Small molecule inhibition of glycogen synthase I for the treatment of Pompe disease and other glycogen storage disorders

Abstract

Glycogen synthase 1 (GYS1), the rate-limiting enzyme in muscle glycogen synthesis, plays a central role in energy homeostasis and has been proposed as a therapeutic target in multiple glycogen storage diseases. Despite decades of investigation, there are no known potent, selective small molecule inhibitors of this enzyme. Here, we report the preclinical characterization of MZ-101, a small molecule that potently inhibits GYS1 in vitro and in vivo without inhibiting GYS2, a related isoform essential for synthesizing liver glycogen. Chronic treatment with MZ-101 depleted muscle glycogen and was well tolerated in mice. Pompe disease, a glycogen storage disease caused by mutations in acid alpha glucosidase (GAA), results in pathological accumulation of glycogen and consequent autophagolysosomal abnormalities, metabolic dysregulation and muscle atrophy. Enzyme replacement therapy (ERT) with recombinant GAA is the only approved treatment for Pompe disease, but it requires frequent infusions, and efficacy is limited by suboptimal skeletal muscle distribution. In a mouse model of Pompe disease, chronic oral administration of MZ-101 alone reduced glycogen buildup in skeletal muscle with comparable efficacy to ERT. In addition, treatment with MZ-101 in combination with ERT had an additive effect and could normalize muscle glycogen concentrations. Biochemical, metabolomic, and transcriptomic analyses of muscle tissue demonstrated that lowering of glycogen concentrations with MZ-101, alone or in combination with ERT, corrected the cellular pathology in this mouse model. These data suggest that substrate reduction therapy with GYS1 inhibition may be a promising therapeutic approach for Pompe disease and other glycogen storage diseases.

One Sentence Summary:

A small molecule inhibitor of glycogen synthase 1 reduces muscle glycogen and corrects cellular pathology in a mouse model of Pompe disease.

INTRODUCTION

Glycogen, a branched polymer of glucose molecules, provides a reservoir for the cellular storage of energy and carbon. Glycogen particles vary widely in size, consisting of up to about 50,000 tightly packed glucose monomers that can be released to support glucose homeostasis, muscle contraction and other metabolic processes (1). The synthesis and catabolism of glycogen are highly regulated in response to changes in nutrient supply and demand. Glycogenins (GYG1 and GYG2) prime synthesis by creating chains of ~10 glucose monomers (2). Glycogen synthase, the rate limiting enzyme for glycogen synthesis, extends these oligosaccharide chains through α-1,4 glycosidic linkages, and glycogen branching enzyme introduces α-1,6 glycosidic linkages (1). Higher eukaryotes express two isoforms of glycogen synthase, GYS1 and GYS2, with about 70% sequence identity and high structural homology (3). GYS1, or muscle glycogen synthase, is expressed throughout the body with highest abundance in skeletal muscle, whereas GYS2 is restricted in expression to the liver (2). GYS2 synthesizes glycogen in liver, which provides a reservoir to buffer circulating glucose levels, especially following fasting.

GYS1 has been the subject of intensive biochemical and structural studies over several decades (2). Two key modes of regulation have been identified for GYS1: allosteric activation by binding of glucose-6-phosphate (G6P) and inactivating phosphorylation at multiple residues (1). Several recent studies, including one from our group, have elucidated the high-resolution structure of human glycogen synthase in complex with glycogenin in the presence and absence of glucose-6-phosphate G6P and phosphorylation (4, 5). These studies defined a set of structurally distinct activated and inactivated states and raised the possibility of identifying small molecules that could interfere with these transitions and thereby inhibit the synthetic activity of the complex.

Inherited or acquired dysfunction of glycogen metabolism can cause disease in humans. Rare variants in the genes encoding glycogen metabolic enzymes can lead to more than twelve glycogen storage diseases (GSDs) (3). Aberrant glycogen metabolism has also been implicated in complex diseases such as diabetes, heart failure, and neurodegeneration (6-12). Furthermore, several studies have shown a role for glycogen in the proliferation and survival of cancer cells (13-15). Pharmacologic agents to modulate glycogen metabolism could therefore have applications across a range of diseases.

Pompe disease (glycogen storage disease type II) is a GSD caused by the pathological buildup of cellular glycogen due to loss of function mutations in the lysosomal enzyme acid alpha glucosidase (GAA) (16). GAA catabolizes lysosomal glycogen, and in the absence of GAA, glycogen builds up in these organelles, triggering a cellular disease cascade that begins with lysosome and autophagosome dysfunction and ultimately leads to cell death and muscle atrophy (17). Patients with Pompe disease have skeletal and cardiac muscle dysfunction that is associated with substantial morbidity and premature death (18). Patients with infantile-onset Pompe disease (IOPD) have little or no endogenous GAA activity and if left untreated, develop rapidly progressive cardiomyopathy, respiratory compromise, hypotonia, and failure to thrive. Patients with late onset Pompe disease (LOPD) have greater residual GAA activity and exhibit normal heart function but develop progressive muscle weakness and respiratory decline. Current standard-of-care treatment for patients with Pompe disease consists of enzyme replacement therapy (ERT) with human recombinant GAA (18). ERT treatment has been life saving for patients with IOPD by addressing the cardiomyopathy and slowing the rate of disease progression. However, for both patient populations the beneficial effects of ERT on skeletal muscle are more modest and generally plateau within a few years of treatment, after which the disease continues to progress (19).

An alternative therapeutic strategy aims to reduce glycogen synthesis by inhibiting GYS1 activity and preventing subsequent accumulation of glycogen in the lysosome. This concept of substrate reduction therapy (SRT) has been successfully applied to other genetic diseases characterized by toxic accumulation of metabolites, such as Gaucher disease, Niemann-Pick Type C disease and Fabry’s disease (20). Indeed, in mouse models of Pompe disease reduction of GYS1 expression by genetic knockout, small interfering RNA (siRNA) or peptide-conjugated phosphorodiamidate morpholino (PPMO) reduced glycogen accumulation in muscles (21-23). However, questions remain about the tolerability of chronic reduction in muscle glycogen. Several case reports describe humans with inactivating mutations in GYS1 who had complete loss of muscle glycogen and developed cardiomyopathy and cardiac arrhythmias (24, 25). Germline ablation of GYS1 in mice increases prenatal lethality with some evidence of skeletal and cardiac muscle abnormalities in surviving animals (26-29). Collectively, these findings suggest that SRT could be an alternative approach to treating GSDs but motivate continued investigation into the effects of chronic reduction in muscle glycogen.

To date, evaluation of therapeutic inhibition of GYS1 has been limited by the absence of highly selective small molecule inhibitors of GYS1 suitable for use in animals and humans. Here we described the in vitro and in vivo properties of MZ-101, a potent and selective small molecule inhibitor of GYS1. Treatment with MZ-101 reduced glycogen concentrations in cells and in mice. In a mouse model of Pompe disease, chronic administration of MZ-101 was well tolerated and reduced skeletal muscle glycogen stores to amounts comparable to enzyme replacement therapy (ERT); when MZ-101 and ERT were delivered in combination, glycogen stores in skeletal muscle were normalized. Using transcriptional and metabolic profiling, we found that GYS1 inhibition in combination with ERT could substantially correct the abnormal cellular pathways associated with Pompe disease. These preclinical data suggest that SRT via small molecule inhibition may be an effective therapy in patients with Pompe disease and other glycogen storage disorders.

RESULTS

MZ-101 is a potent and selective inhibitor of Glycogen Synthase 1 in vitro.

To discover potent and selective inhibitors of GYS1, we performed a high-throughput screen and medicinal chemistry campaign to optimize activity against recombinant GYS1 without inhibiting GYS2. We identified MZ-101 (Fig. 1A), which inhibits human GYS1 (IC50 0.041 μM) without inhibiting human GYS2 at concentrations up to 100 μM (Fig. 1, B and C). We confirmed similar potency and selectivity of MZ-101 for the mouse isoforms of GYS1 and GYS2 (fig. S1, A and B).

Fig. 1. MZ-101 is a potent and selective inhibitor of Glycogen Synthase 1 in vitro.

(A) MZ-101 structure. Inhibition potency of MZ-101 against human recombinant GYS1 (B) and GYS2 (C) in PK-LDH coupled enzyme assay (Exp n=253 GYS1, n=5 GYS2). Dose response of MZ-101 against phosphorylated (D) and fully dephosphorylated (E) human GYS1 across a range of G6P mM (n=4-6 exp). Data shown as mean +/− SEM.

The activity of GYS1 is regulated by two central mechanisms: activation by G6P and inhibition by phosphorylation (30, 31). Both the degree of GYS1 phosphorylation (GYS1-p) and abundance of G6P can vary in response to nutritional status, exercise, and disease pathology (32, 33). Although our initial compound screens were performed with GYS1-p in the presence of 0.5 mM G6P, we sought to identify a compound with potent inhibition across a range of potential cellular states. We assessed GYS1 activity with MZ-101 in the presence of varying concentrations of G6P as well as both fully phosphorylated (huGYS1-p) and fully dephosphorylated (huGYS1-dp) GYS1 (Fig. 1, D and E). Increasing the concentration of G6P from 0.15mM to 5mM (G6P Ka=0.311 mM; (34)) increased the IC₅₀ from 0.036 to 0.071 μM for huGYS1-p and from 0.117 to 0.202 μM for huGYS1-dp. These ~2-fold shifts in potency were consistent with a mechanism of inhibition that is non-competitive with G6P binding. Similar studies were performed to determine whether MZ-101 was a competitive inhibitor for the binding of substrate UDP-glucose. Increasing UDP-glucose induced a ~2-fold shift in MZ-101 inhibition potency, which was consistent with a non-competitive mechanism of inhibition (fig. S1, C and D). These data demonstrated that MZ-101 potently and selectively inhibits GYS1 by a non-competitive mechanism and maintains its activity across a range of physiologically relevant biochemical conditions.

MZ-101 inhibits glycogen accumulation in fibroblasts from healthy controls and patients with Pompe disease.

We next evaluated the ability of MZ-101 to inhibit glycogen synthesis and reduce glycogen storage in human cells. We obtained primary fibroblasts derived from three healthy human controls and three patients with IOPD, which were grown for 7 days in high glucose medium before they were switched to low energy (low glucose, low serum) culture medium in the presence or absence of MZ-101 for 7 days (Fig. 2A). Under these conditions, fibroblasts from patients with Pompe disease retained ~1.5-fold higher glycogen than those from healthy individuals (Fig. 2, B and C). Treatment with MZ-101 reduced glycogen abundance in a dosedependent manner in fibroblasts from healthy controls and patients with Pompe disease with a mean EC₅₀ of ~500nM (Fig. 2, B and C). However, glycogen abundance remained higher in cells from patients with Pompe disease, likely due to slower clearance of glycogen. We also observed similar potency for MZ-101 across a range of immortalized human, dog, and rat cell lines (Table S1). These data showed that MZ-101 inhibits glycogen synthesis across species and cell types expressing GYS1, independent of defects in glycogen catabolism.

Fig. 2. MZ-101 inhibits glycogen accumulation in fibroblasts from healthy controls and patients with Pompe.

(A) Schematic of experimental protocol. Potency of MZ-101 GYS1 inhibition EC₅₀ impact on accumulated total glycogen stores over 7 days treatment in healthy donor fibroblasts (B) and Pompe donor fibroblasts (C) with 1.0 corresponding to the average glycogen content in healthy control DMSO-treated cells. Data shown as mean +/− SEM, N=3 experiments.

Glycogen accumulation in Pompe (Gaatm1Rabn) mouse gastrocnemius drives defects in autophagolysosomal maturation and pathological upregulation of the GYS1 glycogen biosynthetic pathway.

To evaluate the relationship between the in vivo pathophysiology of Pompe disease and glycogen metabolism, we studied a mouse model of Pompe disease carrying a biallelic disruption of GAA (Gaatm1Rabn, hereafter termed GAA KO). As previously described (17, 35), these mice exhibit complete loss of GAA activity, dysregulated autophagolysosomes, and grossly elevated muscle glycogen by 12 weeks of age (Fig. 3A, and fig. S3A to D). Some previous studies have also reported increased glycogen synthetic activity in GAA KO mice(23, 35, 36). To characterize glycogen metabolism in muscle tissue, we quantified abundance of proteins and analytes involved in glucose uptake and glycogen synthesis. GAA KO mice displayed ~1.5-fold increase in protein abundance of the glucose transporter GLUT1, as well as ~2-fold elevated intracellular glucose, and ~5.0-fold increased G6P (Fig 3B) in gastrocnemius muscle. The elevation in G6P abundance was also observed in diaphragm and heart muscle, but not in the liver (fig. S3E). We then quantified GYS1 protein abundance, post-translational modification, and activity in muscle tissue lysates. Samples were analyzed by incorporation of radiolabeled UDP-glucose into glycogen to measure glycogen synthesis activity (Fig. 3C, fig. S3F) and via western blot (Fig. 3D, E) for protein level. In GAA KO mice, GYS1 protein abundance was unchanged, phosphorylation levels were increased, and activity was increased ~2-fold in the absence of exogenous G6P (from 3275 to 6357 GS cpm/mg/min, p=0.0108) (Fig. 3, C to E and fig. S3F). These data indicate that glycogen synthesis is increased in GAA KO mice despite substantially elevated levels of intracellular lysosomal glycogen. These data support a dual mechanism for glycogen accumulation in Pompe disease with contributions from both deficient clearance of glycogen and increased in glycogen synthesis (Fig 3F).

Fig. 3. Glycogen accumulation in Pompe (Gaatm1Rabn) mouse gastrocnemius drives pathological upregulation of the GYS1 glycogen biosynthetic pathway.

(A) Gastrocnemius muscle glycogen content in WT and Pompe GAA KO mice. (B) Markers of the glycogen biosynthetic pathway: glucose transporter GLUT1, free glucose, and glucose-6-phosphate (G6P) metabolite. (C) GYS1 de novo glycogen synthesis as quantified from gastrocnemius tissue lysates. (D) Total GYS1 protein abundance and abundance of inhibitory phosphorylation at S641 GYS1. (E) Representative Western blots from gastrocnemius tissue lysates. (F) Key cell biology pathways driving the pathophysiology in Pompe disease. Data are shown as mean +/− SEM. Students t-test ns p>0.05, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

MZ-101 inhibits GYS1 but not GYS2 de novo glycogen synthesis in WT and Pompe mice.

To further study the effect of MZ-101 on glycogen synthesis in vivo, we developed a high-sensitivity metabolic tracer assay in which we feed mice an oral bolus of ¹³C₆-glucose and monitor its incorporation into glycogen using liquid chromatography mass spectrometry (Fig. 4A). This assay enables evaluation of glycogen synthesis over several hours without contamination by background tissue glycogen levels. Because the oral glucose bolus induces a hyperglycemic and hyperinsulinemic state, glycogen synthesis likely predominates over glycogen catabolism, although both may contribute (6, 37). We first applied this protocol to measure differences in glycogen synthesis between wildtype and GAA KO mice. Over four hours following labelled glucose administration, GAA KO mice incorporated ~2-fold more ¹³C₆-glucose than wildtype mice in gastrocnemius muscle, consistent with elevated GYS1 activity observed in tissue lysates (Fig. 4B, Fig. 3D, right panel). We then used the metabolic tracer assay to quantify inhibition of glycogen synthesis by MZ-101 in mice. After a single oral dose, MZ-101 reduced incorporation of glucose into glycogen in skeletal and cardiac muscle of both WT and GAA KO mice a dose-dependent manner (Fig. 4, C and D) without affecting incorporation in the liver (Fig. 4E, fig S4, A to E). These data provided further evidence that glycogen synthesis was elevated in GAA KO mice and showed that MZ-101 could selectively inhibit glycogen synthesis in muscle tissue in vivo.

Fig. 4. MZ-101 inhibits GYS1 but not GYS2 de novo glycogen synthesis in WT & Pompe mice.

(A) Schematic of in vivo de novo glycogen synthesis experiment. (B) De novo glycogen synthesized over a 4-hour time course pulse in male WT (C57BL6) and Pompe (GAA KO) mouse gastrocnemius. (C) Inhibition of glycogen synthesis in gastrocnemius muscle via dose response of MZ-101 in WT mice (n=7-22). MZ-101 dose response in Pompe mice gastrocnemius (D) and liver (E), (n=8-12 mice per treatment group). Data are individual animals (circles) with mean +/− SEM. One way ANOVA with Dunnett’s post hoc analysis vs 0 mg/kg control group, ns p>0.05, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Substrate reduction therapy (SRT) with MZ-101 reduces glycogen in WT and Pompe mice over 4-14 weeks.

To deplete accumulated glycogen by SRT, the rate of glycogen catabolism must exceed the rate of glycogen synthesis. There are two well-established pathways for glycogen breakdown in cells: cytosolic catabolism initiated by glycogen phosphorylase and lysosomal hydrolysis by GAA (1). However, the kinetics of glycogen catabolism in the absence of GAA have not been measured in a live animal. We adapted the metabolic tracer assay to a pulse chase protocol with a bolus of ¹³C₆-glucose followed by an ad libitum diet containing unlabeled glucose. We then tracked the reduction in labeled glucose in muscle tissue over time to measure glycogen turnover. The half-life of glycogen in the gastrocnemius muscle of WT mice was about 16 hours compared to GAA KO mice in which turnover took several weeks, although more than 95% of labeled glycogen was eventually cleared (Fig. 5A). From these data we hypothesized that at least several weeks of treatment would be required to see substantial reduction of glycogen stores in GAA KO mice with GYS1 inhibition.

Fig. 5. Substrate reduction therapy (SRT) with MZ-101 reduces glycogen levels in WT and Pompe mice over 4-14 weeks.

(A) Glycogen turnover across 12 weeks in WT vs Pompe gastrocnemius. (B and C) 10-12 week old mice provided chow formulated with vehicle (veh.) or MZ-101 (SRT) for 4 or 14 weeks. Glycogen content in WT (B) and Pompe (C) mouse gastrocnemius after 4 and 14 weeks treatment. Data are individual animals (circles) with mean +/− SEM. % reduction calculated at each timepoint for each genotype. Unpaired Students t-Test WT vs Pompe per time point cohort ***p<0.001, ****p<0.0001.

To evaluate the impact of chronic inhibition of glycogen synthesis on glycogen abundance, WT and GAA KO mice were fed chow containing MZ-101 or vehicle over 4-14 weeks. Plasma concentrations of MZ-101 were similar across both genotypes and timepoints (fig. S5A). In wildtype mice, muscle glycogen decreased by 71% at 4 weeks with persistent reduction (81%) at 14 weeks (Fig. 5B). Chronic treatment was well tolerated with no changes in activity levels noted. In GAA KO mice, muscle glycogen decreased by 38% at 4 weeks with continued further reduction (to 58%) by 14 weeks (Fig. 5C). The glycogen abundance in GAA KO mice after 14 weeks of treatment remained ~3-fold greater than untreated WT animals. The delayed clearance of glycogen in GAA KO mice was consistent with the slower measured glycogen catabolism in the absence of GAA. These data demonstrated that chronic inhibition of glycogen synthesis can reduce the accumulation of glycogen, and that GAA accelerates this process.

Biobank analyses support the tolerability of chronic reduction in muscle glycogen abundance.

Inhibiting muscle glycogen synthesis may reduce the pathologic accumulation of glycogen in Pompe disease, but are there other consequences of chronic moderate reduction in muscle glycogen that would limit tolerability. Biobanks containing data from humans with naturally occurring variation in glycogen metabolism could provide an opportunity to understand the effects of lifelong lowering of muscle glycogen content. A common frameshift variant (AF ~1.1% in non-Finnish Europeans) has been observed in the gene encoding protein phosphatase PPP1R3A (PPP1R3A FS, rs527638422), which activates GYS1 and is expressed predominantly in muscle. A previous study used magnetic resonance spectroscopy to measure glycogen synthesis in PPP1R3A FS heterozygous carriers and found about a 65% reduction in muscle glycogen without changes in insulin or blood glucose (38). We reasoned that more comprehensive analysis of the phenotypes in these individuals could provide insights into the expected tolerability profile of lifelong reduction in muscle glycogen levels. We surveyed the UK Biobank and identified 3,561 heterozygous carriers of the PPP1R3A FS variant (carrier frequency 1%) and 9 bi-allelic homozygous carriers. We performed a phenome wide association study of the PPP1R3A FS variant and observed no statistically significant relationship with any disease diagnosis or ICD10 code (fig. S5B), fluid biomarker, or measure of heart structure or function (fig. S5C). Although the smaller number of homozygous PPP1R3A FS carriers limits the power to detect signals within this group, homozygous carriers were not outliers for any of the phenotypes which suggests that a greater level of chronic reduction in muscle glycogen may also be tolerated (fig. S5C). Collectively these data suggest that chronic reduction of glycogen stores by ~65% or higher could be well tolerated in humans.

SRT, ERT, and combination therapy reduce gastrocnemius glycogen and urine and blood biomarkers after 12 weeks of treatment.

Reduction of glycogen levels in Pompe disease could be achieved either by reducing glycogen synthesis, increasing its catabolism or both. We therefore sought to compare the effect of SRT to ERT in the GAA KO mice, and to characterize the effects of SRT in combination with ERT. For enzyme replacement, we used alglucosidase alfa, an approved compound for treatment of Pompe disease, at the labeled intravenous dose of 20 mg/kg biweekly. MZ-101 was delivered in chow diet and achieved mean steady state plasma exposures in the same range as the previous experiment (fig. S5A), which were not changed by combination with ERT (fig. S6A). Mice began treatment at age 6-9 weeks and continued for 12 weeks (Fig. 6A). SRT treatment alone significantly reduced gastrocnemius glycogen stores by 49% compared to vehicle treated GAA KO levels (p<0.0001); ERT alone reduced gastrocnemius glycogen by 26%, similar to published reports (Fig. 6B) (39-41). Combination of the two therapies resulted in a robust additive effect, reducing glycogen levels by 72% and achieving abundances comparable to WT animals (p<0.0001 vs untreated, p=0.65 vs WT, Fig. 6B). In the diaphragm, both ERT and SRT alone reduced glycogen abundance, and significant additivity was observed when the two treatments were given in combination (p<0.0001 vs untreated, p=0.27 vs WT fig. S6B). As seen in previous studies, ERT had a marked effect on cardiac muscle tissue, normalizing glycogen load within 12 weeks (39, 42). Nevertheless, SRT alone also had a substantial effect in cardiac muscle, reducing glycogen stores by ~60% from untreated amounts (fig. S6B). Thus, across muscle tissues either SRT or ERT alone attenuated glycogen accumulation in GAA KO mice; combining SRT with ERT enhanced glycogen reduction and normalized glycogen levels in some muscle beds.

Fig. 6. SRT, ERT, & combination therapy reduce gastrocnemius glycogen and urine and blood biomarkers after 12 weeks of treatment.

Six to nine-week old mice provided chow formulated with vehicle (veh.), MZ-101 (SRT), or alglucosidase alfa 20mg/kg biweekly (ERT) for 12 weeks. (A) Study design schema. (B) Glycogen content of gastrocnemius tissue lysate assessed via biochemical assay. (C) MALDI analysis of glycogen branch chain length (top) and ratio of phosphorylated glycogen to dephosphorylated (bottom). (E) Urinary tetrasaccharide (uGlc4) glycogen breakdown product. (F) Glycogen in peripheral blood mononuclear cells (PBMC) collected at day +14 after ERT dose. Data are individual animals (circles) with mean +/− SEM. One-way ANOVA with Tukey’s post hoc analysis ns p>0.05, **p<0.01,***p<0.001, ****p<0.0001.

To evaluate the structure of glycogen in greater detail we used matrix-assisted laser desorption/ionization (MALDI) coupled with traveling-wave ion mobility high resolution mass spectrometry (TW IMS) to interrogate both the structural properties of glycogen granules and the regional variation in glycogen content within gastrocnemius muscle. MALDI analysis of gastrocnemius tissue slices revealed greater total glycogen in GAA KO gastrocnemius as compared to WT mice, as observed by biochemical analysis. However, no differences were observed in the distribution of glycogen chain lengths or the phosphorylated chain length (Fig 6C). Similarly, treatment with MZ-101 with or without combination therapy resulted in substantial reduction of total glycogen abundance but did not modify the distribution of chain length or phosphorylation properties of glycogen granules (Fig. 6C, fig. S6C). Spatial mapping of whole tissue slice sections revealed homogenous accumulation of glycogen across both gastrocnemius and diaphragm sections in GAA KO and WT mice (fig. S6D). SRT uniformly reduced glycogen across the muscle section suggesting consistent distribution and inhibition across heterogenous muscle fiber types (fig. S6D). These data indicated that inhibition of glycogen synthesis could achieve uniform reduction in glycogen levels across muscle tissue beds without disrupting the structure of glycogen granules.

Urinary glucose tetrasaccharides (uGlc4) are an established biomarker of muscle glycogen content that has been used in glycogen storage diseases, including Pompe. uGlc4 is a product of glycogen released from cells and digested by circulating pancreatic amylase. Concentrations of uGlc4 are elevated in Pompe disease and reduced by ERT (43, 44). We hypothesized that this marker would also reflect reductions in muscle glycogen due to SRT. We evaluated Glc4 in terminal urine samples collected following 12 weeks of treatment with ERT, SRT or the combination in WT and GAA KO mice. We found that both ERT and SRT reduced uGlc4, and that the combination further reduced uGlc4 to concentrations comparable to WT mice. These changes were consistent with changes in muscle tissue glycogen stores (Fig. 6, B and D, fig. S6B). In addition, we also developed an assay to quantify glycogen content in peripheral blood mononuclear cells (PBMCs) isolated from treated animals. We found no significant glycogen accumulation in GAA KO mice PBMCs relative to WT suggesting that PBMCs do not accumulate significant amounts of lysosomal glycogen due to loss of GAA (Fig. 6E). There was no effect of ERT on PBMC glycogen at this timepoint (Fig. 6E). However, chronic treatment with SRT, either alone (p=0.0008 vs untreated) or in combination with ERT (p<0.0001 vs untreated), significantly and robustly reduced stores of glycogen in PBMCs (Fig. 6E). Thus, both uGlc4 and PBMC glycogen reflect the reduction in muscle glycogen seen with SRT. It is also important to note that MZ-101 was well tolerated throughout the study with no impact on total body weight or blood glucose concentrations (fig. S7, A and B).

SRT, ERT, and combination therapy rescue Pompe disease pathway biomarkers in gastrocnemius muscle.

Aberrant accumulation of lysosomal glycogen is the first pathological insult in a series of cellular cascades in Pompe disease that ultimately result in muscle atrophy. Studies from mouse models of Pompe and human tissue samples have identified lysosomal function and autophagy as two key disrupted pathways (17, 45). We hypothesized that reduction in muscle glycogen by SRT would improve these downstream pathologies in GAA KO mice. We evaluated autophagolysosomal pathway using three protein markers, lysosomal associated membrane protein 1 (LAMP1), ubiquitin binding protein 62 (p62), and microtubule-associated proteins 1A/1B light chain 3B (LC3B). LAMP1, a marker for lysosome accumulation, was significantly elevated in the GAA KO gastrocnemius and diaphragm lysates relative to muscle lysates from WT animals (p<0.0001, Fig. 7A, fig. S7C). Treatment with SRT monotherapy significantly reduced LAMP1 in GAA KO mice (p=0.032), and combination treatment with SRT + ERT led to further reduction (p<0.0001, Fig. 7A, fig. S7C). When autophagosomes are healthy they mature and fuse regularly with lysosomes and dispense their cargo into the hydrolytic environment of the lysosome for degradation. Autophagosomal block occurs when autophagosomes form but cannot fuse properly with lysosomes; consequently their cargo is not recycled properly by the cell. The proteins p62 and LC3B are both markers of autophagosomes that are degraded upon proper autophagosome fusion with lysosomes. Both p62 and LC3B accumulated in GAA KO gastrocnemius lysates relative to WT, but not in diaphragm, indicating dysfunction of this key recycling pathway (Fig. 3C, Fig. 7A, fig. S7C). Biweekly treatment with 20 mg/kg ERT had no impact on these markers after 12 weeks, whereas SRT monotherapy generated a small reduction, but combination therapy normalized these markers to within the range observed in WT gastrocnemius tissue (Fig. 7A). The effects of treatment with SRT and ERT on autophagolysosomal markers was confirmed by staining of LAMP1 and p62 in gastrocnemius (Fig. 7B) and diaphragm sections (fig. S7E). In aggregate, these results suggested that reduction of muscle glycogen with the combination of SRT and ERT could also improve autophagolysosomal function.

Fig. 7. SRT, ERT, and combination therapy rescue of Pompe disease pathway biomarkers in gastrocnemius muscle.

Six to nine-week old mice provided chow formulated with vehicle (veh.), MZ-101 (SRT), or alglucosidase alfa 20mg/kg biweekly (ERT) for 12 weeks. (A) Western blot quantification of gastrocnemius tissue lysate for LAMP1, p62, and LC3B. (B) Representative IHC of gastrocnemius tissue section for detection of LAMP1 and P62, scale bar 50um. (C) Western blot quantification of gastrocnemius tissue lysate for GYS1, phosphorylated GYS1, and GLUT1. Data are individual animals (circles) with mean +/− SEM. One-way ANOVA with Tukey’s post hoc analysis ns p>0.05, *p<0.05, **p<0.01,***p<0.001, ****p<0.0001.

Substrate reduction therapy improves abnormalities in gastrocnemius cellular metabolism and skeletal muscle function in a mouse model of Pompe disease.

Lysosomal glycogen accumulation in Pompe disease promotes cellular metabolic dysregulation, which has been observed at both the metabolite and transcriptional levels in muscle tissue from Pompe disease mouse models (36, 46) as well as in muscle biopsies from patients with Pompe disease. We hypothesized that this metabolic dysregulation would be corrected by reducing muscle glycogen with SRT. We performed direct analysis of components of the glycogen synthesis pathway as well as unbiased metabolomic and transcriptional profiling to comprehensively assess the impact of SRT on cellular metabolism.

We first evaluated quantity and phosphorylation status of GYS1. We observed no difference in GYS1 protein in gastrocnemius muscle from WT compared to GAA KO mice at baseline or with ERT or SRT treatment (Fig. 7C) however a small reduction in GYS1 was observed with SRT only in diaphragm (fig. S7D). Phosphorylated GYS1 was elevated in the GAA KO mice gastrocnemius and diaphragm relative to WT controls. ERT alone did not alter GYS1 phosphorylation, whereas SRT alone significantly reduced GYS1 phosphorylation (p=0.001 vs untreated), and ERT+SRT combination further reduced GYS1 phosphorylation to WT amounts (Fig. 7C, fig. S7D). The glucose transporter GLUT1 was elevated in GAA KO animals, reduced by SRT alone, but not ERT, and restored to WT abundance by ERT+SRT combination treatment (Fig. 7C). These data suggested that inhibition of GYS1 by SRT, alone or in combination with ERT, could correct abnormalities in glycogen metabolism in the GAA KO mouse model.

To comprehensively investigate the effect of ERT, SRT, and combination therapy on cellular metabolism, we used gas chromatography-mass spectrometry (GC-MS) to determine the abundance of metabolic analytes in gastrocnemius muscle tissue from the 3-month treatment cohorts. We identified more than 100 unique analytes, spanning central carbon, glycan, and amino acid metabolism. We observed dysregulation of metabolites annotated to the central carbon and amino acid metabolism pathways in Pompe GAA KO mice gastrocnemius relative to wildtype (13/41 metabolites tested; two-tailed t-test, p<0.05; fig. S8A). Particularly prominent dysregulation was observed in central carbon metabolism as evidenced by the pathological accumulation of analytes across glycolysis, the pentose phosphate shunt pathway, the Krebs cycle, amino acid metabolism, as well other energy pathways (Fig. 8A, fig. S8A). In contrast, we detected no change in the abundance of any of the measured glycans.

Fig. 8. Substrate reduction therapy improves abnormalities in gastrocnemius cellular metabolism and skeletal muscle function in a mouse model of Pompe disease.

Six to nine-week old mice fed chow with vehicle (veh.) or MZ-101 (SRT), or alglucosidase alfa 20mg/kg biweekly (ERT) for 12 wks. Polar metabolite GC/MS and RNA-seq were performed on n=5 mice in each treatment group. (A) Heatmap of cohort average Log2FC vs WT abundance in gastrocnemius. (B) GC/MS glycogen biosynthetic pathway analyte bar graphs n=5, mean, s.e.m. One-way ANOVA Tukey’s post hoc (C) Heatmap average Log2FC expression in each condition vs WT for all genes from the glycogen biosynthetic pathway. Hyper-geometric test (ora) on whole pathway enrichment based on genes significantly differentially expressed (p < 0.05) verses WT. (D) Gastrocnemius glycogen post 7-month treatment. (E) Latency to fall in accelerating rotarod test post 7-months treatment. One-way ANOVA, Tukey’s post hoc ns p>0.05, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001

Analytes in the glucose metabolic pathway were among the most significantly altered in GAA KO mice (Fig. 8B, fig. S8A). In Pompe mice, glycogen was substantially elevated (~4-fold relative to wildtype), consistent with biochemical and MALDI measurements. In addition, five out of six glycolytic intermediates were also significantly elevated (~2-4-fold) in GAA KO mice relative to wildtype, including glucose, G6P, dihydroxyacetone phosphate and lactic acid (Fig. 8B). None of these glycolytic metabolites remained substantially altered in the presence of SRT treatment or SRT+ERT combination therapy, but all remained dysregulated in the presence of ERT.

To quantify whether SRT could correct the broader set of metabolite abnormalities found in the GAA KO mice, we used a clustering approach to compare the global polar metabolite pattern between WT mice and GAA KO mice under each treatment condition. We found that muscle from ERT-treated animals clustered most closely with vehicle-treated GAA KO animals, whereas muscle from animals receiving SRT, either alone or in combination with ERT, clustered more closely with wildtype animals. (Fig. 8A). These data suggest that GAA KO mice exhibited a pervasive metabolic derangement that was ameliorated by chronic GYS1 inhibition.

To further characterize the cellular pathology in Pompe GAA KO mice, we profiled the transcriptomes in gastrocnemius muscle derived from mice subjected to each treatment condition, and then analyzed transcriptional changes between treatment groups. We used ontological enrichment analysis to detect transcriptional changes consistent with well-defined biological states, as well as a set of four manually curated gene sets previously shown to be associated with Pompe disease (fig. S8B; see supplementary Materials and Methods).We identified multiple pathways significantly dysregulated in Pompe GAA KO vehicle-treated muscle, including inflammation-associated signatures such as inflammatory response, allograft rejection, and IL6 JAK STAT3 signaling, as well as a set of transcriptional signatures associated with metabolic activity, including oxidative phosphorylation, glycogen biosynthesis, and fatty acid metabolism (fig. S8B). A majority of these gene sets remained dysregulated in the ERT or SRT single-treatment arms, but we found normalization of metabolic pathways after combination treatment (fig. S8B). SRT and ERT combination treatment resulted in a broadly increased expression of oxidative phosphorylation-associated genes (fig S8C) and a recovery of the transcripts encoding glycogen biosynthetic pathway enzymes towards WT amounts (Fig. 8C). Taken together, these data demonstrated a substantial transcriptional change in muscle tissue from Pompe GAA KO mice that was substantially corrected by SRT+ERT combination treatment.

Lastly, we evaluated the impact of glycogen reduction with SRT and ERT on muscle function in Pompe mice. Previous studies had reported that functional deficits in GAA KO mice became more pronounced with aging (21).We therefore performed a longer-term study in which we treated WT or GAA KO mice with ERT, SRT or SRT+ERT for seven months. SRT was well tolerated over the longer treatment period. Reductions in glycogen were durable at seven months and similar in magnitude to those observed after three months of treatment (Fig. 8E). We subjected mice from each treatment group to the accelerating rotarod test as an integrated measure of muscle strength and coordination. Compared with WT mice, GAA KO mice had a shorter latency to fall (45 sec vs 298 sec, p<0.0001, Fig. 8D). Neither ERT nor SRT alone improved performance; however, the SRT+ERT combination significantly prolonged latency to fall (125 sec, p<0.05). These data suggested that the reduction in muscle glycogen and improved cellular metabolic profile seen with SRT in combination with ERT could over time improve muscle function in Pompe mice.

DISCUSSION

He we reported the preclinical characterization of MZ-101, a selective small molecule inhibitor of GYS1, which effectively depletes glycogen in cells and animals. In a mouse model of Pompe disease, substrate reduction by GYS1 inhibition alone was well tolerated, reduced the pathologic buildup of glycogen, and augmented the effectiveness of ERT. We extended previous proof-of-principle studies based on genetic reduction of GYS1 by administering a small molecule that allows finer control over the degree of enzyme inhibition, distributes uniformly in muscle, and can be taken orally. These results suggested that SRT may be a viable therapeutic approach, alone or in combination with ERT, for the treatment of Pompe and other glycogen storage diseases.

Our analyses of Pompe mice highlighted the central role of aberrant glycogen metabolism in Pompe disease pathology. We observed an upregulation of glycogen synthesis in muscle tissue from Pompe mice through direct measurement of glucose incorporation both in tissue lysates and in live mice. These findings extended previous reports of increased glycogen synthase activity in Pompe muscle lysates by demonstrating increased glycogen synthesis in an intact animal (23). The activation of glycogen synthesis in the face of cellular glycogen accumulation promoted a pathologic feed-forward loop that exacerbates glycogen accumulation. These observations provided further motivation for a therapeutic approach, such as GYS1 inhibition, that could interrupt this pathologic cycle.

There are several possible mechanisms that may underlie the increased glycogen synthesis measured in Pompe mice. In intact animals, either increased transport of glucose into cells or greater glycogen synthase activity could increase the incorporation of glucose into glycogen. The increases in GLUT1 protein and in intracellular glucose suggest a contribution from higher cellular glucose concentrations. However, the greater glucose incorporation observed in muscle lysates more directly reflects increases in the activity of GYS1. One potential driver of this increased activity is the allosteric activator G6P, which we found to be elevated in muscle lysates and which can override inactivating phosphorylation of GYS1. Some previous studies in GAA KO mice have also reported elevated G6P in muscle, although others have not (35, 36). These discrepancies may be due to different methods of muscle isolation or preparation and require further investigation. Additional studies will also be needed to identify the sensor of pathologic lysosomal glycogen that induces increases in glycogen synthesis. Nonetheless, the increased rate of glycogen synthesis in Pompe mice points to GYS1 as a central node that can interrupt the pathologic cascade in Pompe disease.

Currently, the only approved treatment for Pompe disease involves ERT using recombinant human GAA. The use of ERT has had a profound impact on the early mortality of patients with IOPD driven primarily by cardiomyopathy, because cardiac muscle cells efficiently endocytose recombinant GAA and transport it to the lysosome. In contrast, the clinical benefit of ERT is less pronounced in patients with LOPD becuase its efficacy in skeletal muscles is often modest or transient due to the generally poor cellular uptake and delivery of ERT to the lysosomal compartment (47). Accordingly, long-term observational studies have shown that patients with LOPD on average experience disease progression with worsening muscle strength and pulmonary function at a median of two years after the initiation of ERT (19, 48, 49). More recently, newer generations of ERT designed to improve muscle biodistribution and lysosomal targeting have been evaluated in patients with LOPD but have not demonstrated superiority to standard of care ERT in clinical outcomes of Phase 3 studies (50, 51). Therefore there is an important need for alternative approaches to treating Pompe disease, particularly mechanisms that can complement existing therapeutic strategies.

The studies presented herein have several important limitations. Foremost is the challenge of translating rodent metabolism studies to predict outcomes in humans. We employed an established rodent genetic model with no residual GAA activity, which reflects the pathology in patients with IOPD. In contrast, patients with LOPD express some residual enzyme activity. As a result, the GAA KO mouse model may underestimate the degree of glycogen clearance and reduction expected with SRT alone in patients with LOPD and therefore the potential therapeutic opportunity for SRT as a monotherapy.

Analyses performed in the UK Biobank on carriers of protein truncating variants in the GYS1 regulatory phosphatase PPP1R3A support the tolerability of lifelong genetic ~65% reduction in muscle glycogen. However, patients with GSD 0B who have complete loss of function mutations in GYS1 experience exercise intolerance and succumb in childhood to cardiac failure due to the complete absence of glycogen (25). These data suggest that there could be an upper limit to the degree of glycogen reduction that will be well-tolerated. Moreover, the consequences of pharmacologic inhibition of GYS1 may differ from lifelong genetic inhibition, and clinical studies will be required to define the therapeutic window for GYS1 SRT in humans.

At the cellular level the metabolic machinery that energizes a cell is quite similar from human to mouse, but species divergence at the organismal level for energy consumption and storage is quite striking. For example, 85% of glycogen in the human body is stored in muscle tissue with only 15% stored in the liver, reflecting the need in large mammals for a substantial muscle energy store to enable situational activity demands (52, 53). The tissue distribution of glycogen in rodents is markedly different, with only ~15-30% of glycogen stored in muscle and the rest in liver tissue (54, 55). The difference in muscle fiber type abundance between human and rodent skeletal muscle is also notable. Mouse muscles are made up almost entirely of glycogen rich type II fast-twitch muscle fibers, while human muscle composition has more of a mix of type I slow twitch oxidative muscle fibers and glycogen rich type II fibers (56). These inter-species differences may contribute to the modest and inconsistent functional deficits that have been observed in Pompe mice as compared to humans with IOPD (21, 57). In our hands, diminished performance on the accelerating rotarod test was the most robust phenotype observed in Pompe mice. After seven months of treatment with ERT and SRT, we saw partial correction of this deficit. It is possible that this correction could be improved by earlier intervention or use of a second generation approved ERT (36). Nonetheless, it remains encouraging that some functional improvement was observed in this severe mouse model.

Pompe disease is one of more than a dozen diseases caused by an inborn error of metabolism that result in aberrant build-up of glycogen in various tissues of the body. Amongst the GSDs, there are several with significant muscle atrophy and currently no disease modifying interventions, including Adult Polyglucosan Body Disease (APBD) or Andersen’s Disease (GSD IV), McArdle Disease (GSD V), and Phosphorylase B Kinase (GSD IXb). Genetic deletion of GYS1 in the APBD mouse model rescued deleterious accumulation of glycogen, and improved both life span as well as neuromuscular function (58), implying that GYS1 inhibition might potentially be beneficial, but further studies are warranted to understand the potential therapeutic benefit of a GYS1 SRT in these GSDs. MZ-101 and other potent and selective small molecule inhibitors of GYS1 thus provide valuable tools to elucidate the role of glycogen metabolism across diverse diseases and potentially to enable therapeutic approaches for multiple patient populations.

MATERIALS AND METHODS

Detailed materials and methods can be found in the Supplementary Materials and Methods.

Study Design:

This manuscript includes many unique in vitro and in vivo studies. The over-riding hypothesis for the collective studies is that reduction of muscle glycogen will ameliorate disease biology seen cellular and mouse models of Pompe disease. The detailed information for the cell and mouse models can be found in the respective methods sections below. Individual study designs for mosue efficacy studies are detailed below. As a general comment on study conduct, all animal study operators were blinded to treatment groups. All samples were given blinded ID numbers so that sample processing and initial data analysis by group was performed blinded. All in vitro biochemistry and cell culture experiments were performed with technical triplicates in each experiment as well as replicating the experiment three times. Central findings for all mouse studies were also performed with female mice for gender replication. The primary biochemical findings presented in the three-month ERT, SRT, SRT + ERT study were replicated by the second cohort of animals that were treated out to seven months.

PK/LDH glycogen synthase assay:

For IC₅₀ determinations, recombinant GYS1 was combined with glycogen and Glucose-6-Phosphate , (Sigma-Aldrich, #G7879) in assay buffer before adding MZ-101 in DMSO (Sigma-Aldrich, #D4540).. The reaction was initiated by the addition of UDP-Glucose (Calbiochem, #270120), phosphoenolpyruvate (Sigma-Aldrich, #P7127), NADH, (Roche, #27045926) and Pyruvate Kinase/ Lactate Dehydrogenase (Sigma-Aldrich, Cat. #P0297). Plates were read at 340 nm over 10 minutes time. Data was fit to a Hill equation for dose response according to the Levenberg-Marquardt algorithm with maximum set to 100 and the minimum set to 0.

Patient fibroblast cell culture and EC₅₀:

IOPD patient fibroblasts (Coriell: GM03329, GM00248 A and GM13522) and healthy control fibroblasts (Coriell: GM05659 J, GM05565 C, and GM00302 C) were cultured in DMEM, high glucose (Gibco #10569-010) supplemented with PenStrep (Gibco #15070063), and 10% FBS (Gibco #26140095). Fibroblasts were cultured for 7 days in culture media and subsequently cultured in Low Energy Media for 7 days (DMEM no glucose (Gibco #A1443001) supplemented with 0.5mM L-Glutamine (Sigma-Aldrich #G7513), 5mM Glucose (Sigma Cat. No. G8644), PenStrep, and 1% FBS). Glycogen content was determined with the Promega Glycogen-GloTM assay kit, following the manufacturer’s instructions.

In vivo efficacy studies:

Animal care and materials:

all procedures in animals were performed in adherence to protocols approved by Maze Therapeutics’ Institutional Animal Care and Use Committee. C57Bl6 mice (Envigo), Pompe (B6;129-Gaatm1Rabn/J, JAX) and their WT controls (B6129SF1/J, JAX) mice purchased from vendors were allowed a minimum of three days for acclimation before initiation of experiments. Mice were housed under a 12-hour light/dark cycle and had free access to water and rodent chow (Envigo Teklad 2920x rodent chow), unless stated otherwise. Mice were randomized into their treatment groups and were monitored weekly for the duration of the studies.

Inhibition of de novo glycogen synthesis with MZ-101:

Male C57Bl6 mice between 7-10 weeks old were obtained from Envigo and male Pompe (B6;129-Gaatm1Rabn/J) mice between 10-32 weeks old from Jackson Labs and randomized into treatment groups. On the study day, mice were fasted for 30 minutes and dosed orally with 2, 7, 20, 70, or 200 mg/kg of MZ-101 formulated in 5%NMP, 20%PPG, 30%PEG, and 45%H₂O, or vehicle. One hour post MZ-101 or vehicle dosing, mice were given an oral dose of ¹³C₆-glucose (Cambridge Isotope Laboratories #CLM-1396-PK) at 3 g/kg. Four hours post ¹³C₆-glucose dose, mice were euthanized via CO2 and decapitation, and the gastrocnemius, heart, diaphragm, and liver were collected rapidly and flash frozen in liquid nitrogen. Plasma was also collected by centrifuging whole blood at 5,000 rpm at 4°C for 5 minutes. Frozen tissues were then submitted to Charles River Labs for 13C6-glucose analysis and plasma was analyzed for MZ-101 exposure (N=12/group).

Tissue glycogen turnover in Pompe and WT mice:

Male WT (B6129SF1/J) and Pompe mice between 8-20 weeks old were purchased from Jackson Labs and randomized into groups. On study day 1, mice were fasted for 1.5 hours prior to oral dosing of 3 g/kg ¹³C₆-glucose. Food was then returned to all mice except those to be euthanized 4 hours timepoint. Mice were euthanized 4 hours, 72 hours, 1 week, 2 weeks, 4 weeks, 8 weeks, or 12 weeks after ¹³C₆-glucose administration via CO2 and cervical dislocation, and the gastrocnemius, heart, liver, and diaphragm were dissected and flash frozen in liquid nitrogen. Frozen tissues were then submitted to Charles River Labs for ¹³C₆-glucose analysis (N=10/group).

Tissue glycogen extraction and ¹³C₆-glucose analysis:

Tissue glycogen extraction and subsequent analysis on the LC/MS were all performed at Charles River Labs (South San Francisco, CA). Briefly, tissue samples were homogenized with Omni probe in 15mL conical tubes in 5% Trichloroacetic acid (TCA). Homogenized tissues were then centrifuged at 4500 rpm for 5 minutes and supernatant transferred to a fresh vial. Nine times the volume of ethanol was then added to each supernatant and mixed thoroughly. The vials were then incubated at 37°C for 1 hour. The samples were then centrifuged at 45000 rpm for 10 minutes and the supernatant carefully removed. The resulting glycogen-containing pellet was dissolved in 1M HCl and boiled for 1.5 hours at 100°C. Glycogen samples were then analyzed on the LC/MS (API-5500).

4- and 14-week SRT study:

Male WT (B6129SF1/J) and Pompe (B6;129-Gaatm1Rabn/J) mice between 10-12 weeks old were purchased from Jackson Labs and randomized into control chow or MZ-101 (estimated dose of 1000mg/kg/day) chow groups. Body weight and food consumption were monitored biweekly. After 4 weeks and 14 weeks, mice were euthanized via CO2 and decapitation and the gastrocnemius, diaphragm, heart, and liver were dissected rapidly, and flash frozen in liquid nitrogen. The tissues were then homogenized for glycogen measurement, and plasma was analyzed for MZ-101 exposure.

ERT, SRT, and combination 3 and 7-month study:

Male WT (B6129SF1/J) and Pompe (B6;129-Gaatm1Rabn/J) mice between 6-9 weeks old were purchased from Jackson Labs. Pompe mice were randomized into vehicle (PBS, Gibco # 10010023), Lumizyme^® (referred to as ERT), MZ-101 chow (estimated 1000mg/kg/day; referred to as SRT) or ERT/SRT combo. WT mice were all given PBS intravenously biweekly as a control. Lumizyme^® was obtained from Sanofi Genzyme and administered biweekly at 20mg/kg dose IV. All Pompe mice not dosed with ERT were given PBS intravenously as a control. To prevent ERT-induced anaphylaxis, all mice, including those not receiving ERT, were given a 30mg/kg intraperitoneal injection of diphenhydramine (Sigma # D3630) 10 minutes prior to ERT/PBS injection. During the course of the study, urine was collected monthly for uGlc4 measurement. Urine was collected directly into Eppendorf tubes by scruffing mice and massaging their bladders. At the end of the study, mice were euthanized via CO₂ and decapitation, and tissues were dissected. Some of the tissues were flash frozen in liquid nitrogen whereas others were fixed overnight in 10% neutral buffered formalin (NBF).

Accelerating rotarod test for the 7-month study cohort:

All experiments were performed on a 6-lane rotarod (Maze Engineers, Skokie IL) in the morning between 8am-12pm. Mice were acclimated to the rotarod for 5 minutes at 4 rpm prior to testing each day. Thirty minutes after acclimation, mice were placed on the rotarod and tested with the following parameters: acceleration from 0 to 35 rpm over the first minute and maintained speed at 35 rpm for the next 4 minutes, totaling 5 minutes for each trial on the rotarod. Latency to drop, speed at drop, and distance at drop were auto-recorded. Analysis of data was done in graph pad prism. Results did not pass tests for normality and as such non-parametric Kruskal-Wallis test with Dunn’s post-hoc assessment was used to evaluate significant differences in cohorts relative to test performance by the WT control group.

Tissue glycogen measurement:

Flash frozen tissues were stored at −80’C before homogenization in TPER buffer with protease / phosphatase inhibitors, boiled at 100°C / 400 rpm for 10 minutes, and then centrifuged at 11,600 x g for 10 minutes / 4°C. Glycogen containing supernatants were stored at −80°C prior to assaying. The glycogen assay was done with the Promega Glycogen-GloTM assay kit (# J6022), following manufacturer’s instructions.

¹³C₆-glucose analysis:

Tissue samples were homogenized in 5% Trichloroacetic acid and centrifuged at 4500 rpm for 5 minutes to precipitate proteins. Nine volumes of ethanol were added to each supernatant followed by incubating at 37°C / 1 hour, centrifuging at 45000 rpm for 10 minutes. The glycogen pellet was dissolved in 1M HCl and boiled for 1.5 hours at 100°C and subsequently analyzed via LC/MS (API-5500).

Biochemical G6P analysis:

The concentration of Glucose-6-phosphate in tissue lysates was determined using the Sigma-Aldrich Glucose-6-phosphate Assay Kit (Sigma-Aldrich #MAK014) following the manufacturers' instructions.

Glycogen synthase enzyme activity assay:

Glycogen synthase activity was measured by incorporation of ¹⁴C-UDP-Glucose into glycogen in tissue lysates homogenized in TPER buffer containing protease and phosphatase inhibitors following the method as outlined by Thomas et al 1968 (59).

Immunoblotting:

Tissue was lysed with RIPA Buffer (TekNova R3792) containing Pierce Universal Nuclease for Cell Lysis (ThermoFisher Scientific 88701) and Halt Protease and Phosphatase Inhibitor Cocktail (ThermoFisher Scientific 78441), after which protein quantity was measured (ThermoFisher Scientific 23227). Equal amounts of protein samples were resolved by SDS-PAGE and transferred to a nitrocellulose membrane (Bio-Rad 1704159),, before primary and secondary incubations, and imaging via LI-COR Odyssey Fc System.

Polar metabolite extraction, GCMS quantification, and metabolomic analysis:

Sample preparation and GCMS protocols were similar to those described previously (12, 15). In brief, frozen tissues were pulverized in liquid nitrogen using a Freezer/Mill Cryogenic Grinder (SPEX SamplePrep), extracted in 50% methanol/H₂O (V/V 1:1), then separated into polar (aqueous layer) and protein/glycogen (pellet) fractions. The polar fraction containing free metabolites was dried, derivatized, and the protein fraction was hydrolyzed with HCl before quenching with methanol. An Agilent 7800B gas-chromatography (GC) coupled to a 5977B single quadrupole mass spectrometry detector equipped with a high-efficiency source was used for this study. Tests identifying differentially abundant metabolites were performed using GraphPad Prism; Metabolite clustering and pathway enrichment was performed using the Metaboanalyst 5.0 web tool (https://www.metaboanalyst.ca/) using MSEA QEA to determine p-values for dot plot comparisons (60). Further details can be found in supplemental methods.

Tissue preparation and analysis for MALDI-imaging:

Tissues were sectioned and mounted on positively charged glass slides for MALDI imaging and analyzed based on previous described methods (12). An HTX spray station (HTX) was used to coat the slide with a 0.2mL aqueous solution of both isoamylase (4 units/slide), and PNGase F (20mg total/ slide). For detection and separation of glycogen and N-linked glycans, a Waters SynaptG2-Xs high-definition mass spectrometer equipped with traveling wave ion mobility was used. Raw MALDI-MSI data files were processed using an algorithm available within the HDI software (Waters Corp). To adjust for mass drift during the MALDI scan, raw files were processed using a carefully curated and established list of 50 MALDI matrix peaks (m/z) and 155 glycogen and N-linked glycan peaks (m/z). Images of glycogen were generated using the waters HDI software. All data were normally distributed and met the assumption of the used statistical approaches. Statistical analyses were carried out using GraphPad Prism. All numerical data are presented as mean ± S.E. Column analysis was performed using one-way ANOVA.

RNA extraction, library generation and transcriptomics analysis:

Tissues were homogenized with TRIzol (ThermoFisher Scientific # 15596026) and RNA extraction was done according to the manufacturer’s instructions (Zymo Direct-zolTM RNA Microprep Kit). RNA concentration and integrity was determined by the Agilent 4150 TapeStation System. Libraries were created with Illumina Stranded mRNA Prep kits using a poly(A) mRNA capture method. Libraries were multiplexed and sequenced using an Illumina NovaSeq 6000 System with v1.5 reagents and an SP flow cell. Transcriptomic analysis was performed using R/Bioconductor (61-63).

Statistical Analysis:

All statistical analyses are detailed in their corresponding figure legends. All bar and line graph statistical analyses were done using GraphPad Prism and are shown as mean ± SEM. For in vitro and in vivo experiments, Student’s T-test was used to determine significance between two groups. A one-way ANOVA with Tukey’s post hoc analysis was used when there were multiple experimental groups and all groups were compared to each other combinatorially. A one-way ANOVA with Dunnet’s post hoc analysis was used when there were multiple experimental groups and all groups were compared to a single control group. A P-value less than 0.05 was considered statistically significant. Please see supplementary materials and methods for detailed descriptions of metabolomic and transcriptomic analyses.