Glycoconjugates play a key role in cellular function and are tightly regulated both in terms of biosynthesis and catabolism. β-Galactosidase (β-GAL) is a lysosomal hydrolase that cleaves β-linked galactose residues from the non-reducing end of glycan moieties found in various glycoconjugates. Reduction in β-GAL activity leads to the accumulation of GM1 ganglioside and its asialo derivative GA1, primarily in lysosomes of neuronal tissue (Caciotti et al., 2011; Jarnes Utz et al., 2017). The first description of GM1 and GA1 storage was made in a case of amaurotic idiocy (as it was then called), now called GM1 gangliosidosis (Jatzkewitz and Sandhoff, 1963). In addition to the storage of GM1 ganglioside, other glycoconjugates with β-galactose at the non-reducing end are detectable in high concentration in patients’ urine, including N-linked glycans, and various O-linked glycans (Lawrence et al., 2019). These disease-related β-linked galactose-terminal oligosaccharides arise from the lysosomal breakdown of glycoproteins that are stored in the brain (Tsay and Dawson, 1976), liver (Holmes and O’Brien, 1978), and other biological fluids including urine (Brunetti-Pierri and Scaglia, 2008; Bruggink et al., 2012) and amniotic fluid (Piraud et al., 2017). The significant proportion of these soluble glycans are metabolites of incompletely degraded N-linked glycans, such as A1G1, A2G2, A3G3, and A4G4. The N-glycan metabolite A2G2 was proposed to be a surrogate glycan biomarker of GM1 gangliosidosis (Lawrence et al., 2019).

GM1 gangliosidosis is an example of a family of inherited metabolic disorders termed lysosomal storage diseases. The estimated incidence of GM1 gangliosidosis is 1:100,000–200,000 live births (Brunetti-Pierri and Scaglia, 2008). However, some forms of the disease appear to be more prevalent in specific geographical areas, including Southern Brazil and Japan and among the Roma (Yoshida et al., 1991; Severini et al., 1999; Sinigerska et al., 2006).

GM1 presents with a continuum of disease severity but patients are loosely classified based on the age of onset of the symptoms into Type I (infantile), Type II (late-infantile and juvenile) and Type III (adult). In principle, disease severity should inversely correlate with residual enzyme activity levels. However, β-GAL activity is mostly measured with the synthetic fluorogenic substrate 4-methyl-umbelliferyl-β-D-galactopyranoside, which may not be accurate enough to estimate the clinical course of the disease on the basis of the measured residual enzyme activity (Inui et al., 1990; Caciotti et al., 2011; Ferreira and Gahl, 2017; Figure 1).

Bi-allelic mutations in GLB1 result in a reduction in β-GAL activity and the build-up of GM1 ganglioside in multiple tissues including the brain (Brunetti-Pierri and Scaglia, 2008; Karimzadeh et al., 2017; Rha et al., 2021) leading to severe neurodegeneration resulting in morbidity and premature mortality (Ferreira and Gahl, 2017). More than 200 disease-causing mutations have been identified across the GLB1 gene, particularly in exons 2, 6, 15, and 16 (Figure 2).

A smaller subset of mutations predominantly near the 3′ end of GLB1 can result in another lysosomal storage disease, Morquio B disease (mucopolysaccharidosis type IVB), which is characterized by primary lysosomal accumulation of the glycosaminoglycan keratan sulfate. Morquio B patients have predominantly skeletal abnormalities without the neurodegenerative aspects of GM1 gangliosidosis (Ferreira and Gahl, 2017). Cases with phenotypic features of both GM1 gangliosidosis and Morquio B disease have also been described (Mayer et al., 2009).

Distinct GLB1 gene mutations that affect the affinity of the enzyme for the two main substrates, GM1 ganglioside and keratan sulfate, explain at least in part the different clinical manifestations and the degree and type of storage material (Figure 2).

Five murine models have been developed that mimic the human phenotype and have formed that basis of our current understanding of disease pathogenesis. They have also been instrumental in the trialing of new therapies.

Comprehensive reviews of GM1 gangliosidosis have been recently published (Regier et al., 1993; Breiden and Sandhoff, 2019; Lawrence et al., 2019; Lang et al., 2020; Rha et al., 2021). In this mini review, we will focus on the clinical spectrum of disease and genetic variants, our current understanding of disease pathogenesis, the utility of available animal models and approaches to therapy for these devastating disorders.

Among the three subtypes of GM1 gangliosidosis, the infantile form is the most severe, with onset of symptoms before 6 months of age and death early in childhood (Jarnes Utz et al., 2017; Lang et al., 2020). One of the earliest signs that can be seen in infantile disease prenatally is hydrops fetalis, which is associated with many severe cases of LSDs (Stone and Sidransky, 1999), and should stimulate an investigation into the underlying causes of this phenotype (Iyer et al., 2021). The early diagnosis of infantile onset disease would have the greatest chance of successful therapeutic intervention.

Type II consists of late infantile and juvenile subtypes (Figure 1). Late infantile disease patients meet developmental milestones at 12 months but have onset of symptoms generally between 12 and 24 months. Patients quickly lose the ability to ambulate and have difficulty swallowing and handling secretions necessitating gastrostomy placement. Late infantile patients succumb to their illness by the mid second decade.

Juvenile patients have onset at 3–5 years of age, having learned to walk, run, and speak in sentences. The first symptoms may be unsteady gait, frequent falls, and “slurring” of speech progressing to ataxia and dysarthria. Children become unable to walk without assistance then lose the ability to walk. Dysarthria progresses to inability to swallow effectively and subsequent weight loss, necessitating gastrostomy tube placement. Children with juvenile onset disease can live into their 4th decade of life (Regier et al., 1993, 2016; Jarnes Utz et al., 2017).

Adult onset GM1 represents with an attenuated form with slower disease progression and very mild dysmorphic features (Suzuki et al., 2001) and greater clinical variability (Tonin et al., 2019). The onset of symptoms occurs in early childhood to late teens, as described in individuals of Japanese descent (Inui et al., 1990; Arash-Kaps et al., 2019).

Patients with all types of GM1 gangliosidosis can have variable skeletal disease including pectus carinatum with spine and rib changes that lead to restrictive lung disease (Ferreira et al., 2020). Radiographic changes can allow differentiation between the late-infantile and juvenile forms of Type II disease. Odontoid hypoplasia, as observed in all late infantile patients but not juvenile patients, requires cervical spine evaluation in late infantile patients prior to anesthesia, and anesthetic care optimally provided by pediatric anesthesiologists with experience in patients with cervical spine pathology to prevent perioperative morbidity or mortality (Ferreira et al., 2020).

The progressive nature of the neurological disease in GM1 gangliosidosis requires close monitoring. Biomarkers in blood, urine, and CSF biomarkers may be potentially useful in this regard (Gray-Edwards et al., 2017a,b; Rha et al., 2021). Disease progression can also be monitored using minimally invasive neuroimaging. In Type II disease both late infantile and juvenile patients demonstrate progressive atrophy in the cerebrum and cerebellum with greater variability and slower progression in the juvenile subtype. Quantitative magnetic resonance spectroscopy (MRS) shows increasing deficits of N-acetyl aspartate (NAA) across many brain regions with greater and swifter deficits seen in the more rapidly progressive late infantile subtype (Regier et al., 2016). These biomarkers of disease progression also correlate with developmental progression of disease and may serve as useful outcome measures for clinical trials.

In contrast to GM1 gangliosidosis, Morquio B disease (Figure 2) patients do not have CNS disease but can have neurologic compromise due to underlying skeletal disease, such as spinal nerve compression (Abumansour et al., 2020).

There is a poor genotype-phenotype correlation in GM1 gangliosidosis demonstrated by the clinical variability in age of onset and progression of disease even between siblings with the same genotype. One can speculate that polymorphisms or mutations in the other genes of the β-GAL complex, protective protein/cathepsin A (PPCA) and neuraminidase 1 (NEU1) (see Figure 1) may account for this variability. An additional limitation to reach an accurate genotype-phenotype correlation is posed by the way the enzyme assay is commonly performed using a soluble fluorogenic substrate that does not reflect the topology and membrane microenvironment of the natural substrate, GM1 ganglioside. Moreover, regulatory, and post-translational mechanisms that modulate GM1 catabolism further hamper an accurate prediction of the clinical course of the disease in patients, if based only on residual enzyme activity against the synthetic substrate (Breiden and Sandhoff, 2019).

Most patients with GM1 gangliosidosis are compound heterozygotes and aside from biallelic null mutations that produce type I disease, it is difficult to attribute specific phenotypes to any single mutation. Generalizations based on crystallographic structure of the β-GAL enzyme have been attempted (Ohto et al., 2012). Mutations associated with type I/infantile onset GM1 gangliosidosis, for the most part, are located in the core protein region causing β-gal instability, whereas mutations associated with milder phenotypes, such as types II and III GM1 gangliosidosis, tend to be on the protein surface (Morita et al., 2009; Ohto et al., 2012; Rha et al., 2021). Recently, the determination of the 3D structure of murine β-gal in complex with PPCA has revealed that some mutations at conserved amino acid residues found in GM1 gangliosidosis patients affect formation of the complex (Gorelik et al., 2021). These findings further complicate genotype-phenotype correlation, in relation to the penetrance of specific disease phenotypes.

Out of the total 261 reported pathogenic variants associated with a phenotype of GM1 gangliosidosis and/or Morquio B disease, most of them are missense/nonsense (194), and the rest are splicing substitutions (20), small deletions (25), small insertion/duplications (17), small indels (2), gross insertion/duplications (2), and a single large deletion (Figure 2). The largest number of mutations are found in exons 2 (26 variants), 6 (23 variants), 15 (21 variants) and 16 (24 variants). Previous reports implicate exons 2, 6, and 15 (Brunetti-Pierri and Scaglia, 2008) as hot spots for mutations, however exon 16 also harbors multiple pathogenic variants.

The first two genetically engineered mouse models of GM1 gangliosidosis were reported in 1997 by two groups (Hahn et al., 1997) and (Matsuda et al., 1997a,b). These knockout mice were generated by homologous recombination at the murine Glb1 locus that disrupted the gene by introducing a selectable marker cassette in either exon 6 or exon 15 (Hahn et al., 1997; Matsuda et al., 1997a,b). Both mouse models closely recapitulated the infantile/juvenile onset form of GM1 gangliosidosis and have been used extensively for studying disease pathogenesis and for testing therapeutic modalities (Table 1).

The β-gal^–/– mice described by Hahn et al. (1997) showed substantial early neuronal loss in the brain and spinal cord (Tessitore et al., 2004). GM1 is abundant in the neuronal plasma membrane (PM) and is the only ganglioside that can influence Ca²⁺⁺ transfer across membranes by interacting with Ca²⁺-binding proteins (Ledeen and Wu, 2015; Annunziata et al., 2018). In β-gal^–/– neurons impaired lysosomal degradation of GM1 results in the abnormal accumulation of the ganglioside in internal membranes, specifically those of the ER resulting in two pathogenic effects: (1) it enhances the flux of Ca²⁺ out of the ER, thereby altering ER Ca²⁺ levels, which activates an unfolded protein response (UPR) (Tessitore et al., 2004); and (2) it increases the number of membrane contact sites between the ER and mitochondria, known as mitochondria associated ER membranes (MAMs). These GM1-enriched microdomains mediate the abnormal flux of Ca²⁺ from the ER to the mitochondria, which ultimately results in mitochondria Ca²⁺ overload (Sano et al., 2009). The combination of these Ca²⁺-dependent pathogenic events steers the simultaneous activation of UPR and mitochondria-mediated neuronal apoptosis (Tessitore et al., 2004; Sano et al., 2009). In addition, the progressive neurodegeneration in these mice elicits a widespread neuroinflammatory response, accompanied by the release of cytokines and chemokines in the brain interstitial fluid and the CSF (Jeyakumar et al., 2003; Sano et al., 2005), which likely accelerates CNS disease. This neuroinflammatory response was shown to favor the recruitment of genetically modified BM monocytes expressing a therapeutic β-gal enzyme, following ex vivo gene therapy in β-gal^–/– mice (Sano et al., 2005) (see below).

With the development of new gene editing approaches, additional β-gal deficient mouse models have been more recently generated using either TALEN or CRISPR/Cas9 technologies (Przybilla et al., 2019; Eikelberg et al., 2020; Liu et al., 2021). Using CRISPR/Cas9 Przybilla et al. (2019) engineered a knock-out mouse model by introducing a deletion in exon 8 of the Glb1 gene. Phenotypic alterations in these mice were evaluated using behavioral tests that showed profound neurocognitive impairment (Przybilla et al., 2019). In Eikelberg et al. (2020) described a knock-out model using TALENs to target exon 15 of Glb1. These mice display brain and spinal cord pathology, characterized by swelling of axons and loss of myelin, leading to abnormal electrophysiological activity of neurons (Eikelberg et al., 2020). This is the first electrophysiological study performed in a mouse model of the disease, showing abnormalities that the authors attribute to increased neuronal cell size and reduced membrane resistance. The most recent β-gal mutant mouse generated using CRISPR/Cas9 is a knock-in model that introduces a human missense mutation in exon 14 of Glb1, described in a patient with late-infantile GM1 gangliosidosis (Liu et al., 2021). The CNS phenotype in this model included impaired motor function, as well as extensive microgliosis, accompanied by activation of autophagy (Liu et al., 2021). The characteristics of available mouse models are summarized in Table 1.

Until very recently, therapy for GM1 gangliosidosis was limited to symptomatic management. However, several experimental therapies have been trialed in murine (Table 1) and feline models (Gray-Edwards et al., 2017a,b, 2020). Since GM1 primarily affects the brain, targeted delivery must traverse the blood-brain barrier (BBB) or be delivered directly to the brain. Experimental therapies are discussed below.

GM1 gangliosidosis is a severe LSD underpinned by complex pathophysiological mechanisms. Numerous disease causing GLB1 mutations have been reported. However, genotype-phenotype correlation has been difficult to establish due in part to the way the enzyme is commonly assayed in patients’ samples. In addition, the clinical outcome of the disease in patients may be strongly influenced by post-translational and regulatory mechanisms controlling GM1 catabolism that may vary from patient to patient. These caveats ask for further elucidation of the cellular pathophysiology underlying this disease that may improve our understanding of the fundamental cell biology of GM1 ganglioside and the enzyme complex that regulates its catabolism in the lysosome. Multiple mouse models of this disorder have been instrumental for the pre-clinical testing of multiple therapies, several of which are currently in clinical trials.

E-RN, KS, FP, CT, Ad’A, and IA were involved in designing the concept of the review and oversight. KS and E-RN drafted the manuscript. All authors reviewed the manuscript, read, and approved the final manuscript.

Ad’A holds the Jewelers for Children Endowed Chair in Genetics and Gene Therapy. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.