Genetic deficiency of lysosomal acid [beta]-galactosidase ([beta]-gal) causes two phenotypically distinct disorders in humans, GM1-gangliosidosis and Morquio B disease (1 ). GM1-gangliosidosis is primarily a neurological disorder with progressive brain dysfunction. Clinically, infantile, juvenile and adult forms are recognized. The infantile disease is most rapidly progressive and involves not only the central nervous system (CNS), but also visceral organs with mucopolysaccharidosis-like features. Patients succumb to the disease usually within a few years. The visceral involvements are minor or absent in the older forms. The adult form is the most chronic, and radiologic bone abnormalities are often prominent. Morquio B disease occurs at varying ages and is characterized by severe bone deformities and an almost total lack of nervous system symptoms, except for those resulting from bone abnormalities, such as spinal cord compression. Patients with clinical manifestations that are intermediate between GM1-gangliosidosis and Morquio B disease have also been reported.
At the neuropathological level, severe infantile GM1-gangliosidosis patients exhibit distended neurons that contain typical lamellar inclusions referred to as membranous cytoplasmic bodies which are also found in other lipidoses (2 ,3 ). Although neurons are the primary target for storage, astrocytes may also appear abnormally vacuolated. Neuronal pathology in late onset forms is delayed and tends to be more severe in deeper structures of the brain than in the cortex (4 ). Inclusions in the liver are of fibrillar nature and are different from the lamellar bodies in neurons. Abnormal accumulation of GM1-ganglioside and, to a much lesser extent, its asialo-derivative GA1, in the brain is the most prominent biochemical feature (5 ,6 ). On the other hand, oligosaccharides derived from keratan sulfate and glycopeptides are stored primarily in visceral organs and are excreted abnormally into the urine (7 ). The human lysosomal acid [beta]-galactosidase cDNA and gene have been cloned and characterized (8 -11 ) and many disease-causing mutations have been identified in patients (12 -21 ).
Naturally occurring canine, feline, bovine and ovine models for GM1-gangliosidosis exist, but none is known among small laboratory animals (1 ). The canine and feline models appear to mimic the human phenotype to a large extent, but a faithful mouse model could facilitate study of the pathogenesis of the disease and its therapy. We have utilized gene targeting in embryonic stem (ES) cells to generate a mouse model of [beta]-galactosidase deficiency. The model appears to be a close duplicate of human GM1-gangliosidosis with respect to CNS involvement, since the mice present with pathological signs very early in life. This is reflected by the excessive accumulation of both GM1-ganglioside and GA1. In contrast to human infantile patients, the liver pathology is less apparent in the young mice. Although gross abnormalities are not visible during the first 4-5 months, the affected mice develop severe tremor, ataxia and an abnormal gait. A preliminary report from another laboratory has been presented in abstract form and describes the generation of a [beta]-gal-deficient mouse line by disrupting the gene further towards the 3' terminus. However, to the best of our knowledge, no details on phenotype pathology are available yet to compare with our model (22 ).
The [beta]-gal gene consists of 16 exons and was inactivated by introduction of a neomycin resistance gene into the middle of exon 6 (177 bp) (Fig. 1 a), which would ensure complete inactivation of the enzyme even if a truncated version was generated. A herpes simplex 1 virus thymidine kinase (tk) gene was inserted downstream of the gene to enable negative selection. Both selectable markers were placed in the same orientation as the [beta]-gal gene. Following electroporation into E14 ES cells and selection with G418 and FIAU, genomic DNA from double resistant colonies was screened with a 3' probe located outside the [beta]-gal gene targeting sequences. Eight independent homologous recombinant ES cell clones were obtained from a total of 325 examined, giving a targeting frequency of 1:40. Two of these (clones 68 and 190) were injected into blastocysts and gave rise to two [beta]-gal (-/-) mouse lines which show biochemical and pathological features that are essentially the same. At birth, the genotypic analysis (Fig. 1 b) of 117 newborn animals (from 15 litters of heterozygous crossings) indicated a Mendelian inheritance ratio among (-/-) (18%; n = 21), (+/-) (59%; n = 69) and (+/+) (23%; n = 27) mice. Thus, embryonic or fetal lethality did not occur. The [beta]-gal (-/-) mice are fertile and produce normal size litters.
Northern blot analysis of kidney, brain and liver total RNA detected no [beta]-gal mRNA of the expected 2.4 kb size in (-/-) mice, while heterozygotes contained approximately half the level of wild-type mRNA when standardized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA (Fig. 1 c). No smaller truncated version equivalent to the first five and a half exons of [beta]-gal was noted, suggesting that any generated mRNA was unstable.
Consistent with the absence of [beta]-gal mRNA in [beta]-gal (-/-) tissues, the residual activity in brain, kidney, liver and spleen using the fluorogenic artificial substrate was severely depressed. Activity levels ranged from 1% in spleen to 4% in brain compared with wild-type littermates (Fig. 1 d). Kidney samples consistently displayed ~8% activity. This residual activity towards 4-methylumbelliferyl [beta]-galactoside may be contributed by the other lysosomal [beta]-gal, galactosylceramidase, which is genetically normal in these mice and has some activity toward this substrate (23 ,24 ). Heterozygotes contained ~50% activity compared with normal littermates.
Since [beta]-gal is the first enzyme in the natural degradation pathway of GM1-ganglioside, it was important to assess whether there was any [beta]-gal activity towards the natural substrate in affected mice. The activity in the liver of a (-/-) mouse was undetectable compared with 85 nmol/h/mg measured in a wild-type mouse. This clearly indicates that there is no redundancy for the [beta]-gal-catalyzed degradation of gangliosides.
Brain tissue from affected mice was analyzed for any alteration in lipid composition during progression of the disease. Thin layer chromatography (TLC) of total brain gangliosides from 3-week- and 3.5-month-old (-/-) mice clearly demonstrated a marked increase in GM1-ganglioside, like that seen in an infantile patient (Fig. 2 ). Interestingly, a significantly higher level of GA1 occurred in affected mice compared with humans (5 ,6 ), suggesting that the murine neuraminidase is more active toward GM1-ganglioside than the corresponding human enzyme. No other significant abnormalities regarding content or distribution of other lipids, including phospholipids, galactolipids (galactocerebrosides and sulfatides) and cholesterol, were seen in the brain of the affected mice (unpublished data).
The gross appearance of the brain was normal. On histological preparations, swollen neurons containing storage materials throughout the brain were strongly stained with periodic acid Schiff (PAS) as early as 3 weeks of age. By 5 weeks, neuronal storage had increased dramatically and was noted in almost all neurons in the cerebrum, cerebellum, brainstem, spinal cord and dorsal root ganglia (Fig. 3 ). Storage was particularly conspicuous in large neurons of both the central and peripheral nervous system. In the cerebellum, in addition to the perikarya, PAS-positive material was also noted in the dendrites of Purkinje cells (Fig. 3 d). Electron microscopic examination of the cerebral cortical neurons at 3 weeks of age revealed pleomorphic inclusions consisting either of stacked or concentric lamellae, or of many small vesicles and complex lamellar structures (Fig. 3 f). The histopathology and ultrastructural features of the neuronal inclusions in these [beta]-gal (-/-) mice were closely similar to those of infantile GM1-gangliosidosis in humans (1 ).
We have established a mouse model of GM1-gangliosidosis through targeted disruption of the [beta]-gal gene. This disease has been reported in large domestic animals (reviewed in Suzuki et al., ref. 1 ). The canine and feline models appear particularly faithful to the human disease, and manifest many of the pathological and biochemical abnormalities seen in our murine model. These animals have been carefully studied and utilized for experimental therapy. While, to some extent, they may more closely parallel the conditions in humans for transplantation, they are technically more difficult and costly to maintain and treat. Therefore, the availability of a small laboratory animal should greatly facilitate studies on disease pathology and therapy.
[beta]-gal null mice show a complete absence of [beta]-gal mRNA and low residual enzyme activity for the fluorogenic substrate in all tissues tested. GM1-ganglioside is the primary substrate for the enzyme in the brain and no activity towards this compound is detected. As a consequence, there is rapid and progressive accumulation of GM1-ganglioside in the CNS of deficient mice which is obvious already at 3 weeks of age. In addition, the asialo derivative of GM1-ganglioside (GA1) accumulates to a similar level during progression of the disease. In the murine models of Tay-Sachs and Sandhoff disease (deficiencies of [beta]-hexosaminidase [alpha] and [beta] subunits, respectively) (25 ,26 ), Proia and colleagues have recently obtained evidence that desialylation of GM2-ganglioside to GA2 is a significant pathway in the mouse (26 ), unlike in humans (27 ). In the Tay-Sachs disease mouse, however, GA2 can be degraded further by the genetically normal [beta]-hexosaminidase B ([beta][beta]) isoenzyme, while its degradation is blocked in the Sandoff disease mouse. The distribution of undegraded GM1-ganglioside and GA1 in [beta]-gal (-/-) mice thus parallels that of GM2-ganglioside and GA2 in [beta]-hexosaminidase B (-/-) mice. Active desialylation of both GM1- and GM2-gangliosides has been reported in mouse Neuro2a cells (28 ), but interpretation was ambiguous due to the transformed nature of the cells. Our results indicated that in the mouse, degradation of GM1-ganglioside to GA1 is also a major pathway.
Contrary to GM1-gangliosidosis patients, the liver pathology in deficient mice is much less pronounced. The liver in infantile patients is enlarged and contains numerous foamy cells with abnormal fibrillar inclusions which differ from the lamellar bodies of neurons (1 ). The major storage products are oligosaccharides derived from keratan sulfate and glycopeptides. In [beta]-gal (-/-) mice, inclusions in the liver are not apparent even at the age of 3.5 months and storage of oligosaccharides is minimal. A possible explanation for this discrepancy is the presence of normal levels of the other [beta]-gal enzyme (galactosylceramidase) which, in mice, may compensate, at least in part, for [beta]-gal activity on some substrates. However, this seems unlikely since twitcher mice, deficient for galactosylceramidase, do not accumulate oligosaccharides (1 ), indicating that either this enzyme is not critical for catabolism of these hydrophilic substrates or that any activity towards these substrates can be fully compensated by [beta]-gal. Therefore, it is possible that the metabolism of these hydrophilic substrates differs between mice and humans. While this may be the case in visceral organs, there seems to be no compensating activity towards GM1-ganglioside, the primary [beta]-gal substrate in the CNS.
The nervous system involvement is the most predominant feature in the mouse model and is consistent with that seen in early infantile patients. The significantly higher level of GM1-ganglioside and GA1 in the brain of 3-week-old [beta]-gal (-/-) mice suggests that the CNS disease may already be present at birth. Accumulation of storage products is rapid and their distribution is very widespread, with almost all neurons being affected to some extent. One would expect there to be major perturbations in mechanosensory and psychointellectual pathways as the disease progresses. As evidence of this, some of the mice are starting to display overt neurological problems including tremor, ataxia and abnormal gait beyond the age of 5 months. Recently, an 8-month-old mouse died, apparently from hemiparesis. It will be crucial to monitor subtle motor, sensory and intellectual functions in a significant number of animals to define accurately the time of onset and type of neurological aberrations that result, and their lifespan.
Overall, this mouse model for GM1-gangliosidosis closely mimics the most fundamental aspects of the neuropathological and neurochemical abnormalities of the human disorder despite some distinct differences. In particular, because of the generalized CNS involvement, the [beta]-gal (-/-) model will provide an excellent system in which to correlate neuronal storage with physiological effects, and in which to investigate CNS-targeted therapy.
A mouse acid [beta]-gal cDNA (C57BL/6J) (29 ) was used as a probe to isolate the corresponding gene from a 129/SV mouse genomic [lambda] library (Stratagene, La Jolla, CA). One clone of ~15.9 kb encompassed exons 3-7 of the mouse [beta]-gal gene (Fig. 1 a). The targeting construct essentially contained a neomycin resistance (neo) gene inserted into a unique AatII site in exon 6 and a herpes simplex virus tk gene to enable positive/negative selection (30 ). To generate this, the 15.9 kb [lambda] clone was cut with AatII within exon 6 and the ends blunted. It was then digested with NotI (polylinker flanking site) and the 11.2 kb fragment containing exons 3 ,4, 5 and half of 6 was force-cloned into the pPNT (31 ) vector that had been cut with XhoI, blunted and then cut with NotI. The resultant construct was cut with BamHI, between the neo and HSV tk genes, and into it was inserted a 2.2 kb AatII-EcoRI fragment (downstream portion of exon 6 and part of intron 6) that was blunted and linked with BamHI linkers. The final targeting construct was linearized with NotI.
Electroporation into E14 ES cells and Southern blot screening for homologous recombinant clones were performed as described (32 ). Chimeric mice were generated by injection of two independent clones into C57BL/6 blastocysts as described (33 ). Heterozygotes were mated to obtain homozygotes [beta]-gal (-/-) mice.
Mouse tissues were homogenized in 4 volumes of water and [beta]-gal activity determined using an artificial 4-methylumbelliferyl [beta]-galactoside substrate (34 ). Total protein concentration was determined using bicinchoninic acid (35 ) following the manufacturer's protocol (Pierce Chemical Co.). The assay for acid [beta]-gal activity in total liver homogenates from the 3.5-month-old control and affected mice toward the natural substrate, GM1-ganglioside, was performed as previously described (36 ).
Brain and liver samples were obtained from three affected mice and three control littermates aged 3 weeks, 2.75 and 3.5 months, and from a human infantile GM1-gangliosidosis patient and one adult control subject. Lipid analyses were conducted on one cerebral hemisphere from each of the mice and on ~200 mg of human cerebral cortex. Lipid extraction, ganglioside isolation and lipid quantification were performed as described (37 ). Using chromatographic separation of total lipids on a 1 g silica gel 60 column, glycolipid GA1 eluted together with the ganglioside fraction. Thin-layer chromatograms were sprayed with resorcinol reagent and densitometric quantification of gangliosides was performed using a CAMAG II TLC scanner/CATS3 software. For visualization of gangliosides and GA1 on the same chromatograms, the orcinol spray which detects both lipids similarly was used instead of the sialic acid-specific resorcinol reagent.
The supernatant (50 000 g for 45 min) of 15% tissue homogenates were recentrifuged at 100 000 g for 2 h. Aliquots (5 µl) of the final supernatant were analyzed by TLC on high-performance silica gel 60 plates (Merck), developed for 3.5 h in n-butanol:acetic acid:water (3:3:2, by vol) and visually evaluated following spraying with the orcinol-sulfuric acid reagent (38 ).
Mice were anesthetized with Avertin and perfused through the left cardiac ventricle with 4% paraformaldehyde in 0.1 M sodium phosphate, pH 7.4, and fixed overnight in the same fixative. The brain, spinal cord including dorsal root ganglia and liver were removed. Some were processed for routine paraffin sections and 4 µm thick sections were stained with hematoxylin/eosin. Others were placed in 30% sucrose in 0.1 M sodium phosphate, pH 7.4 for 2 days, and then embedded in TBS tissue freezing medium by rapidly freezing in liquid nitrogen. Cryosections (8 µm) were cut and stained with PAS reagent.
For electron microscopy, following perfusion as described above, the brain and spinal cord were processed, embedded, sectioned and examined, as previously described (37 ).
We are grateful to Dr Gerard C. Grosveld for his continuous support, to Drs Xiao Yan Zhou and Jan van Deursen for their guidance in the initial stages of ES cell gene targeting, to Christie Nagy and John Swift for blastocyst injections and chimeric breeding, to Peggy Burdick for help in typing the manuscript and Sjozef van Baal for his computer skills. This work was supported by grants from the Assisi Foundation of Memphis [328502- (C.N.H.)], the National Institutes of Health [NIH RO1-NS24289, P30-HD03110 (Ku.S.); NIH RO1-NS24453 (Ki.S.)], the ELA (M.T.V.) and by the American, Lebanese, Syrian Associated Charities (ALSAC) of St. Jude Children's Research Hospital (A.d'A., M.P.M.).
+These authors contributed equally to this work
Human Molecular Genetics Pages
©
Introduction
Results
Generation of [beta]-gal (-/-) mice
Biochemical features of the CNS in young [beta]-gal (-/-) mice
GM1-ganglioside and GA1 accumulate in the brain of [beta]-gal (-/-) mice
[beta]-gal (-/-) mice display neuropathology consistent with GM1-gangliosidosis
Discussion
Materials And Methods
Targeting vector construct
Gene targeting in ES cells and generation of homozygous mice
[beta]-gal activity assays
Lipid analysis
Liver oligosaccharide analysis
Histological procedures
Acknowledgements
References
REFERENCES
This page is maintained by OUP admin. Last updated Fri Jan 31 11:45:11 GMT 1997. Part of the OUP Journals World Wide Web service.Copyright Oxford University Press, 1996