Tay-Sachs disease is a severe, inherited disease of the nervous system caused by accumulation of the brain lipid GM2 ganglioside. Mouse models of Tay-Sachs disease have revealed a metabolic bypass of the genetic defect based on the more potent activity of the enzyme sialidase towards GM2. To determine whether increasing the level of sialidase would produce a similar effect in human Tay-Sachs cells, we introduced a human sialidase cDNA into neuroglia cells derived from a Tay-Sachs fetus and demonstrated a dramatic reduction in the accumulated GM2. This outcome confirmed the reversibility of GM2 accumulation and opens the way to pharmacological induction or activation of sialidase for the treatment of human Tay-Sachs disease.
Tay-Sachs and Sandhoff diseases are autosomal recessive genetic disorders of GM2 ganglioside catabolism due to deficient activity of lysosomal β-hexosaminidase A (Hex A, structure αβ) (1). Tay-Sachs disease is caused by mutation of the α subunit, resulting in deficient Hex A activity, while Sandhoff disease is caused by mutation of the β subunit, with consequent deficiency of Hex A and Hex B (structure ββ) activities. Both diseases cause GM2 gangliosidosis leading to a devastating neurodegenerative disease which is fatal by 2–4 years of age.
We (2) and others (3,4) have established mouse models of Tay-Sachs and Sandhoff disease by targeted disruption of the Hexa (α subunit) and Hexb (β subunit) genes, respectively. In contrast to the similarity of the human diseases, Hexb-/- mice suffer a profound, fatal neurodegenerative disease with onset at 3–4 months of age and death 4–6 weeks later, while Hexar-/-mice escape disease to >1 year of age. The difference between the human and mouse diseases stems from a more efficient conversion in mice of GM2 ganglioside to its asialo derivative, glycolipid GA2, which can be catabolized further by the Hex B isozyme present in Hexa-/- but not Hexb-/- mice (Fig. 1) (2,3). This enzymatic sequence constitutes a metabolic bypass of the Hex A defect that permits disease avoidance in Hexa-/- mice. The enzyme responsible for initiating the bypass is lysosomal sialidase. It occurs in a multienzyme complex with β-galactos-idase and cathepsin A (5–8). It is inactive outside the complex and, indeed, it is rapidly degraded in cells deficient in cathep-sin A in the genetic disease galactosialidosis (9). The activity of lysosomal sialidase is independently deficient in the inborn error sialidosis (10).
In this study, we show that Tay-Sachs neuroglia cells, boosted in the level of functional sialidase by transfection, are able to catabolize accumulated GM2, suggesting that pharmacological stimulation of endogenous sialidase might have therapeutic benefit in Tay-Sachs disease.
In order to evaluate the potential for the sialidase bypass to function in human cells, we made use of a neuroglia cell line, derived from fetal cerebellum from the terminated pregnancy of an Ashkenazi Jewish couple (11), as an in vitro model of the disease. The cells have been shown to have deficient Hex A activity and to accumulate GM2 (12), confirming that they have a functional ganglioside biosynthetic pathway. We tested for the common Ashkenazi Jewish mutations using established DNA tests and determined that the fetus was a compound heterozygote for functionally null mutations in the a subunit (HEXA gene mutations, 1278insTATC and IVS12+1G→C; data not shown), indicating that the cells were devoid of Hex A activity. We confirmed immunocytochemically that we could detect the accumulated GM2 in the affected Tay-Sachs cells, but not in normal neuroglial cells (Fig. 2A and B). We also confirmed the location of the GM2 in lysosomes by co-localization with the lysosomal membrane protein LAMP2 (13; Fig. 2C1–C3) and cathepsin A (Fig. 2D1–D3).
Our approach was to transfect the Tay-Sachs neuroglia cells with an expressible sialidase cDNA and to assay the cells for clearance of GM2 ganglioside. We showed previously that an SV40-transformed derivative of Tay-Sachs neuroglia cells transfected with a cDNA clone encoding the α subunit of Hex A(pCMV-Hex α) produced active Hex A as assayed with a synthetic substrate (14). We repeated this experiment as a positive control, except that we used the untransformed cell line as host, in order to confirm whether the increased Hex A activity would result in depletion of the accumulated GM2. While untransfected cells remained positive for GM2, those expressing a high level of Hex A (44 cells counted) were cleared of the lipid (Fig. 3A-C). This illustrated the capacity of the accumulated GM2 to be successfully degraded in cells with a restored catabolic sequence.
In order to determine whether increasing the activity of lyso-somal sialidase would have a similar impact on GM2 degradation, we performed similar experiments, this time making use of a cDNA encoding the lysosomal sialidase instead of the Hex α subunit. In order to detect the transfected sialidase alone and not the endogenous enzyme, a pCMV vector was prepared with a polyhistidine-tagged human sialidase cDNA (pCMV-HisSial) which could be detected with an anti-polyhistidine antibody (Fig. 4A and B). We first confirmed that the polyhis-tidine-tagged sialidase retained the ability to form a complex with β-galactosidase and cathepsin A. This was accomplished by co-transfecting the Tay-Sachs neuroglial cells with pCMV-HisSial and pCMV-CA (cathepsin A), the latter to promote formation of the multienzyme complex (15,16). The cells were radiolabeled with [35S]methionine and the resulting radioactive proteins were immunoprecipitated using the anti-polyhistidine antibody and resolved using SDS-PAGE. Radioactive polypeptides corresponding in size to those of β-galactosidase (78 kDa) and cathepsin A (20 kDa), as well as sialidase (doublet of 44 and 46 kDa), were detected, indicating that the polyhisti-dine tag does not interfere with assembly of the complex (Fig. 4C).
We could now test the effect of increased sialidase on GM2 levels. Tay-Sachs neuroglial cells were co-transfected with pCMV-HisSial and pCMV-CA. Two days after transfection, the cells were fixed and immunolabeled with anti-GM2 and anti-polyhistidine antibodies. Figure 3D-F illustrates that neu-roglia cells expressing high levels of sialidase show virtually complete depletion of their accumulated GM2. All cells strongly positive for the transfected sialidase (270 cells counted) gave complete or near complete reduction of GM2 ganglioside to background levels.
In an independent test of the efficacy of human sialidase to act on GM2, cultured Tay-Sachs and Sandhoff fibroblasts were preloaded with a ganglioside mixture (1 mg/ml), co-trans-fected with pCMV-HisSial and pCMV-CA and immuno-stained for GM2 and sialidase as before. The fibroblast cultures were from skin biopsies and do not normally synthesize gan-gliosides, so that cells positively staining for GM2 confirmed successful loading of the lipid in the lysosomes. Two days after transfection, both Tay-Sachs (Fig. 5A-C) and Sandhoff fibroblasts (Fig. 5D-F), shown to be overexpressing sialidase by immunostaining (35 and 48 cells counted, respectively), had dramatically reduced levels of GM2 when compared with surrounding untransfected cells.
We conclude that elevation of the level of lysosomal sialidase in Tay-Sachs neuroglia cells, as well as preloaded mutant fibroblasts, is sufficient to facilitate degradation of the accumulated ganglioside and mimic the catabolic bypass that allows the murine model of Tay-Sachs disease to survive (Fig. 1). While our assay was limited to detecting clearance of GM2, we anticipate that continued catabolism through lactosylcera-mide likely occurs because of the much reduced level of gly-colipd GA2 observed in Tay-Sachs compared with Sandhoff brain in the human disease (17). When the sialidase bypass was first documented in Hexa-/- mice, it was proposed that it was made possible by a higher affinity of the mouse enzyme for GM2 or that mice have a higher level of the enzyme in ganglio-side-producing tissues (2,3). While our studies implicate quantity over quality of enzyme, they do not rule out a modest difference in substrate affinity that would favor the bypass pathway in mouse tissues. Resolution of the mechanism will likely require in vitro analysis of the two enzymes.
Our experiments took advantage of the potentiating effect of cathepsin A, transfected in combination with sialidase, to maximize lysosomal expression of the latter enzyme. Cathepsin A is required to facilitate incorporation of sialidase into the lyso-somal β-galactosidase-cathepsin A-sialidase complex, either through ferrying the enzyme to the lysosome (15) or by protecting it from degradation prior to its entry into the lysosome (9). It remains to be determined if stimulating the endogenous synthesis of sialidase would require a concurrent increase in cathepsin A to result in a functional increase in sialidase activity.
As established for the Hexar-/- mouse, it is the presence of functional Hex B that makes it possible for lysosomal sialidase to function in the metabolic sparing of the Hex A defect (2,3,18). This is why the bypass can protect the Hexar-/- but not the Hexb-/- mouse, since the latter lacks both Hex A and Hex B. Thus, only in Tay-Sachs disease would it be possible to shunt GM2 to GA2 and lactosylceramide to permit completion of the catabolic sequence to ceramide.
Inherited neurodegenerative diseases pose an enormous challenge to treatment because of limited access to the brain. In Tay-Sachs and Sandhoff diseases, approaches to therapy have been directed toward the development of viral vectors encoding the deficient enzyme subunit (19–21), the injection of neural progenitor cells expressing functional enzymes (22), the use of bone marrow transplantation to provide enzyme through the circulation (23) and the drug-mediated inhibition of ganglioside biosynthesis (24). Bone marrow transplantation extended the lifespan of Hexb-/- mice, but central nervous system pathology was unchanged by the treatment (23). Drug inhibition of ganglioside synthesis is a promising prospect for limiting the accumulation of GM2. Platt et al. (24) treated Hexa-/- mice with N-butyldeoxynojyrimycin and estimated that they reduced the synthesis of gangliosides in brain by ∼10%. This limited bio-synthetic impairment was sufficient to reduce the presence of swollen lysosomes characteristic of the pathology of GM2 storage. A side-effect of the drug used was severe destruction of the spleen, an issue that would need to be addressed in a chronic treatment method. Also uncertain is the potential impact of interfering with ganglioside biosynthesis in humans. Nevertheless, these are early stage experiments that challenge the notion that Tay-Sachs disease is inaccessible to treatment. Our studies, along with those of Platt et al. (24), indicate that manipulations of brain metabolism are candidates for approaches to treatment. If safe, effective pharmacological intervention can be developed for stimulation of the sialidase bypass, perhaps in combination with marginal inhibition of ganglioside biosynthesis, then an effective treatment may be on the horizon for Tay-Sachs disease.
Materials and Methods
Normal and Tay-Sachs neuroglia cells were generous gifts from L. Hoffman (Neuroscience Centre, Kingsbrook Jewish Medical Center, New York, NY). Tay-Sachs (WG107) and Sandhoff (WG150) fibroblasts were obtained from the Mutant Human Cell Repository (Montreal Children's Hospital, Montreal, Canada) and their molecular defects have been characterized previously. Cell lines were maintained in modified Eagle's medium supplemented with 15% fetal calf serum and antibiotics.
Mouse monoclonal antibody against human LAMP2 was obtained from the Developmental Studies Hybridoma Bank (Baltimore, MD). Texas Red-conjugated goat anti-rabbit IgG, Texas Red-conjugated goat anti-mouse IgG, BODIPY FL-conjugated avidin and biotinylated anti-human antibody were purchased from Molecular Probes (Eugene, OR). The mouse anti-polyhistidine peptide antibody was purchased from Qiagen (Germany). The mouse/human chimeric monoclonal anti-GM2 antibody KM966 was provided by N. Hanai (Tokyo, Japan) and has been shown to be highly specific for GM2 gan-glioside (25,26). The rabbit anti-cathepsin A antibody was provided by A. Pshezhetsky (Montreal, Canada) (9). The anti-Hex A antibody was provided by D. Mahuran (Toronto, Canada) (27).
Subcloning into expression vectors
The human sialidase cDNA (1.9 kb) flanked by XmaIII restriction sites was subcloned into the NotI site of a pCMV vector (28). In order to insert a nucleotide sequence encoding a poly-histidine tag, a synthetic adapter made up of two oligo-nucleotides (5′-CATGCGCGGATCTCATCATCATCACCATCACCTCGTGCAGCCGCTG and 5′-GTCACCAGCGGCTGCACGAGGTGATGGTGATGGTGATGATGATGAGATCCGCGCATG) was prepared. The oligonucleotides were allowed to anneal, creating a blunt 5′-end and a cohesive ZMEII-compatible 3′-end. The pCMV-Sial vector was digested at unique restriction sites using PshAI (position 152) and BstEII (position 171), dephosphorylated and then ligated with the polyhistidine adapter. The ligation mixture was used to transform TOP10F′ cells. Positive clones of pCMV-HisSial were identified by digesting with HhaI and XhoII (Fig. 4A and B). The final pCMV-HisSial vector contained the amino acid sequence MRGSHHHHHH immediately following the signal peptide cleavage site. The pCMV-Hex A and pCMV-cathepsin A vectors were prepared as described before (14,16). At the end of this study, we learned from Dr H. Sakuraba (Tokyo) that our cDNA contained an amino acid difference, Leu90Pro (269T→C), compared with the published sequence (29). A polyhistidine-tagged cDNA prepared subsequently with the corrected sequence confirmed the results obtained with the original clone.
Ganglioside loading of mutant fibroblasts
Confluent Tay-Sachs and Sandhoff fibroblast cultures were incubated for 72 h with Optimem medium (Gibco BRL, Burlington, Ontario, Canada) containing 1 mg/ml bovine ganglio-side mixture (Sigma, Oakville, Ontario, Canada). Twenty-four hours before transfection, the cells were trypsinized and replated at 50–70% confluency on single chamber LabTek slides (Nunc, Burlington, Ontario, Canada).
Expression in mammalian cells
For transient expression, cells were transfected with either pCMV-Hex a vector or pCMV-HisSial/pCMV-CA. Transfection was done using lipofectamine [1 µg vector(s) and 6 µl lipofectamine solution in 1 ml Optimem solution for one-chamber LabTek slides] as described by the manufacturer (Gibco BRL). For controls, a pCMV vector containing no insert or cathepsin A cDNA was used. After transfection, cells were incubated for 24 h in modified Eagle's mediumcontaining 15% fetal calf serum with no antibiotics. On the second day, the medium was replaced with medium containing 15% fetal calf serum and antibiotics. Transfection efficiency was <1% as detected by immunostaining for the expressed product (Hex A or polyhistidine tag).
Immunocytochemical localization of lysosomal sialidase
Transfected cells expressing human sialidase and cathepsin A cDNAs, grown on chambered slides, were washed in phosphate-buffered saline (PBS) and fixed for 30 min in 3.8% para-formaldehyde in PBS. Cells were permeabilized with 0.5% Triton X-100 in PBS for 30 min. After washing twice in cold PBS, cells were blocked with 10% goat serum in PBS for 1 h. For double labeling, primary antibodies (diluted 1:200 in PBS) were incubated with the cells at 4°C overnight. Cells were washed three times for 5 min each in PBS containing 0.1% Tween-20. Cells were then incubated with secondary antibodies (either biotinylated or Texas Red conjugated) for 1 h at room temperature and again washed in PBS containing 0.1% Tween-20. Cells were incubated with BODIPY FL-conjugated avidin for 30–45 min, then were washed again in PBS containing 0.1% Tween-20 three times for 5 min each. A final wash was performed in distilled water to remove residual salts. Cells were mounted on slides using Pro-Long antifade solution (Molecular Probes). Double antibody labeling experiments were analyzed on a Zeiss LSM 410 inverted confocal microscope (Carl Zeiss, Thornwood, NY) as described previously (16). The fluorescein signal was imaged by exciting the sample with the 488 nm line from an argon or an argon/krypton laser and the resulting fluorescence was collected on a photomulti-plier after passage through the FT510, FT560 and BP515–540 filter sets. Likewise, the same field was excited with a helium/ neon (543 nm line) laser and the Texas Red signal was imaged on a second photomultiplier after passage through the FT510,
FT560 and LP590 filter sets. The green and red images were overlaid and pseudo-coloured using built-in LSM software. Images were obtained with 25×/0.8 or 63×/1.4 plan-Apochro-mat (Zeiss) oil objectives and printed on a Kodak XLS8300 color printer.
S.A.I. is a recipient of a Montreal Children's Hospital Research Institute post-doctoral fellowship. C.M. is a recipient of an Eileen Peters McGill Major Fellowship. J.M.T. is a Medical Research Council of Canada Scientsist and scholar of the Fonds de la Recherche en Santé du Québec. These studies were supported by the Medical Research Council of Canada.