Glb1 knockout mouse model shares natural history with Type II GM1 gangliosidosis patients

Abstract

GM1 gangliosidosis is a rare lysosomal storage disorder affecting multiple organ systems, primarily the central nervous system, and is caused by functional deficiency of β-galactosidase (GLB1). Using CRISPR/Cas9 genome editing, we generated a mouse model to evaluate characteristics of the disease in comparison to GM1 gangliosidosis patients. Our Glb1^−/− mice contain small deletions in exons 2 and 6, producing a null allele. Longevity is approximately 50 weeks and studies demonstrated that female Glb1^−/− mice die six weeks earlier than male Glb1^−/− mice. Gait analyses showed progressive abnormalities including abnormal foot placement, decreased stride length and increased stance width, comparable with what is observed in type II GM1 gangliosidosis patients. Furthermore, Glb1^−/− mice show loss of motor skills by 20 weeks assessed by adhesive dot, hanging wire, and inverted grid tests, and deterioration of motor coordination by 32 weeks of age when evaluated by rotarod testing. Brain MRI showed progressive cerebellar atrophy in Glb1^−/− mice suchlike seen in some patients. In addition, Glb1^−/− mice also show significantly increased levels of a novel pentasaccharide biomarker in urine and plasma which we also observed in GM1 gangliosidosis patients. Glb1^−/− mice also exhibit accumulation of glycosphingolipids in the brain with increases in GM1 and GA1 beginning by 8 weeks. Surprisingly, despite being a null variant, this Glb1^−/− mouse most closely models the less severe type II disease and will guide the development of new therapies for patients with the disorder.

INTRODUCTION

The gangliosidoses are lysosomal storage disorders characterized by the pathological accumulation of complex sialic acid-containing glycosphingolipids known as gangliosides [1]. GM1 gangliosidosis is an autosomal recessive ultra-rare multisystem disorder with an estimated incidence of one in 100,00 to 300,000 [2]. GM1 gangliosidosis primarily affects the central nervous system (CNS) and is caused by functional deficiency of lysosomal β-galactosidase (β-gal) which cleaves the terminal galactose from sialyl-glycosphingolipid GM1-ganglioside (GM1). The consequent buildup of GM1 results in apoptosis of neurons through a process not completely characterized. GM1 gangliosidosis is typically grouped into three sub-types: infantile (type I/severe form; MIM #230500), late infantile to juvenile (type II/intermediate form; MIM #230600) and adult (type III/chronic form; MIM #230650) [3–6]. Disease onset and severity is, however, a continuum based on the amount of residual β-gal activity that is primarily determined by the specific variant(s) in GLB1 gene (NM_000404), from little or no detectable β-gal activity in type I to increasing levels of activity in types II and III.

Mice harboring mutations in Glb1 were produced in the late 1990s by homologous recombination and embryonic stem cell technology. The Glb1^tm1Adz [7] and Glb1^tm1Jmat [8, 9] null/knockout mouse models mimicked human GM1-gangliosidosis disease, lacked β-gal mRNA, and had diminished enzymatic activity with an accumulation of storage material in the tissues and periodic acid-Schiff-positive intracytoplasmic storage in the brain. These animals showed tremor, ataxia, abnormal gait [7] and progressive spastic diplegia [8, 9].

Gene editing using CRISPR/Cas9 provides a means of generating mouse models with increased efficiency and specificity compared with traditional knockout methods and allows for generation of knockout models by perturbing only a small fraction of genomic sequence. CRISPR/Cas9-mediated editing of mouse Glb1 provides an important tool for understanding and studying the pathogenesis of GM1 gangliosidosis. For example, a recent report describes a CRISPR/Cas9 gene editing mouse model of Glb1 that introduces a 20 bp mutation removing the catalytic nucleophile of β-gal (β-gal^−/−)[10]. This model was devoid of β-gal enzyme activity, had ganglioside storage and presented with severe neuromotor and neurocognitive dysfunction.

More recently, a model generated using transcription activator-like effector nucleases (TALEN), inserted a fragment of the lacZ gene into exon 15 of Glb1, produced new features of axonopathy and showed a reduction of membrane resistance useful for studying axon–glial interactions [11]. Additionally, Liu developed a knockin model of Glb1^G455R using CRISPR/Cas9 mutation that displays GM1 accumulation, neurological behavioral decline and microglia activation [12].

There are no fully effective therapies for GM1 gangliosidosis; however, substrate reduction with small molecules has been shown to reduce gangliosides in the brain of Glb1^tm1Adz mice [13, 14]. AAV-mediated gene delivery to the CNS has been shown to improve survival [15–17] and most recently Tsunogai et al reported amelioration of CNS abnormalities using ex vivo hematopoietic stem cell therapy (HSCT) with lentivirus containing Glb1 [9, 18].

We used CRISPR/Cas9 to generate a null Glb1 allele containing a 17bp deletion in exon 2 and a 28bp deletion in exon 6 two exons commonly harboring pathogenic variants in GM1 gangliosidosis patients. We extensively phenotyped this mouse model for survival, ganglioside composition and behavior including gait abnormalities. We also used magnetic resonance imaging (MRI) to measure brain volume, resembling phenotyping used in patients in our ongoing natural history study (02-HG-0107, NCT000299965). Surprisingly, we find that the natural history of this Glb1 KO mouse has important phenotypic features parallel to what is observed in patients with Type II disease.

MATERIAL AND METHODS

Generation of Type II GM1-gangliosidosis mouse model.

Glb1 mutant (B6J-GE(glb1)^{del exon 2,6/Tif}, Glb1^−/−) mice were generated by pronuclear co-injection of two sgRNAs specific for the loci of interest in exons 2 and 6 of Mus Musculus Glb1 (Figure 1. a.) and a Cas9 WT mRNA into mouse zygotes.

Figure 1. Generation of CRISPR/Cas9 Mouse Model.

a. Mus musculus Glb1 gene diagram to develop β-gal deficiency mouse model of Type II GM1 gangliosidosis diseases applying CRISPR/Cas9 genome editing by utilizing two sgRNAs in exons 2 and 6, common sites of mutations in Type II GM1 gangliosidosis patients; The left white bar represents the first 5’ UTR, and the right white bar represents the 3’ UTR. b. and c. Sanger sequencing verification of CRISPR editing revealing a 17bp deletion in exon 2 and a 28bp deletion in exon 6 of Glb1^−/− mice. PCR amplification of exons 2 and 6, and Sanger sequencing were performed using genomic DNA collected from homozygous Glb1^−/− mice. d. Schematic of frameshift and predicted premature termination. e. Percentage of relative β-gal activity in Glb1^−/− tissues as compared to Glb1^+/+ control from 4 independent experiments, each with n=4 (mixed gender and ages) Glb1^−/− tissues compared to Glb1^+/+ control. f. Body weight (g) of female (n=4–5) and males (n=3–10) at 8-, 20- and 32-weeks old (two-tailed unpaired t-test with Welch’s correction). g. Glb1^−/− mice demonstrated decreased lifespan with disease progressing more rapidly in female mice progressing than in male mice. Mice were euthanized when they became moribund and unable to move around the cage and/or lost more than 15% of their body weight. Kaplan-Meier survival curve shows that Glb1^−/− female mice (42.7 weeks old, ±1.1, n=8) die approximately six weeks sooner than Glb1^−/− male mice (49.1 weeks old, ±1.1, n=11).

The exon 2 sgRNA, 5’ – GATACCCCGCTTCTACTGGG (AGG) – 3’ and the exon 6 sgRNA, 5’ – GCAGGACCTGTACGCCACAG (TGG) – 3’ (Table S1) were designed using CRISPRScan [19] and direct Cas9 to introduce the double-strand breaks. sgRNAs were purchased from Horizon at a concentration of 1ug/ul. The size and integrity of the sgRNAs were confirmed on a 2% agarose gel.

Pronuclear injection was performed using standard procedures [20] on fertilized eggs that were collected from superovulated C57BL/6J (Jackson Laboratories; abbreviated B6) females approximately 9 hours after mating with C57BL/6J males. Microinjections were performed using a capillary needle with a 1–2um opening pulled with a Sutter P-1000 micropipette puller. The pronucleus was injected using a FemtoJet 4i (Eppendorf) with continuous flow that we estimate to result in approximately 2pl of injection mix. Following visualization of pronuclear swelling, the needle was pulled out through the cytoplasm, likely resulting in a small amount of additional RNA delivery to the cytoplasm. 40 injected eggs were surgically transferred to pseudo pregnant CB6F1 hybrid recipient females (bred from a cross of Balb/cJ females with C57Bl/6J males). There were 16 pups born, 4 were found dead and 6 were positive for Glb1 mutations. Specific RNA and DNA concentrations for each injection session are provided (Table S1).

Initial genotyping of exons 2 and 6 was performed on founder animals and progenies using Sanger sequencing of genomic DNA extracted from tails. Tail snips from 10- to 15-day-old mice were obtained, and genomic DNA was extracted using the Extract-N-Amp Tissue PCR Kit (Sigma-Aldrich), as per manufacturer’s instructions. Amplifications of the regions of interest were carried out by PCR using the following primers:

5’ - CCCTTACTTGCCCTGCAGAA - 3’ and 5’ - GTGTGTGTAGGGGGAAAGGG - 3’ for exon 2 and 5’ -CTGCTGATCTCTGGTCCTCCTT - 3’ and 5’ - TCTAGATGCTACCTACACACACC - 3’ for exon 6 (Table S1). Cycling parameters were as follows: 95°C for 3 min, 40 cycles of (95°C for 15s, 62°C for 15s, 72°C for 15s), 72°C for 5 min. Reactions were treated with ExoSAP-IT (ThermoFisher) and directly sequenced using one of either of the primers above for the respective exon. PCR product from deletion-positive DNA was cloned into PCR4-TOPO vector (ThermoFisher), as per manufacturer’s instructions. Ten clones from each were sequenced using one of either of the primers from each set above. Founder mice were bred to C57Bl/6J. Biallelic mutant mice do not breed, hence, heterozygous F₁ mice from the F₀ crossed with C57Bl/6J mice were used to establish the mouse colony. Homozygous Glb1^−/−, control (Glb1^+/+) and Glb1 heterozygous mice (Glb1^+/−) were generated from heterozygote breeding of twice-backcrossed animals.

Animals Husbandry and Procedures

Type II GM1-gangliosidosis mouse model, Glb1 mutant (B6J-GE(glb1)^{del exon 2,6/Tif}, Glb1^−/−), control (Glb1^+/+) and Glb1 heterozygous mice (Glb1^+/−) were generated from heterozygote breeding. Genotyping on the established colony was performed as using primers in exon 2 (Table S1).

All mice had available ad libitum water and food and were maintained under a standard 12h light/12h dark cycle. All procedures were performed according to NIH guidelines, and in accordance to the Assessment and Accreditation of Laboratory Animal Care (AAALAC) guidelines in an approved facility. All procedures were approved by the NHGRI Institutional Animal Care and Use Committee (ACUC) under protocol G-11-4 (Mouse Models of Glycosphingolipid Storage Disorders: Understanding Disease Pathogenesis Provides Clues to Therapeutic Options).

The humane endpoint was defined by either weight loss of >15% of body weight at the highest weight that the mouse reached, or paralysis of the back limbs. Animals were euthanized either by CO₂ exposure followed by neck dislocation, or by overdose of 0.25% Avertin (2,2,2-Tribromoethanol), followed by transcardial perfusion with PBS only or followed by 4% paraformaldehyde and tissue collection.

Human Subjects

Patients were enrolled in the National Institutes of Health (NIH) under clinical protocol 02-HG-0107, “Natural History of Glycosphingolipid and Glycoprotein Lysosomal Storage Disorders” approved by the National Human Genome Research Institute (NHGRI) Institutional Review Board. For all patients, written informed consent was obtained from the parents.

Animal brain MRI and Analysis

High resolution MRI of excised mouse brains were performed in a 14T /40 micro imaging system (Bruker Inc., Billerica, MA) operating on a Paravision 360 platform. The excised brains, fixed in a 0.2X10⁻³ % Magnevist (Bayer) and 4% PFA mixture for 48 hours, were removed from the fixative, surface moisture minimized, and placed in a 15 mm NMR tube. A non-protonated oil, Fluorinert (3M), was added to submerge the brain to minimize magnetic susceptibility artifacts. The sample was then centered in a 15 mm transmit receive coil (Bruker Inc. Billerica, MA), the coil performance and magnetic field within the sample optimized, and three orthogonal scout scans were acquired. These scout scans were used to specify a 3-dimensional (3D) volume, encompassing the whole brain, and two sets of T₁ weighted images (TR= 1500 ms, TE = 5 and 10 ms, isotropic resolution = 50 mm) were acquired. Images were collected at three timepoints: 8, 20 and 32 weeks, corresponding to pre-symptomatic, symptomatic and end stage disease, respectively. Raw data were processed offline using the software packages ImageJ (NIH) and ITK-SNAP (itksnap.org). Regions of interest (ROI) were drawn, slice by slice, picking out areas of interest including white matter of the corpus callosum, cerebellum, hippocampus, basal ganglia and whole brain volume. The regions were summed for all slices and converted into volumes based on the slice thickness. The volume values were tabulated in Excel files and used for statistical comparison with GraphPad Prism 9.0.

Human MRI method and analysis

Brain MRI was performed using a Philips Achieva 3T system (Philips Healthcare, Best, the Netherlands) equipped with a 32-channel head coil (Philips Healthcare). 3D T1-weighted fast field echo imaging was performed with the following parameters: TR=11ms, TE=7ms, slice thickness=1mm, flip angle=6°, NEX=2. The estimated white matter in the corpus callosum was segmented consistently for all patients using AMIRA (V6.0.1, Thermo Fisher Scientific, Waltham, MA, USA) based on a semi-automated method of thresholding and manual segmentations on a slice-by-slice basis. The slice segmentations were then reconstructed into a 3D rendering to estimate the white matter volume in the corpus callosum region.

NP-HPLC quantification of ganglioside levels (GSLs)

HPLC profiles and quantification of ganglioside levels in cortex, cerebellum and the rest of the brain tissue (midbrain) from Glb1^+/+ and Glb1^−/− mice were performed using the method described by Neville and co-workers [21]. Briefly, GSLs were extracted and digested with ceramide glycanase. Released oligosaccharides were labelled with 2-aminobenzoic acid (2-AA) and analyzed using NP-HPLC. Data are presented as mean ± SD for n = 10 (5 males and 5 females) of 8-, 20- and 32-week-old mice (**** p < 0.0001, one-way ANOVA with Tukey multiple comparison test).

Mouse behavioral analysis

Animals were allowed to acclimatize to housing conditions for one week before any testing was carried out. Data were recorded at three timepoints: 8, 20 and 32 weeks.

Gait analysis (footprint ink test)

Forepaws and hind paws of Glb1^+/+ and Glb1^−/− were coated with non-toxic green and black ink, respectively, to evaluate placement of paws during walking. Mice were trained to travel a straight path to analyse their gait. The mouse was allowed to walk along a narrow, paper-covered corridor (open-top runway for mice: 50 cm long, 5 to 10 cm wide, with walls 5 to 10 cm high) in an enclosed goal box (20 cm square, with a 4 × 5–cm entrance hole), leaving a track of footprints. Once the footprints dried, measurements were taken. The gait parameters were modified from Lubjuhn et al. [22] and are indicated in Table S2 and Figure 2 b.

Figure 2. Gait analysis (Footprint ink test).

a. Representative footprint comparison of Glb1 mice. Glb1^+/+ and Glb1^−/− mice were trained to travel a straight path to analyze their gait. Forepaws and hind paws were coated with green and black ink respectively, to evaluate placement of paws during walking. Glb1^+/+ mice show an alternate gait pattern in which right and left limbs move separately in both the forelimbs and hind limbs. In contrast, Glb1^−/− mice have abnormal foot placement in the footprint ink test. b. Representation of gait parameters (modified from Lubjuhn et al.[22]). Paw angle was defined as the angle that paw makes with the long axis of the direction of motion of the animal; step angle, the angle between left and right hind paws as a function of stride length and stance width; stride length, the spatial length that a paw traverses through a stride; stance width, the perpendicular distance between the center point of either set of axil paws during the peak stance. c. Female (n=9–10) and d. Male (n=9–11) front paw angle (degrees). e. Female (n=9–10) and f. Male (n=9–11) hind paw angle (degrees) in Glb1^−/− is greatly reduced and axis of the front and hindfeet are turned inward towards center axis, and the Glb1^−/− mice also show extra placement/steps of forepaws. The Glb1^−/− footprints also appear less well defined than the Glb1^+/+. g. Female (n=9–10) and h. Male (n=9–11) forelimb stride length distance (cm). i. Female (n=9–10) and j. Male (n=9–11) hindlimb stride length distance (cm) in Glb1^−/− mice are significantly shorter than Glb1^+/+ mice. k. Female (n=9–10) and l. Male (n=9–11) step angle (degrees). Data are presented as mean ± SD of 8-, 20- and 32-week old mice (two-tailed unpaired t-test with Welch’s correction).

Inverted Grid

A non-invasive inverted grid suspension test was used to assess grip strength and coordination in mice. Mice were placed on a grid (35cm x 28cm), which was slowly turned over so they were suspended above a padded surface no higher than 3 feet and the latency to fall was timed with the maximum time at 180 seconds. No pain or discomfort was observed for this testing platform.

Rotarod testing

The rotarod test is widely used to assess the motor coordination, strength and balance of rodents and is especially sensitive in detecting cerebellar dysfunction. However, mice with GM1 gangliosidosis showed only mild motor deficit on the typical accelerating rotarod, which makes the rotarod a good test for motor coordination.

A Rotamex 5 rotarod from Columbus Instruments was used to assess motor coordination, balance, and equilibrium. The mouse was placed on the rod and the rotarod was accelerated gradually. Latencies for the mice to fall from the rod were recorded. A soft disposable pad was placed on top of a thick foam pad where the mice landed. The test was repeated 3 consecutive times. Below are more details about the rotarod test:

1. On Day 1, the acclimation phase was conducted which consisted of three trials separated by a minimum of 10 minutes, with each mouse being placed gently on the still rod for 60 seconds. The latency to fall and rearing, grooming, and turning behavior was noted.

2. On Day 2, the habituation phase was conducted, which consisted of three trials separated with 1-hour rests. Each mouse was placed on the rod which was rotating at a stable speed of 4 rotations per minute for 300 seconds (5 minutes). The latency to fall was recorded.

3. On Day 3, the test phase was conducted which consisted of three to five trials separated by 1-hour rests. Each mouse was placed on the rod, which accelerated from 4 to 40 revolutions per minute for 300 seconds (5 minutes). Latency to fall and speed at fall was recorded.

Neurological scores (Adhesive dot, hanging wire and tail elevation tests)

The SHIRPA score comprises 10 single tests that can be summarized in 5 different functional categories, giving a good overview about motor functions, neuropsychology, reflex and sensory, muscle tone strength and motor behavior of each animal. We performed non-invasive murine phenotyping tests, the modified SHIRPA behavioral and functional screening test battery. These standardized tests are designed to detect, by observational assessment, defects or abnormalities in gait, posture, motor control or coordination; changes in excitability or aggression, salivation, lacrimation, piloerection, defecation; and manifestation of stereotyped or abnormal behavior such as circling, retropulsion, etc. The screen includes the following: home cage observation for body position, tremor, aggression, or any abnormal behavior; a gross physical examination to include visual inspection of fur coat, eyes, ears, teeth, nares, and rectum; measurement of body weight and length; a general behavioral assessment in an arena cage for transfer arousal, gait, tail elevation, excessive running, grooming, freezing, rearing or jumping; sensorimotor tests to detect an approaching object, eye blink, ear twitch, touch escape and auditory cue; and a sequence of manipulations to assess postural reflexes including cage lid and wire hanging tests, tail suspension, righting reflex, balance and trunk curl. Padded cushions were used as a base for all suspension tests, and no pain or discomfort was observed for any of these momentary non-invasive testing platforms. Another screening test for motor coordination and function was conducted to include the adhesive dot test (time to recognize and remove a small sticker lightly applied to its muzzle or head).

Enzyme assay

β-gal enzyme assays were performed as previously described [23]. Briefly, tissues were lysed with lysis buffer (T-Per protein extraction reagent, Thermofisher) containing proteinase inhibitor. Protein levels were determined using standard BCA protocol as recommended by the manufacturer instructions (Thermofisher). Total β-gal enzyme activity was measured with a synthetic fluorogenic substrate, 4-methylumbelliferyl-b-D-galactopyranoside (4MU), measuring its release at excitation 355 nm and emission 460 nm on a plate reader and normalized to total protein concentration.

Pentasaccharide quantification

Pentasaccharide levels in urine and plasma of Glb1 mice and in plasma, urine and CSF from GM1 patients were measured using liquid chromatography-tandem mass spectrometry (LC-MS/MS). Mouse plasma and urine, human plasma, CSF, and urine samples (50 μL) were aliquoted into 2 mL polypropylene tubes (VWR, West Chester, PA). To each tube 2-AA (30 mg/mL) and NaBH3CN (20 mg/mL) in methanol (200 μL) was added. The mixtures were vortexed for 3 min, heated at 80°C for 1 hour, and centrifuged for 10 min at 9400 g. The supernatants were transferred to 1.2 mL glass inserts (VWR, West Chester, PA) and dried with nitrogen flow at 50°C. The residues were partitioned between water (200 μL) and methyl tert-butyl ether (MTBE) (600 μL), and aqueous phases were transfer to new 1.2 mL glass inserts for LC-MS/MS assay on a Shimadzu Prominence HPLC system (Columbia, MD) coupled with a 6500QTRAP+ mass spectrometer (AB Sciex, Framingham, MA). The oligosaccharides were separated on an ACE C18 column (4.6 × 150 mm, 3 μm) (Mac-Mod Analytical, Chadds Ford) protected with a SecurityGuard C18 guard column (4 × 3 mm) (Phenomenex, Torrance, CA). The mobile phases consisted of 0.1% formic acid in water (solvent A) and 0.1% formic acid in methanol-acetonitrile (4:1) (solvent B), and the flow rate was 1 mL/min. The gradient was as follows: 0 to 10 min, 3 to 40% solvent B; 10 to 10.1 min, 40 to 100% solvent B; 10.1 to 11 min, 100% solvent B; 11 to 11.1 min, 100 to 3% solvent B; 11.1 to 13 min, 3% solvent B. The eluate was directed into the mass spectrometer for data acquisition from 8 to 11 min; elsewhere, eluate was sent to waste to minimize source contamination. The injection volume was 2 μL, and the total run-time was 13 min. The ESI source temperature was 500ºC; the electrospray voltage was 5500 V; the declustering potential was 75 V; the entrance and the collision exit potential were 10 and 10 V, respectively. The collision and curtain gas were set at medium and 30, respectively. Both desolvation gas and nebulizing gas were set at 35 L/min. The collision energies for multiple reaction monitoring transitions of m/z 1032.4 to 667.3 (quantifier for 2-AA derivatized H3N2a and H3N2b) and m/z 1032.4 to 343.1 (qualifier for 2-AA derivatized H3N2a and H3N2b) were 30 and 45 eV, respectively. The dwell time was set at 50 ms for both MRM transitions. Data were acquired and analyzed by Analyst software (version 1.6.3).

Statistical Analysis

All statistical analyses were performed with GraphPad Prism 9.0 (GraphPad, San Diego, CA). Two-tailed unpaired t-test with Welch’s correction, two-tailed unpaired Mann-Whitney t-test, or two-way ANOVA followed by Tukey post-hoc tests was used to compare multiple groups. Data are presented as mean ±SEM or SD as denoted by each figure legend with * p - value < 0.05, ** p - value < 0.01, *** p - value < 0.001,**** p - value < 0.0001.

RESULTS

Type II GM1-gangliosidosis mouse model: generation and appearance

To develop a β-gal-deficient mouse model, Glb1 mutant (B6J-GE(glb1)^{del exon 2,6/Tif}, Glb1^−/−) mice, we applied CRISPR/Cas9 genome editing and utilized one sgRNA each targeting exon 2 and exon 6 of the Mus musculus Glb1 gene, NM_009752, that encodes a total of 16 exons (Figure 1. a). The target exons were chosen based on common sites of pathogenic variants in Type II GM1 gangliosidosis patients [6].

Sanger sequencing verification of CRISPR/Cas9 editing demonstrated generation of a 17bp deletion in exon 2 and a 28bp deletion in exon 6 of Glb1. PCR amplifications of exons 2 and 6 for Sanger sequencing were performed using genomic DNA collected from homozygous Glb1^−/− mice (Figure 1. b and c). The 17bp deletion in exon 2, c.181_197del17; p.(Ile61GlyfsTer24), results in a frameshift and predicted premature termination within exon 3. (Figure 1. d). Glb1^−/− mice were essentially devoid of β-gal enzyme activity in all the tissues analyzed (0% brain, 0% heart and 2.16% liver) except in kidney which demonstrated 32.2% enzyme activity (Figure 1. e). Both wild-type and mutant Glb1 mice continued to gain weight throughout their lifespan, and there were no significant differences between the two genotypes for female or male mice (Figure 1. f).

Glb1^−/− mice did demonstrate a decreased lifespan with disease progressing more rapidly in female mice than in male mice. Mice were euthanized when they became moribund, unable to move around the cage and/or lost more than 15% of their body weight. The Kaplan-Meier survival curve shows that Glb1^−/− female mice live approximately six weeks less (42.7 weeks old, ±1.1, n=8) than mutant male mice (49.1 weeks old, ±1.1, n=11) (Figure 1. g).

Behavioral testing of Glb1^−/− mice: Neuromotor disease

To test the neuromotor disease we performed a series of behavioral assays in Glb1^−/− mice and littermate controls at 8-, 20- and 32-weeks. Glb1^−/− mice are asymptomatic at 8-weeks of age but by 20-weeks they have visible neurological dysfunction that worsen through the 32-week timepoint.

Gait analyses (Figure 2. a and b) revealed decreased average front and hind paw angles for both female and male Glb1^−/− mice (Figure 2. c, d, e and f). The paw angle is greatly reduced, the axis of the front and hindfeet show in-toeing, and Glb1^−/− mice also show extra placement/steps of forepaws. The Glb1^−/− footprints also appear less distinct than the Glb1^+/+ control mice suggesting that steps are more tentative and gait more unsteady (Figure 2. a). The step angle was significantly decreased in Glb1^−/− females starting at 20-weeks and for males at 32-weeks as compared with littermate controls (Figure 2. k and l). The fore- and hindlimb stride distance (Figure 2. g, h, i and j) were greatly reduced in females and males starting at 20-weeks displaying a significantly shorter length in Glb1^−/− mice than Glb1^+/+ mice with an increased step angle in both females and males starting 20-weeks (Figure 2. k and l).

Grip strength in Glb1^−/− mice was diminished beginning at 8 weeks as demonstrated by a faster latency to fall as compared with controls by inverted grid testing (Supplemental Figure 1 a and b). Motor coordination in females was diminished at all timepoints, as assessed by the rotarod test. Glb1^−/− male mice showed better coordination with impairment beginning at 32-weeks (Supplemental Figure 1 c and d). Additionally, using the adhesive dot test, we assessed the time that the Glb1^−/− mice were able to remove a sticker from their head, a measure of fine motor coordination and function. We observed a significant increase in time for both male and females as compared to controls by 20-weeks (Supplemental Figure 1. e and f). To examine grip strength, we performed the hanging wire test where both male and female mice demonstrated weakness from 20-weeks (Supplemental Figure 1. g and h). Glb1^−/− mice also showed increasing tail stiffness, a measure of gait instability, by 20 weeks of age as compared to littermate controls (Supplemental Figure 1. i and j).

Neurodegeneration in Glb1^−/− mice

Brain volume by magnetic resonance imaging (MRI) correlates with clinical disease progression in type II GM1 gangliosidosis in humans [24]. We have observed progressive thinning of the corpus callosum on T1-weighted images in several of our patients with serial imaging on the same scanner using the same imaging sequences (Figure 3. a and c). Most patients show progressive global atrophy under the same imaging conditions. To extend this observation to our mouse model, we sought to evaluate the age-dependent progressive decline in brain volume in Glb1^−/− mice by MRI. We found a rapid decline in the estimated white matter volume of the corpus callosum region and cerebellum in mice (Figure 3. b, d and e and Table S3), whereas hippocampal, basal ganglia and total brain volumes did not reach statistical significance (Supplemental Figure 2 a, b and c and Table S4). Moreover, thinning of the corpus callosum became markedly conspicuous in the Glb1^−/− mouse brain by 20-weeks and progressed by 32weeks of age. Atrophy in the cerebellum reached significance by 32-weeks (Figure 3. b, d and e). The calculated volumes (in mm³) of control versus Glb1^−/− mice are displayed in Figures 3 d and e, Table S3, S4 and Supplemental Figure 2 a, b and c at 8-, 20- and 32- weeks in Glb1^−/− and control mice. T1-weighted MRI demonstrates that the Glb1^−/− mice undergo progressive reduction in the estimated white matter volume of the corpus callosum and cerebellar volume throughout their lifespan.

Figure 3. Neurodegeneration in Glb1^−/−.

a. Representative sagittal T1 MRI images of GM1 patient at one year and eleven months old and four years and three months old. b. Representative sagittal T1 MRI images of Glb1^+/+ and Glb1^−/− mice. c. Estimated white matter volume of the corpus callosum region from T1 MRI images of late infantile and juvenile GM1 patients (mm³). d and e. ITK-Snap quantification of mouse T2 MRI images: d. estimated white matter in corpus callosum region and e. cerebellum volumes (mm³) for of 8-, 20- and 32-week-old mice (± SEM, two-tailed unpaired t-test with Welch’s correction).

Pentasaccharide biomarker accumulation in Glb1^−/− mice and GM1 patients

Our recent effort to discover biomarkers for GM1 gangliosidosis led to identification of 2 isomeric pentasaccharides, H3N2a and H3N2b, containing 2 acetylglucosamines, 2 mannoses, and a terminal β-galactose (Figure 4 a). Both H3N2a and H3N2b were significantly elevated in urine, plasma, and CSF from both juvenile and late infantile GM1 patients as compared with controls (Figure 4 b–g). We sought to validate these biomarkers in our mouse model. We measured H3N2a and H3Nb pentasaccharide levels in urine and plasma of Glb1^+/+ and Glb1^−/− mice and showed a highly significant increase in both biomarkers in both fluids. (Figure 4 h–k).

Figure 4. Increased pentasaccharide levels.

a. Structure of H3N2a and H3N2b pentasaccharides. Pentasaccharide levels in b. and c. urine, d. and e. plasma and f. and g. CSF samples from control and GM1 juvenile and late infantile patients; Data are presented as mean ± SD for n = 4–20 (CSF), n = 4–10 (plasma) and n = 4–10 (urine) of GM1 juvenile, late infantile and control samples (two-tailed unpaired Mann-Whitney t-test). Pentasaccharide levels in h. and i. urine and j. and k. plasma from Glb1^+/+ and Glb1^−/− mice. Data are presented as mean ± SD for n = 7–9 (urine) and n = 9–20 (plasma) of 8-, 20- and 32-week-old mice (two-tailed unpaired t-test with Welch’s correction).

Glycosphingolipid accumulation in Glb1^−/− mice

We have quantitatively analyzed the ganglioside content in cortex, cerebellum, and the rest of the brain tissue (midbrain) as well as in liver and kidney, from Glb1^+/+ and Glb1^−/− mice using high performance liquid chromatography (HPLC, Figure 5 and Supplemental Figures 3–7). We found a highly significant increase in GA1, GM1a and total GSLs in cortex, cerebellum and midbrain both in female and male Glb1^−/− mice beginning at 8 weeks when the mice remain asymptomatic (Figure 5 a–f and Supplemental Figures 3–5). In cerebellum the levels of GM2Gc, a minor ganglioside species that is not a β-gal substrate, were significantly higher in both females and males (Supplemental Figure 4) in Glb1^−/− mice as compared with controls. This may be secondary lipid accumulation as a result of a more generalized lysosomal degradative impairment.

Figure 5. Glycosphingolipid accumulation in Glb1^−/− mice.

Quantification of ganglioside levels in cortex, cerebellum and midbrain from Glb1^+/+ and Glb1^−/− mice. a. HPLC profile trace of cortex glycosphingolipids from a representative control at 32-weeks of age and Glb1^−/− female mouse at 8-, 20-, 32- weeks of age. b. Cortex GA1 and GM1a levels. c. HPLC profile trace of cerebellum glycosphingolipids from a representative control at 32-weeks of age and Glb1^−/− female mouse at 8-, 20-, 32- weeks of age. d. Cerebellum GA1 and GM1a levels. e. HPLC profile trace of midbrain glycosphingolipids from a representative control at 32-weeks of age and Glb1^−/− female mouse at 8-, 20-, 32- weeks of age. f. Midbrain GA1 and GM1a levels. GSLs were extracted and digested with ceramide glycanase. Released oligosaccharides were 2-AA-labelled and analyzed using NP-HPLC. Data are presented as mean ± SD for n = 10 (5 males and 5 females) of 8-, 20- and 32-week-old mice (two-way ANOVA with Tukey multiple comparison test).

GM1 gangliosidosis is a systemic disease hence we measured the GSL levels in liver and kidney. We observed higher levels of GM1a and GM1b in liver from Glb1^−/− mice compared with controls (Supplemental Figure 6) as well as a significant increase in GA1 in kidney from 20-week-old for females and males and a significant increase of GM1aGc in kidney beginning at 8 weeks (Supplemental Figure 7).

DISCUSSION

Animal models are important tools in pre-clinical development of new therapies for rare diseases. We constructed a new mouse model for GM1 gangliosidosis, homozygous for a null allele, and detailed its neurological, behavioral, and biochemical phenotype.

The published Glb1 mouse models [7, 9–12] range in lifespan from 6 to 11 months and are not reported to have sex-specific endpoints. Interestingly, we saw a difference in lifespan between female and male mice of about 6 weeks (42.7 vs. 49.1 weeks, female and male respectively, Figure 1). The lifespan of our mice and the others containing a Glb1 null allele is longer than might be expected given the much shorter life span (~4.5 months) of the Hexb knockout mice [25–27], which are unable to degrade GM2 ganglioside, the immediate next step in the ganglioside degradation pathway (Figure 5 g). In humans, both GLB1 and HEXB null alleles cause very severe infantile disease [6, 28]. This suggests that an alternate pathway for limited degradation of GM1 ganglioside may be present in the mouse ameliorating the Glb1 knockout disease course similar to what has been reported for the Hexa knockout mouse [29–32].

Glb1^−/− mice exhibited lack of β-gal activity in all tested tissues (brain, heart, liver, spleen). Enzyme activity was, however, seen in the kidney, similar to that described in other models [7]. Other lysosomal β-galactosidases like galactosylceramidase have some activity towards the substrate 4-methylumbelliferyl β-galactoside that was used to measure the enzyme activity. This may contribute to the residual activity observed in the kidney using the artificial substrate [7, 33, 34]. That LacCer does not accumulate in kidney indicates that the lysosomal β-galactosidase which removes its terminal galactose is unaffected. The accumulation of GM1aGc and GA1 noted in the kidney (Supplemental Fig. 7 b) indicates a deficiency of Glb1 β-galactosidase activity.

We find that GM1 gangliosidosis mouse models with null alleles more closely recapitulate the phenotype of Type II disease patients by modeling our evaluation of the mice on parameters measured in human patients. Specifically, the lifespan of the animals is 43–50 weeks, or 50% of a normal murine lifespan. Patients with type II disease live into their early teens (2A), or into their 4^th decade (2B) [35]. Gait analysis shows a characteristic “in-toeing” in the mouse forepaws that is one of the first symptoms in Type II children (Tifft, personal observation) (Figure 2). Similar to reports in other mouse models, physical and behavioral evaluation of Glb1^−/− mice demonstrated onset of symptoms by 20 weeks of age, with deficits of neuromotor function, impaired balance and motor coordination as demonstrated by behavioral changes in the gait analysis (Figure 2) and the inverted grid and rotarod quantifications (Supplemental Figure 1).

Type II GM1 gangliosidosis patients show progressive global cerebral atrophy by MRI [24, 36, 37]. In addition to cortical and cerebellar atrophy, some patients demonstrate decreases in corpus callosal volume. This is the first study to utilize MRI in GM1 mice. Using ex vivo imaging, we observed atrophy of the corpus callosum from 20 weeks and decreases in cerebellar volume by 32 weeks. Similar in vivo studies on live mice are in progress. We posit that MRI could be a useful adjunct to histology in pre-clinical studies of new therapies for this devastating disease.

Both pentasaccharides H3N2a and H3N2b were elevated in urine, plasma and CSF in patients with GM1, and H3N2b is currently being used as an exploratory outcome measure for a gene therapy trial in GM1 patients (NCT03952637; IND18831). We also demonstrated that H3N2a and H3N2b were elevated in the urine and plasma of the Glb1^−/− mice and suggest that this will be a useful biomarker for further pre-clinical studies in GM1 gangliosidosis.

Gangliosides are the major glycoconjugates found in the nervous system [38]. The identification of GM1 and GA1 as the main storage material in GM1 gangliosidosis was discovered in postmortem brain tissue from a patient with infantile amaurotic idiocy [39] and proven later by O’Brien et al. as a functionally deficit GM1-β-gal in the catabolic pathway of GM1 [1]. There have been several other reports demonstrating GM1 and GA1 accumulation in postmortem brain tissue from GM1 gangliosidosis disease in patients [40–42] and animal models [7, 11]. We quantified GSLs using HPLC analysis and demonstrated elevations in GSL levels in Glb1^−/− brains, liver and kidney. This is one of the most comprehensive ganglioside composition analyses in a Glb1 knockout mouse model. We found that total GSLs in cortex, cerebellum and midbrain were highly elevated as are GA1 and GM1a in all the brain regions measured, including cortex, cerebellum and midbrain and GM2Gc in the cerebellum. In liver total GSLs, GM1a, GM1b and GM2Gc were highly elevated, whereas in kidney GA1 and GM1Gc were increased.

In summary, we have developed and extensively phenotyped a knockout mouse model of GM1 gangliosidosis that mirrors many of the features in patients with Type II disease including life span, progressive gait instability, and volume loss in the white matter of corpus callosum and cerebellum (Table 1). Indeed, several reported null Glb1^−/− models have a similar lifespan despite severe pathology and GM1 accumulation [6, 8–10] and more closely mirror Type II disease. These cumulative findings suggest that the murine disease may have additional pathways that ameliorate the phenotype when compared to the human disease as seen in the GM2 knockout mouse model [25, 27]. Our mouse model can become a powerful tool to better understand disease pathogenesis and to guide development of new therapies for patients with this devastating disorder.

Table 1.

Similarities between Glb1 knockout and GM1 gangliosidosis Type II patients

Glb1 knockout mouse	GM1 gangliosidosis Type II patients
Premature death (adults)	Premature death (teen and adults)
Progressive gait disturbance	Progressive gait disturbance
Loss of strength and coordination	Loss of strength and coordination
Cerebellar and corpus callosum atrophy	Global brain atrophy
H3N2a and H3N2b biomarker elevation (serum and urine)	H3N2a and H3N2b biomarker elevation (serum, urine and CSF)
Progressive elevation of GM1 gangliosides and its related charged glycosphingolipids	GM1 and GA1 elevation in postmortem nervous tissue [40–42]