|Home | About | Journals | Submit | Contact Us | Français|
GM1 gangliosidosis is an inherited, fatal neurodegenerative disease caused by deficiency of lysosomal β-D-galactosidase (EC 18.104.22.168) and consequent storage of undegraded GM1 ganglioside. To characterize the genetic mutation responsible for feline GM1 gangliosidosis, the normal sequence of feline β-galactosidase cDNA first was defined. The feline β-galactosidase open reading frame is 2010 base pairs, producing a protein of 669 amino acids. The putative signal sequence consists of amino acids 1–24 of the β-galactosidase precursor protein, which contains seven potential N-linked glycosylation sites, as in the human protein. Overall sequence homology between feline and human β-galactosidase is 74% for the open reading frame and 82% for the amino acid sequence. After normal β-galactosidase was sequenced, the mutation responsible for feline GM1 gangliosidosis was defined as a G to C substitution at position 1448 of the open reading frame, resulting in an amino acid substitution at arginine 483, known to cause GM1 gangliosidosis in humans. Feline β-galactosidase messenger RNA levels were normal in cerebral cortex, as determined by quantitative RT-PCR assays. Although enzymatic activity is severely reduced by the mutation, a full-length feline β-galactosidase cDNA restored activity in transfected GM1 fibroblasts to 18-times normal. β-Galactosidase protein levels in GM1 tissues were normal on Western blots, but immunofluorescence analysis demonstrated that the majority of mutant β-galactosidase protein did not reach the lysosome. Additionally, GM1 cat fibroblasts demonstrated increased expression of glucose-related protein 78/BiP and protein disulfide isomerase, suggesting that the unfolded protein response plays a role in pathogenesis of feline GM1 gangliosidosis.
The lysosomal enzyme β-D-galactosidase (βgal, EC 22.214.171.124) cleaves terminal galactose residues from a variety of molecules, including gangliosides GA1 and GM1. Deficiency of βgal is known to cause two lysosomal storage diseases: GM1 gangliosidosis (neuronopathic) and Morquio B Disease (mucopolysaccharidosis IVB, non-neuronopathic). GM1 gangliosidosis demonstrates varying degrees of clinical severity but is invariably fatal, and children with the most common and severe form of GM1 gangliosidosis usually die within 3 years of birth . Although various experimental strategies have been contemplated or are under development for GM1 gangliosidosis, the blood–brain barrier seriously limits otherwise promising treatments such as enzyme replacement therapy by excluding circulating lysosomal enzymes from the brain.
GLB1, the gene responsible for production of βgal, contains 16 exons and maps to human chromosome 3 and cat chromosome B3 . At least 59 mutations have been reported in GLB1 (Human Gene Mutation Database, http://www.hgmd.cf.ac.uk/ac/index.php). Mutation of nucleotide 1445 (G > A) results in substitution of arginine with histidine at amino acid 482 (Arg482His). The first reported mutation in Caucasians, Arg482His is carried by 8.3% of the population in Pelendri, Cyprus  and also has been reported in people of Japanese [4–6] or Maltese  descent. Previous studies determined that the G1445A mutation results in normal size and amount of GLB1 mRNA, although βgal protein may be slightly reduced on Western blots. When expressed singly in GM1 gangliosidosis fibroblasts or COS-1 cells, the Arg482His substitution produced a βgal protein with little to no residual activity using the 4-methylumbelliferyl-β-D-galactopyranoside synthetic substrate [3,4]. Although the in vivo biochemical effect of the Arg482His mutation often is difficult to discern because it occurs most frequently in compound heterozygotes, patients homozygous for the G1445A substitution present with the infantile (most severe) form of GM1 gangliosidosis [3,7,8]. A similar mutation, Arg482Cys, also produced no residual βgal activity after expression in GM1 gangliosidosis fibroblasts .
Feline GM1 gangliosidosis, first described in a Siamese cat in 1971 , models the juvenile form of the human disease. Onset of clinical neurological disease in affected cats occurs at approximately 3.5 months of age with a fine head or limb tremor. GM1 mutant cats have progressive dysmetria and ambulatory difficulties, with blindness and epileptiform seizures in the terminal disease stage at 9–10 months of age. In the current study, we identify an amino acid substitution analogous to Arg482His/Cys as the pathogenic mutation in feline GM1 gangliosidosis, which remains an important animal model for evaluation of translational therapeutic strategies. In addition to characterization of the mutation’s effect on βgal transcript and protein, we present evidence to support a potential mechanistic basis for the enzymatic deficiency in cats and humans expressing this mutation.
The feline βgal cDNA sequences generated by this study have been deposited in the GenBank database under Accession Nos. AF006749 (normal) and AF029974 (mutant).
The clinical, biochemical and pathological characteristics of feline GM1 gangliosidosis have been characterized extensively [9–24]. For laboratory studies, the humane endpoint is defined by weight loss >15% of maximum and the inability to enter a litter box of 3 in. in height for 3 consecutive days. Affected cats in the GM1 gangliosidosis research colony are humanely euthanized at 7.7 ± 0.8 (standard deviation) months. βgal enzyme activity is substantially reduced in cultured fibroblasts and brain homogenate . Obligate heterozygotes (parents of affected progeny) show no clinical signs of disease, but have reduced βgal enzyme activity (≈50% normal). All animal procedures are approved by the Auburn University Institutional Animal Care and Use Committee. Auburn University is accredited by the American Association for Assessment and Accreditation of Laboratory Animal Care.
Consensus PCR primers were designed from homologous regions of the human and mouse βgal (GLB1) genes [25,26]. βgal was sequenced first from normal cats to establish the normal feline sequence, then from mutant cats to identify the mutation. Sequence determination was performed by amplifying four overlapping segments of the feline GLB1 cDNA with Taq DNA polymerase (Applied Biosystems) using primers listed in Table 1. The 5′ and 3′ ends of the cDNA were sequenced using 5′ RACE and 3′ RACE systems (Invitrogen) with gene-specific primers 292n and 1252c, respectively. Once the mutation was defined from cDNA sequencing, it was confirmed by amplification and sequencing of genomic DNA with primers Ex14c and Int14n (Table 1), which amplify the mutation site in exon 14 of feline GLB1 (exon designation based on comparison to human and mouse GLB1 exon–intron boundaries) [27,28]. Because the mutation lies at the 3′ end of exon 14, primer Int14n anneals in intron 14 and was designed after partial sequencing of intron 14. Subsequently, diagnostic assays to identify the genotype of all colony cats were performed on genomic DNA with primers Ex14c and Int14n.
For cDNA production, tissue or primary skin fibroblasts were frozen in liquid nitrogen immediately after harvest. RNA was harvested from frozen tissue (1–5 mg) or cells (2 × 107) in Trizol Reagent (Invitrogen/Life Technologies) as recommended by the manufacturer. Reverse transcription was performed with an oligo(dT) primer and Superscript II reverse transcriptase (Life Technologies) at 50 °C for 1 h.
Genomic DNA was isolated from snap-frozen tissue with the Stratagene DNA extraction kit or from whole blood with the InstaGene Whole Blood Kit (Bio-Rad) as recommended by the manufacturers.
A full-length feline βgal cDNA was constructed with differing 5′ untranslated regions by RT-PCR of cerebral cortex mRNA isolated from a genotypically normal cat. A common non-coding strand primer (fBgal stop) was paired with one of two coding strand primers containing the feline βgal initiation codon only (fBgal ATG, construct 1.11) or the initiation codon plus 12 base pairs of the native feline 5′ untranslated region (fBgal 5′utr, construct 3.9) (Table 1). PCR reactions included 0.4 μM of each primer, 2–10 μl of cDNA, and 2.5 U of Pfu Turbo DNA Polymerase (Stratagene). Thirty cycles were performed of 95 °C for 30 s, 62 °C (fBgal ATG) or 58 °C (fBgal 5′utr) for 30 s, and 72 °C for 2 min. After gel extraction, A overhangs were added to the approximately 2.0 kb amplicons by a 30-min incubation with Taq DNA Polymerase at 72 °C. Amplicons were ligated by TA cloning into plasmid PCR3.1 (Invitrogen), which drives transgene expression with a cytomegalovirus promoter. Feline βgal clone sequences were verified by automated fluorescent DNA sequencing.
Primary fibroblasts from normal or GM1 cats were isolated by incubating minced skin biopsies for 3 days in Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum, 1× penicillin–streptomycin–amphotericin (D10 media, all components from Sigma) and 200 units/ml collagenase type Ia (Calbiochem). After plating the collagenase-digested skin pieces in D10 media alone, fibroblast colonies were trypsinized and replated. Primary fibroblasts were calcium-phosphate transfected with plasmid pSV3-DHFR (American Type Culture Collection), which contains the large T antigen of Simian Virus 40. Approximately 2 weeks after transfection, colonies of immortalized cells were evident in the tissue culture flasks and were trypsinized for further use .
Immortalized feline fibroblasts were calcium-phosphate transfected with plasmids 1.11 or 3.9 (described above), both of which contain a neomycin resistance gene. Transfected cells were selected by growth for 2 weeks in the neomycin analog, Geneticin (Invitrogen, 750 μg/ml active). Non-transfected control cells died after selection in Geneticin for 1 week. Cell lysates were prepared in 0.1% Triton X-100, and fluorogenic enzyme assays were performed as previously described .
Human fibroblast cell lines were obtained from a control patient (Wt) and two patients with GM1 gangliosidosis: H1, a homozygous non-sense mutation (Arg351X/Arg351X) resulting in neither βgal enzyme activity nor cross reactive material by Western blot; and H2, a compound heterozygote (Arg148Ser/Asp332Asn)  with less than 1% residual βgal activity and low but detectable levels of immunoreactive βgal protein. The cell lines were maintained at 37 °C in α-MEM media (Wisent Inc.) supplemented with antibiotics (penicillin/streptomycin, Gibco BRL) and 10% (v/v) fetal bovine serum in a humidified atmosphere with 5% CO2.
For analysis of βgal mRNA levels, six experiments were performed with a total of two age-matched, mutant–normal cat pairs. Data are expressed as means ± standard deviation (SD). Quantitative RT-PCR (qRT-PCR) assays for feline βgal mRNA were performed on an Applied Biosystems model 7700 sequence detector using primers 3a and 2b (Table 1). Because primer 3a anneals in exon 3, the qRT-PCR assay measures transcript levels of βgal only, not the elastin–laminin binding protein (a GLB1 splice variant lacking exons 3, 4 and 6). Total RNA was harvested from frozen tissue and reverse transcribed with oligo(dT) as described above. Quadruplicate 25 μl PCR reactions were performed with 1 × Sybr Green buffer, 3 mM MgCl2, 0.2 mM dNTP’s, 0.625 U Taq Gold, 0.25 U uracil-N-glycosylase (all from Applied Biosystems), 0.4 μM each primer and 0.5 μl cDNA template per reaction. After initial incubations of 2 min at 50 °C and 10 min at 95 °C to activate uracil-N-glycosylase and Taq Gold, respectively, 40 cycles of amplification were performed as follows: 30 s at 95 °C, 30 s at 60 °C and 45 s at 72 °C. βgal copy numbers from experimental samples were extrapolated from a standard curve generated with the same amplification conditions as above, with cDNA replaced by varying copy numbers (102–108) of plasmid fBgal 3.9 (cloned in our laboratory), which contains the putative open reading frame of feline βgal. Glyceraldehyde-3-phosphate dehydrogenase (EC 126.96.36.199, GAPD) expression was analyzed as above with an annealing temperature of 58 °C for primers fGAPD 101c (5′-CCTTCATTGACCTCAACTACAT-3′) and fGAPD 225n (5′-GAAGATGGTGATGGGCTTT-3′). A standard curve for GAPD amplification was generated with varying copy numbers (103–108) of plasmid fGAPD, which contains a partial feline GAPD cDNA cloned in our laboratory. To verify homogeneity of the amplicon generated from both βgal and GAPD PCR reactions, melting curve analysis of PCR product was performed on the Applied Biosystems model 7700 sequence detector with ABI Prism 7700 Dissociation Curve 1.0 software. PCR product was incubated for 15 s at 95 °C, 20 s at 60 °C, then 15 s at 95 °C. The ramp time between annealing (60 °C) and final denaturation (95 °C) was 19 min, 59 s. Additionally, automated fluorescent DNA sequencing was performed on selected samples of both βgal and GAPD PCR products to verify identity.
An immunogenic peptide for creation of a monoclonal antibody to feline lysosomal βgal was designed based on the published feline cDNA sequence (GenBank Accession No. AF006749). Peptide fBgal 148 (CGL RSS DPD YLA AVD K) includes amino acids 148–161 of feline βgal with CG at its amino terminus for conjugation of adjuvant. Peptide fBgal 148 is 100% homologous to human βgal and includes amino acids derived from the 3′ end of exon 4 and the 5′ end of exon 5. Peptide fBgal 148 bears no homology to the elastin–laminin binding protein, the alternatively spliced variant of GLB1 resulting from removal of exons 3, 4 and 6 with a frame shift in exon 5 . The peptide was synthesized at >95% purity, analyzed for net peptide content, and conjugated to keyhole limpet hemocyanin (KLH) by Global Peptide Services, LLC. Female Balb/c mice were immunized intra-muscularly with the following emulsification: 100 μg KLH-peptide, aluminum hydroxide gel adjuvant (Alhydrogel 85, Accurate Chemicals/Superfos Biosector, Denmark, equal volume of 0.4 mg/ml stock solution), 10 μg CpG-1826 [32,33] (5′-TCC ATG ACG TTC CTG ACG TT-3′, synthesized with a phosphorothioate backbone). Subsequent immunizations were performed as above but with 50 μg KLH-peptide. The final immunization was performed with 25 μg unconjugated peptide alone.
Isolation of spleen cells from immunized mice, fusion with myeloma cells and subsequent isolation of hybridoma clones were performed according to standard procedures by the Auburn University Hybridoma Facility. Hybridoma supernatants were screened by Western blotting for selection of the best antibody clone.
Western blots were performed with 25 μg of protein on a 10% SDS–polyacrylamide gel according to the method of Laemmli . Protein was transferred to nitrocellulose membranes (BioRad or Pierce), which were blocked in 5.0–7.5% non-fat dry milk and probed with hybridoma supernatant or a 1:6000 dilution (264 ng/ml) of mouse anti-rabbit GAPD (Chemicon). Secondary antibody (horseradish peroxidase-conjugated goat anti-mouse IgG/M) was diluted 1:120,000–1:180,000 and visualized with Supersignal West Dura chemiluminescent substrate (Pierce).
For quantification of protein levels, blots were scanned with a HP ScanJet 4570c flatbed scanner and digitized with Un-Scan-It gel Version 5.1 (Silk Scientific). The pixel density of each fBgal 148 band was normalized to the corresponding pixel density of GAPD. Normalized band intensities then were compared between normal and GM1 mutant animals. Tissue protein from two age-matched, mutant–normal pairs of cats was used for Western blotting. Four experiments were performed for kidney protein, and five experiments were performed for cerebral cortex.
Human and cat fibroblast cell lysates were obtained by repeated freezing and thawing in the presence of 0.1% sodium deoxy-taurocholate (w/w) followed by a 20 min centrifugation at 16,000g in a benchtop centrifuge at 4 °C. The reduced and denatured total proteins (40 μg) from fibroblast lysates were separated by SDS–PAGE as described above. Pre-stained high molecular weight protein standard was purchased from Invitrogen. Separated proteins were transferred to nitrocellulose (Whatman) for Western blot analysis using an ECL system (Amersham) as previously described . Primary antibodies were a goat polyclonal IgG anti-human GRP78 (BiP) (1:200, Santa Cruz Biotechnology Inc.) and a mouse monoclonal IgG1 anti-Protein Disulfide Isomerase (PDI) (1:1400, Stressgen). Membranes were stripped using a commercial solution (Pierce) between BiP and PDI labeling. Secondary antibodies were HRP-conjugated donkey anti-goat and donkey anti-mouse, respectively (1:5000, Jackson). GRP78/BiP and PDI run at apparent molecular masses of ~78 and ~58 kDa, respectively.
Indirect immunolabelling was performed using a protocol previously described  with small modifications. In short, cells were seeded onto 18 mm diameter coverslips for 16–20 h, then washed and fixed with 4% paraformaldehyde (EMS) in PBS, pH 7.2, for 20 min at 37 °C. Blocking and permeabilization was performed for 1 h at room temperature with SS–PBS (0.2% saponin (Sigma) and 10% either goat or horse normal serum (Wisent Inc.) in PBS). Coverslips were overlaid with primary and secondary antibodies (diluted in SS–PBS) for 1 h each at room temperature (in the dark for secondary antibodies). Nuclear staining was done with DAPI (Molecular Probes) at 1:50,000 in PBS, and coverslips were mounted onto glass slides with fluorescent mounting medium (DakoCytomation). Primary antibodies were as follows: rabbit polyclonal IgG anti-human βgal ; mouse monoclonal IgG1 anti-human LAMP-1 and LAMP-2 (Developmental Studies Hybridoma Bank, University of Iowa); mouse monoclonal IgG1 anti-Protein Disulfide Isomerase (Stressgen). Anti-human LAMP-2 was used for immunofluorescence with cat cells because the anti-human LAMP-1 antibody did not cross-react. Secondary antibodies were chicken anti-rabbit Alexa flour 488 for βgal and either chicken anti-goat or anti-mouse Alexa fluor 594 (Molecular Probes) at a 1:200 dilution in SS–PBS solution. Samples were analyzed using a Zeiss Axiovert confocal laser microscope equipped with a 63 × 1.4 numerical aperture Apochromat objective (Zeiss) and LSM 510 software. DAPI-stained nuclei were detected on the same system with a Chameleon two-photon laser. Confocal images were imported and contrast/brightness adjusted using Volocity 4 program (Improvision Inc.).
Statistical significance was analyzed using a one sample, two-tailed t-test  with the null hypothesis H0: μ = 1, where μ is the quotient defined by GM1 mutant βgal/normal βgal after both values were normalized to GAPD levels in the experimental samples. Therefore, if no difference in βgal levels exists between GM1 mutant and normal samples, then μ = 1.
The feline βgal cDNA was sequenced from normal cats and found to be very similar to human and mouse βgal sequences [25–28,39]. The feline open reading frame is 2010 base pairs (bp), compared to 2031 bp and 1941 bp for human and mouse, respectively. The overall sequence homology is 74% for cat versus human sequences and 72% for cat versus mouse sequences. (The human βgal open reading frame is 70% homologous to mouse.) Predicted protein length is 669 amino acids for feline βgal, compared to 677 amino acids for human βgal and 647 amino acids for mouse βgal, with an overall homology of 82% for cat versus human protein and 76% for cat versus mouse protein. (Human βgal protein is 80% homologous to mouse protein.) Based on sequence homology, the putative feline signal peptide (which is cleaved from the protein after facilitating translocation of the nascent protein into the endoplasmic reticulum) consists of amino acids 1–24, compared to amino acids 1–23 and 1–24 in human and mouse, respectively. Seven potential asparagine (N)-linked glycosylation sites occur at amino acids 27, 248, 429, 465, 499, 547 and 557, with six of seven sites found also in the human protein and five of seven sites found in the mouse protein.
When βgal cDNA from mutant cats was compared to the normal sequence, a guanine to cytosine substitution was detected at base 1448 of the open reading frame (G1448C), causing an arginine to proline substitution at amino acid 483 of the feline βgal protein (Arg483Pro). Human GM1 gangliosidosis is known to be caused by similar mutations, Arg482His [3,6,8] and Arg482Cys . The feline mutation resides in exon 14 of GLB1, as determined by sequence homology with the known exon–intron boundaries in human and mouse GLB1 [27,28].
To analyze a large number of cat samples for verification that the G1448C mutation is the pathogenic mutation in feline GM1 gangliosidosis, amplification of exon 14 was performed as described in Materials and methods section. By automated fluorescent DNA sequencing of the exon 14 amplicon from genomic DNA, 377 cats from this colony have been genotyped. To date, 136 phenotypically normal cats (36.1%) possess a G at position 1448, while an additional 180 phenotypically normal cats (47.7%) exhibit overlapping G and C peaks at position 1448, identifying them as heterozygotes for the mutant allele. The mutant allele has been identified in a homozygous state in 61 cats (16.2%), all of which developed stereotypical clinical signs of GM1 gangliosidosis. The homozygous G1448C mutation has never been detected in a phenotypically normal cat.
A full-length feline βgal cDNA was cloned from a genotypically normal cat into mammalian expression vector pCR3.1 (Invitrogen). As described in Materials and methods section, two separate full-length cDNAs were generated containing (1) the ATG translation initiation codon only or (2) 12 base pairs of the feline βgal 5′ untranslated region followed by the initiation codon. Transfection of the two feline βgal cDNAs into immortalized fibroblasts from a GM1 gangliosidosis cat restored enzymatic activity to levels approximately 18-fold higher than fibroblasts from a normal cat and 200-fold higher than untransfected GM1 cells (Table 2). The construct that included the feline βgal 5′ untranslated region (3.9) produced the highest levels of enzymatic activity in GM1 fibroblasts (18-fold above normal), but the construct including only the initiation codon (1.11) still was able to generate normal levels of βgal activity.
To analyze the molecular consequences of the G1448C mutation in feline GM1 gangliosidosis, quantitative RT-PCR assays were performed with primers 3a and 2b using cDNA isolated from cerebral cortex of normal and GM1 mutant cats. After normalization of data to GAPD, feline βgal mRNA levels in GM1 gangliosidosis cortex were not statistically different from normal (115.9 ± 15.7%, P > 0.05). Melting temperature analysis of the qRT-PCR product revealed a single dissociation peak, strongly suggesting amplification of a single PCR product. In addition, single-band PCR product was observed during gel purification of the qRT-PCR product, which demonstrated no contamination of the primary PCR product upon automated fluorescent DNA sequence analysis (data not shown).
To discern if cats with GM1 gangliosidosis produce steady-state levels of βgal protein, Western blots were performed with monoclonal antibody fBgal 148, which recognizes βgal from cat and other species such as human and cow. Anti-fBgal 148 does not recognize the elastin–laminin binding protein, an alternatively spliced variant of GLB1, due to lack of homology in the peptide sequence (see Materials and methods). In kidney, a 76 kDa band was evident at equal levels in normal and GM1 affected cats (Fig. 1). When normalized to GAPD protein levels, βgal in mutant kidney was 115.1 ± 29.2% of normal (mean ± SD), not a statistically significant difference (P > 0.20, n = 4). Similarly in cerebral cortex, cats exhibit a 63 kDa band of approximately equal intensity between normal and GM1 affected individuals. After normalization to GAPD, βgal in mutant cerebral cortex was 93.9 ± 17.4% of normal (P > 0.20, n = 5). Molecular weight differences in bands from cortex and kidney samples likely represent precursor versus mature forms of βgal.
To evaluate whether mutant βgal protein is correctly targeted to the lysosome, immunofluorescence was performed with a previously characterized polyclonal antibody to human βgal (anti-P-Gal) . In normal cat fibroblasts, βgal was correctly targeted to the lysosomes, as demonstrated by co-localization of βgal primarily with lysosome associated membrane protein (LAMP2). Very little co-localization of βgal and PDI was observed. However, mutant βgal in GM1 cat fibroblasts did not co-localize with LAMP2, indicating a non-lysosomal address for the mutant protein (Fig. 2). Based on cytological evidence (perinuclear staining with anti-P-Gal antibody) and faint co-localization with PDI, mutant βgal appears to be localized to the endoplasmic reticulum (ER). Similar subcellular addresses for βgal were documented in human fibroblasts, with the majority of wild-type βgal found in the lysosomes of normal cells and the majority of mutant βgal found in the ER of the H2 cell line, derived from a GM1 patient who was a compound heterozygote (Arg148Ser/Asp332Asn). The human H1 cell line, homozygous for a premature termination codon mutation (Arg351X), was included as a control and did not generate βgal cross-reactive material.
Subcellular localization of βgal also was studied in GM1 cat cells transfected with plasmids expressing normal (1.11) or supranormal (3.9) levels of wild-type βgal (Fig. 2). In 1.11 cells expressing normal βgal specific activity, the targeting defect was corrected, as demonstrated by substantial increases in co-localization of βgal and LAMP2. Interestingly, in 3.9 cells expressing supranormal levels of βgal specific activity, the majority of βgal protein (wild-type) co-localized with PDI, demonstrating an ER address.
Localization of a mutant lysosomal protein such as βgal to the ER suggests the possibility of ER stress, which could trigger the unfolded protein response (UPR). For an initial evaluation of the UPR in human and cat GM1 gangliosidosis, Western blots for GRP78/BiP or PDI were performed on fibroblast lysate from normal and affected cells (Fig. 3). Strong up-regulation of both GRP78/BiP and PDI was detected in GM1 gangliosidosis fibroblasts from human H2 cells (Arg148Ser/Asp332Asn) and from feline GM1 cells (Arg483Pro/Arg483Pro). Moderate increases in GRP78/BiP were documented in the human H1 cell line (Arg351X/Arg351X), although PDI levels were not increased.
GM1 gangliosidosis is a fatal, neurodegenerative lysosomal storage disease that, in its most common and severe form, causes the demise of human infants by 3 years of age . GM1 gangliosidosis also occurs in cats , dogs [40,41], sheep [42,43] and knockout mice [44–46], and these animal models are very valuable for disease characterization and therapeutic development.
In the current study, the genetic mutation for the well-characterized feline model of GM1 gangliosidosis was defined as G1448C, resulting in an Arg483Pro substitution. Messenger RNA levels of the βgal transcript were normal in the cerebral cortex of affected animals (115.9 ± 15.7%, P > 0.05). Because the upstream primer (3a) for the qRT-PCR assay annealed in exon 3, results reflect transcript levels for βgal only. No contribution is expected from the alternatively spliced variant of GLB1, the elastin–laminin binding protein, which lacks exons 3, 4 and 6 .
When 377 members of the feline GM1 breeding colony were tested for the mutant allele by a PCR-based diagnostic assay (see above), 36.1% were homozygous for the normal allele while only 16.2% were homozygous for the mutant allele, suggesting a detrimental effect of the G1448C mutation on fecundity. Similar observations have been reported in mouse models of lysosomal storage disease. For example, statistically significant abnormalities were detected in the sperm of mice heterozygous for knockout (KO) of the acid sphingomyelinase (ASM) gene. Although they produced normal numbers of sperm, heterozygotes demonstrated abnormal sperm morphology, mitochondrial membrane potential and ability to undergo capacitation or the acrosome reaction. Strikingly, when sperm from heterozygous ASMKO mice was used for in vitro fertilization of heterozygous oocytes followed by implantation into pseudopregnant wild-type mice, 37.8% of the resulting offspring were normal and 15.6% of the offspring were ASMKO affected mice . Also, mice heterozygous for the βgal knockout mutation produced offspring of which 18% were affected by GM1 gangliosidosis , and sperm abnormalities have been reported in the mouse models of Niemann-Pick Disease type C  and globoid cell leukodystrophy (Twitcher mouse) .
Although the analogous human mutation often occurs in GM1 patients who are compound heterozygotes, the Arg482His/Cys mutation in homozygosity results in a severe, infantile-onset disease phenotype [3,7,8]. Fibroblasts and leukocytes from patients homozygous for the Arg482His mutation express <1% of normal βgal activity, as with the mutant protein expressed in immortalized GM1 fibroblasts  or COS-1 cells . However, cats homozygous for the Arg483Pro mutation more closely resemble juvenile-onset GM1 gangliosidosis, with less severe clinical disease and no hepatosplenomegaly . Clinical disease onset in affected cats begins at approximately 3.5 months, and the humane endpoint is reached at 7.7 months. Less severe clinical disease is explained by substantial residual enzymatic activity in affected cats, as demonstrated in skin fibroblasts (9.2%, Table 2) and cerebral cortex (≤5%, data not shown). It is possible that the Arg483Pro mutation allows retention of residual enzymatic activity compared to the histidine or cysteine substitutions in mutant human βgal. This could be due to residual catalytic activity toward GM1 ganglioside and/or partial traffcking of mutant βgal to the lysosome (as suggested by faint co-localization of βgal and LAMP2 antibodies in Fig. 2). When a wild-type feline βgal cDNA is expressed at normal levels in GM1 cat fibroblasts, lysosomal localization of βgal is achieved. However, when wild-type feline βgal is expressed at high levels (18-fold above normal), the majority of the βgal protein is detected in the ER, probably en route to secretion from the cell. Because the βgal precursor is enzymatically active toward the synthetic 4MU substrate , specific activity assays of cell lysates measure total βgal activity (including βgal in the ER or secretory vesicles), not solely lysosomal βgal. Therefore, the current data does not permit determination of the amount of βgal that reaches the lysosome in overexpressing 3.9 cells (Fig. 2) or whether it would be sufficient to remove GM1 storage.
The UPR has been implicated in many disease processes [51–53], including lysosomal storage diseases such as Gaucher  and Fabry Disease [55,56], infantile neuronal ceroid lipofuscinosis (INCL) [57–59] and GM1 gangliosidosis . Specifically, in a knockout mouse model of GM1 gangliosidosis, GRP78/BiP was shown to be upregulated along with a variety of other markers for UPR activation. Although calcium dysregulation resulting from GM1 ganglioside accumulation was shown to induce the UPR in knockout mice, similar results were obtained from cultured feline fibroblasts (Fig. 2), in which GM1 synthesis is undetectable. This suggests that the UPR is induced by the presence of mutant βgal protein in the ER. Therefore, it is possible that at least two triggers for induction of the UPR exist in GM1 cats: (1) calcium dysregulation resulting from GM1 accumulation in the neuronal ER, and (2) accumulation of mutant βgal protein, which is not present in GM1 knockout mice. Of the two potential UPR triggers listed above, GM1-induced calcium dysregulation may be the most devastating, should it cause generalized protein misfolding and ER-associated degradation of otherwise normal proteins. However triggered, prolonged ER stress is known to induce UPR-mediated apoptotic pathways that effect neuronal death in neurodegenerative lysosomal diseases. For example, in the mouse model of INCL, brain elevations of several UPR markers accompanied increases in the apoptotic mediators, caspases 9, 12, 3 and 4 [57–59]. We hypothesize that a similar mechanism of neuronal apoptosis occurs in feline GM1 gangliosidosis, although a more precise elucidation of mechanistic pathways will require further investigation.
In summary, this report identifies the disease-causing mutation in feline GM1 gangliosidosis as Arg483Pro, similar to known pathogenic mutations of Arg482 in humans. Cerebral cortex and kidney tissues from GM1 cats produce normal amounts of βgal, as evidenced by Western blotting. However, the mutant βgal protein in feline and human GM1 cells is not effectively targeted to the lysosome but is retained in the ER. Continued definition of pathogenic pathways in the well-established feline disease model further enhances its utility for therapeutic research and is expected to stimulate innovative new therapies for GM1 gangliosidosis.
This study was supported by the Scott-Ritchey Research Center, College of Veterinary Medicine, Auburn University. The authors gratefully acknowledge the following contributors: (1) The LAMP-1 and -2 antibodies were obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by the University of Iowa, Department of Biological Sciences, Iowa City, IA 52242. (2) M. Woodside and P. Paroutis (Hospital for Sick Children Imaging Facility) for technical assistance with confocal microscopy.