmission was chronically blocked with the N-methyl-D-aspartate (NMDA) receptor antagonist 2-aminophosphovaleric acid (APV). In other experiments postsynaptic activity was directly inhibited by the gamma-aminobutyric acid agonist muscimol. Chronic exposure to APV or to muscimol did not affect the initial development of spinocerebellar projections; many spinal axons were present in the anterior lobe and arranged in incipient stripes. Both the APV and the muscimol appeared to prevent the elimination of interstripe projections; consequently the boundaries of the stripes remained poorly defined. These findings suggest that cerebellar cortical neuronal activity is a necessary requirement for the refinement of spinal afferent topography in the anterior lobe.

The postnatal spatial and temporal development of corticospinal projections in cats.

Orthograde labeling and immunocytochemical techniques were used to study the postnatal spatial and temporal development of corticospinal projections in cats. Findings from the orthograde labeling studies indicate that there are three major phases in the spatial development of corticospinal projections: an early period (1-10 postnatal days) when cortical axons grow into the spinal gray from the white matter; an intermediate period (2-5 postnatal weeks) where corticospinal axons develop terminal arborizations in a rostral to caudal, medial to lateral and intermediate gray to dorsal and ventral horn sequence; and, a late period (6-7 postnatal weeks) during which some corticospinal projections are eliminated. The time period over which cortical axons grow into the spinal cord was determined immunocytochemically using a monoclonal antibody against a microtubule associated protein (MAP 1B) present in growing axons. The corticospinal tracts were strongly immunoreactive for MAP 1B during the first three postnatal weeks. MAP 1B immunostaining of these tracts started to decline in the fourth postnatal week and was completely absent by five weeks of age. These findings indicate that the postnatal development of corticospinal projections is spatially and temporally protracted in cats.

Viral-mediated delivery of the late-infantile neuronal ceroid lipofuscinosis gene, TPP-I to the mouse central nervous system.

Classical late-infantile neuronal ceroid lipofuscinosis (LINCL) is caused by mutations in tripeptidyl peptidase I (TPP-I), a pepstatin-insensitive lysosomal protease, resulting in neurodegeneration, acute seizures, visual and motor dysfunction. In vitro studies suggest that TPP-I is secreted from cells and subsequently taken up by neighboring cells, similar to other lysosomal enzymes. As such, TPP-I is an attractive candidate for enzyme replacement or gene therapy. In the present studies, we examined the feasibility of gene transfer into mouse brain using recombinant adenovirus (Ad), feline immunodeficiency virus (FIV) and adeno-associated virus (AAV) vectors expressing TPP-I, after single injections into the striatum or cerebellum. A dual TPP-I- and beta-galactosidase-expressing adenovirus vector (AdTTP-I/nlsbetagal) was used to distinguish transduced (beta-galactosidase positive) cells from cells that endocytosed secreted TTP-I. Ten days after striatal injection of AdTTP-I/nlsbetagal, beta-galactosidase-positive cells were concentrated around the injection site, corpus callosum, ependyma and choroid plexus. In cerebellar injections, beta-galactosidase expression was confined to the region of injection and in isolated neurons of the brainstem. Immunohistochemistry for TPP-I expression showed that TPP-I extended beyond areas of beta-galactosidase activity. Immunohistochemistry for TTP-I after FIVTTP-I and AAV5TTP-I injections demonstrated TPP-I in neurons of the striatum, hippocampus and Purkinje cells. For all three vectors, TPP-I activity in brain homogenates was 3-7-fold higher than endogenous levels in the injected hemispheres. Our results indicate the feasibility of vector-mediated gene transfer of TPP-I to the CNS as a potential therapy for LINCL.