GAA Replacement Improves Respiratory Muscle, Neural, and Alveolar Pathology in the Pompe Mouse

Abstract

Pompe disease is a devastating neuromuscular disorder caused by mutations in the gene GAA. These mutations result in a deficiency of the enzyme acid α-glucosidase (GAA), leading to lysosomal glycogen accumulation in cardiac, skeletal, and smooth muscle, motor neurons, and alveolar epithelial cells. Respiratory failure due to neuromuscular weakness, recurrent aspiration pneumonia, and tracheo-bronchomalacia are the leading cause of morbidity and mortality in PD patients. Enzyme replacement therapy (ERT) is currently the only FDA approved treatment for Pompe disease, however, gene therapy with naturally occurring and engineered adeno-associated viral vectors are also widely studied as an alternative treatment. In the present study we directly compared the benefits of existing and novel treatment modalities - ERT, AAV9-GAA, and AAVcc47-GAA, with an emphasis on correction of pathologies related to respiratory function. We find that GAA replacement in early adult mice improves respiration through 9 months of age. This improvement is attributed to glycogen clearance in the tongue, diaphragm, and lungs, which subsequently improved diaphragm neuromuscular junctions and reduced lysosomes within the alveolar epithelia.

INTRODUCTION

Pompe disease is a devastating neuromuscular disorder caused by mutations in the gene GAA^1,2. These mutations result in a deficiency of the enzyme acid α-glucosidase (GAA), leading to lysosomal glycogen accumulation in cardiac, skeletal, and smooth muscle, motor neurons, and alveolar epithelial cells^3–8. Patients with the most severe form of Pompe disease – Infantile-onset (IOPD) – have less than 1% of residual enzymatic activity present with cardiorespiratory distress in the first few months of life^4,9. Late-onset (LOPD) patients who have between 1 – 40% of enzymatic activity experience respiratory and ambulatory deficits but have limited cardiac involvement^4,10–12. Eventually most untreated Pompe patients require ventilatory assistance^5,13. Respiratory failure due to neuromuscular weakness, recurrent aspiration pneumonia, and tracheo-bronchomalacia are the leading cause of morbidity and mortality in Pompe patients^5,13,14. Historically, respiratory deficits have been attributed to an inability to maintain airways patency due to macroglossia, reduced conduction of respiratory motor neurons and nerves as well as dampened contractile ability of the diaphragm – a major inspiratory muscle^5,15–18. Recently, we described novel pathology within the trachea and alveolar epithelial cells (Alveolar type 1 and type 2 (AT1 and AT2), of the proximal airways and the alveoli, respectively^7,8. In the trachea and bronchi, we noted that GAA deficiency results in abnormal smooth muscle and distal airway structure and function^6–8. Specifically, we noted that lysosomal glycogen accumulation disrupted smooth muscle cellular architecture. This pathology also resulted in an inability of the airway smooth muscle cells to adequately contract due to calcium mishandling⁸. In the alveolar sacs, we also found that GAA deficiency in Pompe disease significantly disrupted AT1 and AT2 cellular architecture, reduced surfactant protein D (SP-D) gene expression, and disrupted autophagy⁷ .

Enzyme replacement therapy (ERT) is the only FDA approved therapy for Pompe disease with two version of ERT currently available to patients: Alglucosidase alfa (formerly Myozyme^®, now Lumizyme^®), approved in 2006, and avalglucosidase alfa (Nexviazyme^®), approved in 2021. Alglucosidase alfa is successful at treating cardiac pathology in IOPD patients resulting in improved survival rates, and is also moderately successful in treating skeletal muscle pathology, however many pathophysiologies continue to be unresolved including those in respiratory skeletal muscle, motor neurons, and smooth muscle^17,19–21. The impact of ERT on the alveolar cell pathology is also unclear. Avalglucosidase alfa has superior success in treating skeletal muscle in Pompe patients, but as a protein therapy is unable to cross the blood brain barrier to address CNS pathology^22–24.

Gene therapy is an alternative therapy that has been widely explored preclinically and is under current clinical investigation for Pompe disease^25–27. The most extensively studied vectors are those made up of adeno-associated virus (AAV) capsids carrying GAA cDNA. These vectors have utilized capsid-promoter combinations which target specific muscle groups, liver-directed GAA production, global pathology, and respiratory pathology^25–27. AAV serotype 9 (AAV9) is a current clinical candidate due to its ability to resolve pathology in many tissues of interest including the CNS^28–30. AAVcc47 is a novel capsid which was derived from the AAV9 capsid sequence with modified amino acids 452-458, corresponding to variable region IV, and isolated following cross-species evolution in the brains of pigs, mice, and non-human primates³¹. After one-month of treatment, AAVcc47-treated Gaa^-/- mice had increased GAA activity in the CNS compared to AAV9-treated mice.

Interestingly, there are very few preclinical studies which directly compare the benefits of ERT and AAV-GAA³². In the present study we aimed to do just that with an emphasis on correction of pathologies related to respiratory function in the Pompe mouse. The Gaa^-/- mouse has a complete deficiency of GAA and recapitulates many of the pathologies present in Pompe patients. Here we demonstrate that GAA replacement in early adult mice improves respiration through 9 months of age. We attribute this improvement to glycogen clearance in the tongue, diaphragm, and lungs, which subsequently improved diaphragm NMJs and reduced lysosomes within the airways.

METHODS AND MATERIALS

Animals:

Gaa^-/-;B6/129 (Gaa^-/-) and B6/129 (wildtype, WT) mice were obtained from The Jackson Laboratory³³. They were bred and maintained in accordance with the Duke University Institutional Animal Care and Use Committee (IACUC) under Protocol Numbers A232-17-10 and A182-20-09. They were maintained with food and water ad libitum. At 2 months of age, mice were enrolled into the study. Gaa^-/- mice were enrolled into one of 4 groups: AAV9-GAA, AAVcc47-GAA, ERT, and sham. WT mice also received sham injections. Mice were euthanized at 9 months of age.

Recombinant human acid α-glucosidase:

Clinical grade rhGAA (Lumizyme^®, alglucosidase alfa; Genzyme Corporation, Cambridge, MA, USA) was used as a formulated drug product. Mice were treated weekly with 40 mg/kg of rhGAA by bolus tail vein injection. All mice were treated prophylactically with 15 mg/kg diphenhydramine intraperitoneally (IP) 15 minutes prior to rhGAA administration to prevent anaphylactic reactions. 3 consecutive IP injections of MTX (10 mg/kg) were also given at 0, 24 and 48h after injection of rhGAA in weeks 1, 2, and 3 of ERT as previously described³⁴.

AAV Production and Purification:

A plasmid with a CBh promoter³⁵ and codon optimized GAA^36,37 was synthesized by Blue Heron Biotech, then cloned into a pTR-SV40pA backbone. AAV was produced and purified as previously described³¹. Briefly, adherent HEK293 cells were transfected with an AdV helper plasmid (pXX680), AAV9 or AAV.cc47 rep/cap plasmid, and pTR-CBh-GAA-SV40pA using polyethlyenimine (PEI). Vectors were purified from cell media by precipitation in polyethylene glycol followed by iodixanol gradient and were finally buffer exchanged using a Zeba spin desalting column (ThermoFisher Scientific). Purified AAV was titered via qPCR.

Respiratory Analysis:

Whole body plethysmography was performed as previously described at 9 months of age³⁸. Unanesthetized, unrestrained mice were placed into Plexiglas chambers (DSI, St. Paul, MN) and respiratory data was collected and analyzed using FinePointe Software. Respiration was evaluated under normoxic conditions (FiO₂: 0.21, N₂ balance) for 1.5 hours, during which a 5-minute period of quiet breathing was selected as baseline. Mice were then exposed to a hypercapnic and hypoxic (FiO₂: 0.10, FiCO₂: 0.07, N₂ balance) challenge for 10 minutes. The flow rate of both gases is 0.5 L / min to each chamber. To reduce circadian rhythm effects WBP was initiated on all mice during similar times of days. Mice entered the chambers under normoxic conditions between 8:15am and 11:30am, with an average start time of 9:30am. Challenge conditions were initiated between 10:20am and 1:00pm, with an average of 11:45am. The FinePointe Software calculates the box flow tidal volume by multiplying the raw tidal volume by the compensation factor using the Drorbaugh Fenn algorithm which compensates for temperature and humidity in the WBP chamber. Body temperature was not included in this analysis but was recorded for each mouse before and after the WBP session.

Limb Function Analysis:

Ledge Test: Unanesthetized mice were placed on the edge of an empty cage and observed for 5-minutes total as they walked along the edge and lowered themselves down. Mice were scored from 0 (least severe) – 3 (most severe) based on ability to balance on ledge and coordination of moving off the ledge³⁹. The test was repeated (2 total rounds) and the average of the scores was recorded. Wire Hang: Mice were placed on a wire mesh with 0.5 inch squares. They wire mesh was inverted and held ~ 6 inches from the table. Mice were observed for 2 minutes total. Time was recorded when the mice fell from the mesh. This test was repeated (2 total rounds) and the average of the scores was recorded.

Glycogen Quantification:

Glycogen was quantified using an Abcam Glycogen Assay Kit as previously described⁴⁰. Briefly, tissues were snap frozen, and homogenized in water with protease inhibitor using a FastPrep24. Samples were then heated for 10 min at 95°C and centrifuged for 10min at 4°C and 13k rpm. Samples were incubated with hydrolysis enzyme mix for 30 minutes at room temperature, followed by incubation with development mix for 30 minutes at room temperature. Absorbance of each sample was evaluated using a Varioskan Lux (ThermoFisher) μDrop plate and compared to a standard curve. Glycogen quantities were normalized to total protein per reaction determined by DC Protein Assay (Bio-Rad), performed in accordance with manufacturer instructions.

GAA Activity Quantification:

GAA Activity was quantified as previously described⁴⁰. Briefly, tissues were snap frozen, and 50mg pieces were homogenized in water with protease inhibitor. Samples were frozen and thawed 3x prior to centrifugation for 10 min at 4°C and 13k rpm. Samples were incubated with 4-methylumbelliferyl-α-D-glucosidase (Sigma) and sodium acetate (Sigma) pH 4.3 for 1 hour at 37°C. The reaction was stopped with sodium carbonate (Sigma) pH 10.7. Fluorescence was measured using a Varioskan Lux (ThermoFisher) and compared to a standard curve of 4-methylumbelliferone (Sigma). GAA activity was normalized to total protein per reaction determined by DC Protein Assay (Bio-Rad), performed in accordance with manufacturer instructions.

Vector Genome Biodistribution:

Vector genomes were quantified as previously described³⁶. Briefly, tissues were snap frozen and genomic DNA (gDNA) was purified using Qiagen DNeasy Blood and Tissue kits. The concentration of samples was evaluated using a Varioskan Lux (ThermoFisher) μDrop plate. 100ug of each gDNA sample were evaluated in triplicate with SYBR Green, SV40 primers (Fwd: AGCAATAGCATCACAAATTTCACAA; Rev: GCAGACATGATAAGATACATTGATGAGTT), and a CFX Opus 384 (Bio-Rad), and were compared to a standard curve using pTR-CBh-GAA-SV40pA.

Diaphragm NMJ Evaluation:

NMJ evaluation was completed as described previously⁴¹. Briefly, hemi-sections of each diaphragm were stretched on dental wax and fixed in 4% paraformaldehyde (PFA) for 15 min, followed by 3, 10 min washes with PBS. They were then permeabilized with 2% triton x-100 in PBS for 30 min, and blocked overnight at 4°C with 0.05% triton x-100, 4% BSA, and 4% normal horse serum in PBS. Presynaptic plates were identified with a synaptotagmin antibody (znp1, Zebrafish International, 1:200, mouse), axons were identified with an NF-H antibody (CPCA-NFH, EncorBio, 1:800, chicken), and post-synaptic plates were identified with AlexaFluor594-conjugated α-bungarotoxin. Images were taken with a Zeiss 780 Upright Confocal microscope and Zen Software. Each NMJ was outlined using ImageJ by a blinded reviewer. The Coloc2 plug-in for ImageJ was used to determine Pearson’s correlation coefficient for each NMJ.

Medulla & Spinal Cord Histology:

The medulla and cervical spinal cord were removed from the mouse en bloc and placed in 4% PFA for 72 hours. The brainstem and spinal cord were then extracted from the bone and fixed in fresh 4% PFA for 24 hours before placing in 30% sucrose in PBS for 72 hours. The medulla and spinal cord were then separated and frozen in Optimal Cutting Temperature (OCT) Compound and maintained at -80°C. Blocks were sectioned with a cryostat at 40μm and sections were maintained in 4% PFA at 4°C. Selected sections were first washed (3x with 1xPBS for 5 minutes each at room temperature (RT)) and then quenched (70% 1xPBS, 30% methanol, and 0.6% hydrogen peroxide) for 30 minutes. Sections were blocked (90% 1xPBS, 10% normal horse serum, and 0.4% fish gelatin). Sections were incubated at 4°C overnight with an anti-GAA antibody (1:100, Abcam, ab102815) in antibody buffer (98% 1xPBS, 2% normal horse serum, 0.4% fish gelatin, and 0.3% triton x-100). The secondary antibody, Biotin-SP AffiniPure™ anti-Rabbit (Jackson ImmunoResearch, 711-065-152), was diluted 1:500 in antibody buffer and the sections were incubated for 2 hours at RT. The Vectastain ABC Kit (VectorLabs, #PK-4000) and ImmPACT^® DAB Substrate Kit (Vectorlabs, #SK-4105) were applied according to manufacturer’s instructions. Sections were then counter-stained with cresyl violet and dehydrated. Brightfield images were taken of each section from each animal using an ECHO Revolve.

Lung Histology:

Following perfusion with PBS, lungs were inflated and fixed with 4% PFA as previously described⁷. Lungs were cryopreserved with 30% sucrose in PBS, embedded in OCT, cryosectioned at 8–10 μm, and stored at −80°C. Sections were fixed with 4% PFA and stained with hematoxylin and eosin as previously described³⁷. Brightfield images were taken of each section from each animal using an ECHO Revolve.

For immunofluorescence staining, antigen retrieval was performed using 10 mM sodium citrate buffer in a water bath (95°C for 10 min). Sections were washed with PBS-T (0.1% Triton x-100 in PBS) and incubated with 1% BSA for 30 min at RT followed by primary antibodies at 4°C overnight. Primary antibodies included: Prosurfactant protein C (MilliporeSigma, AB3786, 1:1,000), RAGE/AGER (R&D systems, AF1145 Goat 1:400), LAMP1 (DSHB Cat. No. 1D4B, 1:500) and GAA (Proteintech, 14367-1-AP, 1:500). Sections were then washed, incubated with secondary antibodies (Invitrogen: anti-Rabbit, A32754, 1:500, anti-Rat, A78947, anti-Goat, A32814, 1:500) in blocking buffer for 1 h at RT, washed with PBS-T three times, and mounted using Vectashield reagent with DAPI. The sections were visualized using an ECHO Revolve light microscope. Ten images were taken from different areas of different lung sections for each slide, and the intensity of LAMP1 and GAA was quantified and averaged.

RESULTS:

GAA Replacement Improves Respiration through 9 months of Age:

At 2 mo Gaa^-/- mice were enrolled in one of the following groups: control, AAVcc47-GAA, AAV9-GAA, and ERT. The control and AAV groups intravenously received a one-time administration of either 1e14kg/vg AAV or AAV formulation buffer (sham). The ERT group received an intravenous administration of 40 mg/kg rhGAA once per week until study completion. The ERT group also received 3 doses of methotrexate for the first three weeks of ERT delivery. A cohort of WT mice was also intravenously administered AAV formulation buffer. All mice were aged to 9 months prior to in vivo and post-mortem analyses.

First, respiratory effort was measured under baseline (room air, 21% O₂, N₂ bal) and challenge (10% O₂, 7% CO₂, N₂ bal) conditions in all mice at 9mo. We report respiratory effort via frequency (f), tidal volume (V_T), minute ventilation (V_E), peak inspiratory flow (PIF), and peak expiratory flow (PEF). Minimal differences were observed between groups under baseline conditions (Table 1), however, under challenge conditions more significant differences between Gaa^-/- and WT mice present themselves (Figure 1A). Specifically, Gaa^-/- have significant deficits in f, V_E, and PIF. Among the treated groups many improvements were observed. AAV9-GAA treated Gaa^-/- mice do not have significant differences in f, V_E, and PEF compared to WT (p>0.05), and were either significantly or trending different than untreated Gaa^-/- mice (V_E: p=0.07, PIF: p<0.05). ERT-treated Gaa^-/- mice were also not significantly different than WT mice and their f, V_E, and PIF were all significantly different than untreated Gaa^-/- mice (p<0.05). Mice treated with AAVcc47-GAA had significant respiratory deficits in all measures (vs WT: p<0.05). Significant differences were observed between the AAV-treated groups for most measures (AAV9 vs AAVcc47: V_T, V_E, PIF, PEF: p<0.05) as well as between AAVcc47-GAA treated mice and ERT treated mice (f, V_E, PIF: p<0.05). There were minimal significant weight deficits among any groups of mice which would impact respiration with only AAV9 and ERT treated mice reaching significance (Figure 1B). Body temperatures were recorded before and after the WBP session to verify differences in respiration are not due to temperature fluctuations (Table 2). Variations in WBP measures are also not due to sex differences (Supplemental Figure 1).

Table 1: Baseline Data.

	Frequency		TV		MV		PIF		PEF
	Mean	SEM	Mean	SEM	Mean	SEM	Mean	SEM	Mean	SEM
WT	139.5	10.1	0.2260	0.0150	29.97	0.92	3.090	0.123	2.192	0.145
Gaa-/-	160.8	4.6	0.1773	0.0060	28.19	0.96	2.854	0.116	1.842	0.113
AAVcc47	166.6	6.1	0.1879	0.0100	30.18	1.63	3.001	0.188	1.922	0.115
AAV9	170.7	6.9	0.1985	0.1040	33.30	2.18	3.261	0.142	1.966	0.114
ERT	158.6	6.5	0.2013	0.0130	31.33	1.83	2.770	0.265	1.662	0.120

Figure 1. Respiratory and Limb Function in Treated Gaa^-/- mice.

Respiratory measurements – frequency, tidal volume, minute ventilation, peak inspiratory flow, and peak expiratory flow – evaluated at 9 months of age (A). Untreated Gaa^-/- mice have deficits in multiple measures which are improved in AAV9- and ERT-treated mice. There is no significant variation between each group in weights of mice at 9 months (B). Limb strength measured by a ledge walking test and inverted wire hang test demonstrate the weakness of untreated Gaa^-/- mice which is improved in AAV and ERT treated mice (C). n = 7 – 8 per group (n=4 males, n=4 females in WT, Gaa^-/-, AAVcc47, and ERT groups; n=4 males, n=3 females in AAV9 group), *p<0.05, **p<0.01, ***p<0.001.

Table 2. Temperatures.

	Before		After
	Mean	SD	Mean	SD
WT	37.34	0.374	37.56	0.311
Gaa-/-	36.97	0.472	36.7	0.638
AAVcc47	37.23	0.382	36.66	0.739
AAV9	37.34	0.336	36.78	0.311
ERT	37.87	0.656	36.95	0.373

GAA Replacement Improves Limb Function:

Limb strength was also assessed at 9 months, via a ledge walking test and inverted wire hang time (Figure 1C). Untreated Gaa^-/- mice had significantly greater severity scores on the ledge test (vs WT: p<0.05). The AAV-GAA and ERT treated mice had improved severity scores, which were not significantly different than either untreated Gaa^-/- or WT mice. Untreated Gaa^-/- mice also had significantly reduced hang times (66.7± 48.4 seconds) compared to WT mice (120 ± 0 seconds). All AAV-GAA and ERT treated mice had significantly improved hang times compared to untreated Gaa^-/- mice and were also not significantly different than WT mice. Similar to WT mice, AAV9-treated mice had no mice with reduced hang time.

AAV9 and AAVcc47 Transduce a Broad Range of Tissues in Pompe Mice:

Cardiac, skeletal, and smooth muscle, as well as the CNS and lungs were collected for analysis. A biodistribution analysis of vector genomes was performed to understand where AAVs transduced (Figure 2). We observe relatively equal vector genome presence between AAVcc47 and AAV9 in the tongue, esophagus, medulla, and cervical spinal cord. Only in the liver are there nearly significantly fewer vector genomes present in AAVcc47 treated mice than the AAV9 cohort, with lower (albeit statistically non-significant) average vector genomes detected in diaphragm, TA, and the heart as well.

Figure 2. Vector Genomes in Gaa^-/- mice.

Biodistribution of vector genomes in skeletal muscle, cardiac muscle, airway epithelium, smooth muscle, and the central nervous system. n = 3 – 6 per group, *p<0.05, **p<0.01.

AAV9-CBh-GAA and AAVcc47-CBh-GAA are expressed and ERT is taken up Across a Broad Range of Tissues:

The CBh promoter is ubiquitously expressed including in key tissues for treating Pompe disease – the CNS and skeletal muscle^31,35. GAA activity in the tongues, diaphragms, and tibialis anteriors (TA) from AAV-treated mice are ~3.5 – 24x greater than WT mice, with AAV.cc47 showing higher expression than AAV9 consistent with the vector biodistribution observed in these tissues (Figure 3A). In ERT-treated mice, the GAA activity in the TA and the diaphragm was slightly above WT levels (1.1 – 2.7x greater) but only reached 58% of WT levels in the tongue. Greatest levels of GAA activity is observed in the hearts of AAV-treated mice which have ~29 – 224x greater activity than WT mice. There are moderately supraphysiological levels of activity (1 – 3.5x greater than WT) in the esophagi and lungs of AAV- and ERT-treated mice. The only tissue evaluated that did not reach WT levels of activity is the bladder in which only 60 – 85% of WT levels are reached among the three treatment groups.

Figure 3. GAA Expression in Gaa^-/- mice.

GAA enzyme activity levels in a range of skeletal muscle (tongue, diaphragm, tibialis anterior), cardiac muscle, the airway epithelium (lungs), smooth muscle (esophagus and bladder), and the liver (A) n = 3 – 6 per group, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Representative images of GAA-positive (brown) neurons (purple) within the hypoglossal motor nucleus (B) and within the putative phrenic motor nucleus (C). Scale bar of full sections = 890 μm, Scale bar of motor pools = 180 μm

To understand GAA expression and cross-correction uptake in respiratory motor neurons specifically, we performed anti-GAA immunohistochemistry within hypoglossal motor neuron-containing regions of the medulla and within the C3-C5 segments of the cervical spinal cord which houses phrenic motor neurons (Figure 3B and C). The medullas and cervical spinal cords from AAV-treated mice have robust GAA expression through the entire hypoglossal and putative phrenic motor neuron pools. In contrast the untreated Gaa^-/- and ERT-treated mouse motor pools are nearly devoid of any positive staining.

Increased GAA Activity Results in Glycogen Clearance:

To understand the impact of GAA expression, glycogen was quantified across a range of tissues (Figure 4). In untreated Gaa^-/- mice, glycogen was significantly accumulated in all tissues with a broad range of ~7.7 – 4000x greater than in WT mice, which have negligible amounts of quantifiable glycogen. In skeletal muscle (tongue, diaphragm, tibialis anterior) a range of glycogen degradation occurs with 0 – 89% of glycogen cleared among AAV- and ERT-treated mice, however lower levels of glycogen clearance were observed in ERT treated mice compared to AAV treated mice in the diaphragm and tibialis anterior. In the heart, there is near complete clearance of glycogen (>89%) with all treatments. In the esophagi, lungs, and bladders, there is a moderate level of glycogen clearance (~45 – 65%) compared to untreated Gaa^-/- mice.

Figure 4. Glycogen Clearance in Gaa^-/- mice.

Quantification of glycogen across skeletal muscles (tongue, diaphragm, tibialis anterior), cardiac muscle, lungs, and smooth muscles (esophagus and bladder). WT mice exhibit near 0 levels of detectable glycogen, which untreated Gaa^-/- have significant glycogen accumulation in all tissues. Treatment reduces glycogen accumulation to variable degrees in each tissue. n = 3 – 6 per group, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

GAA Replacement Moderately Improves Diaphragm NMJ Integrity:

Neuromuscular junctions within the TA and diaphragm of Pompe mice are characterized for having reduced overlap between the pre- and post-synaptic plates and being fragmented⁴². Here, we confirm these prior findings with quantitative analyses of reduced pre- and post-synaptic plate correlation (Figure 5A) and qualitative analyses of increased fragmentation (Figure 5B) in Gaa^-/- mice compared to WT mice. AAV-treated mice have an intermediate phenotype with significant improvements to the synaptic plate overlap. Additionally, these treated NMJs are both less fragmented than untreated Gaa^-/- mice but also not as compact as WT mouse NMJs. In contrast, ERT-treated mice have NMJs with significantly disrupted synaptic plate overlap and maintain a highly fragmented appearance.

Figure 5. Neuromuscular Junctions in Gaa^-/- mice.

Average Pearson’s correlation coefficient based on overlap of presynaptic plate (synaptotagmin) and postsynaptic plate (acetylcholine receptors) which make up the neuromuscular junction (NMJ) (A). n = 4 – 6, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. A representative image of motor neuron axons (magenta) terminating at the presynaptic plate (green) overlapping with the postsynaptic plate (red) from a WT mouse, followed by representative images of individual NMJs from each group (B).

Lung Pathology Analysis:

We recently reported the impact of GAA-deficiency on alveolar epithelial cells within the lungs of Gaa^-/- mice and the impact of treatment has not previously been studied with respect to addressing this pathology⁷. Here, we stained lung sections with hematoxylin and eosin to evaluate the morphology and structure of the alveolae (Figure 6A). We confirm the hypercellularity and smaller intra-alveolar space present in Gaa^-/- mice compared to WT mice. The lungs of AAV and ERT-treated mice exhibit pathology consistent with moderate levels of treatment including areas of narrowing within the alveolar space, similar to that found in untreated Gaa^-/- mice, as well as areas of enlarged alveolar space, which is more similar to that found in WT mice.

Figure 6. Alveolar Pathology in Gaa^-/- mice.

Representative images of hematoxylin and eosin-stained lung sections depicting the hypercellularity and reduced intra-alveolar space in Gaa^-/- mice compared to WT, which is moderately improved in treated mice (A). Scale bar = 90μm. Representative images of lung sections stained to identify GAA (red), LAMP1 (cyan), RAGE (green), and DAPI (blue) (B). Intensity of GAA and LAMP1 was quantified in AT2 (RAGE+) cells (C) n = x per group, , *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

To further characterize the impact of increased GAA on the alveolar pathology, we evaluated lungs sections that were triple-stained with GAA, LAMP1 (lysosome marker), and RAGE (alveolar type 1 (AT1) cell marker) (Figure 6B). Then, intensity of GAA and LAMP1 staining was quantified within RAGE+ AT1 cells (Figure 6C). As expected, untreated Gaa^-/- mice are devoid of GAA (p<0.0001) and have 1.6x greater levels of LAMP1 than WT (p<0.001). AAVcc47-treated mice have the greatest levels of GAA expression with modestly supraphysiological levels (WT vs AAVcc47: n.s.) and the most robust correction of pathology with 46% less LAMP1 accumulation than untreated Gaa^-/- mice (Gaa^-/- vs AAVcc47: p<0.0001). AAV9- and ERT-treated mice have intermediate phenotypes. AAV9-treated mice have 92% of WT levels of GAA (WT vs AAV9: n.s.) and an 18% reduction in LAMP1 accumulation compared to untreated Gaa^-/- mice (Gaa^-/- vs AAV9: p<0.05). Lastly, ERT-treated mice have 69% of WT levels of GAA (WT vs ERT: p<0.05) and a 34% reduction of LAMP1 accumulation compared to untreated Gaa^-/- mice (Gaa^-/- vs ERT: p<0.001).

DISCUSSION

Overall conclusions:

In this study we demonstrate that AAV-GAA gene therapy and rhGAA ERT improve maximal respiratory capacity by increasing GAA enzyme in the tongue, diaphragm, lungs, and both hypoglossal and phrenic motor neurons which decreases glycogen accumulation. Substantial glycogen clearance in the diaphragm and putative phrenic motor neurons with AAV-GAA leads to improved overlap of the pre- and post-synaptic plates that make up the neuromuscular junction. Additionally, increased expression of GAA in the alveolar sacs with reduced total LAMP1+ structures in the alveolar cells.

Comparing AAV gene therapies & ERT on Respiratory Pathology:

Glycogen accumulation has been reported in patients and the Gaa^-/- mouse in several tissues which are in total responsible for maximal respiratory function including the genioglossus, diaphragm, hypoglossal motor neurons (MNs), phrenic MNs, intercostal MNs, diaphragm neuromuscular junctions, trachea, bronchi, and alveoli^15,17,18,42. We and others have highlighted the importance of holistically addressing these tissues in order to fully prevent respiratory decline^{5,6,13,17,18,43}.

Enzyme replacement therapy of recombinant human GAA (rhGAA) is currently the only FDA approved treatment for Pompe disease. ERT reduces overall glycogen accumulation resulting in improved muscle function, increased lifespan, and ventilator free survival of Pompe patients. Alglucosidase alfa (Myozyme^®/Lumizyme^®) is an rhGAA which was approved in 2006 and 2010. In 2021, alvaglucosidase alfa (Nexviazyme^®), an updated rhGAA with increased mannose-6-phosphate moieties leading to increased cellular uptake was approved for use in LOPD patients^22–24. Patients on alglucosidase alfa experienced short term benefits, however long-term respiratory complications persisted leading to morbidity and mortality^13,17,20. In contrast, patients receiving avalglucosidase alfa have demonstrated a delay in respiratory decline²⁴. Specifically, LOPD patients who switched from alglucosidase alfa to avalglucosidase alfa had either increased mean improvement or stable percent predicted upright FVC values²⁴.

AAV-GAA gene therapy is a promising alternative to ERT. The primary advantages of gene therapy are the need for fewer administrations (single administrations are likely to last several years) and the ability to cross the blood-brain barrier to treat the nervous system^44,45. Glycogen accumulation across the nervous system including a particular emphasis on motor neurons and its subsequent impact on function has been well documented in Pompe patients and the Pompe mouse model. Some AAV capsids including AAV9 and its derivatives (AAVcc47) have the ability to cross the blood-brain barrier when administered intravenously or intramuscularly^31,41,46,47. AAV-derived GAA results in glycogen clearance across a range of cell populations in the brain and spinal cord as shown here^36,48. In the present study, we observe minor differences in uptake, GAA expression, and glycogen clearance between different tissues from AAV9- and AAVcc47-treated mice. Some notable difference between AAV9 and AAVcc47 in this study include reduced cardiac and liver uptake of AAVcc47 leading to reduced overexpression in the heart, all of which have broad implications in circumventing potentially harmful immune responses. Additionally, AAVcc47-treated mice had significantly greater GAA expression in the tongue and diaphragm – two key respiratory muscles. Variability in GAA activity and glycogen clearance across groups and tissues may be impacted by both the uptake of the vectors in different cell types (based on cell surface receptors) as well as the non-homogenous distribution of uptake across cells within a given tissue. While similar sections of tissue are collected for post-mortem analysis, the entire tissue is not evaluated and may lead to variation in values. Additionally, variability in glycogen accumulation and clearance in is attributed to the range of tissues evaluated here, given that not a cell types utilize glycogen at equal rates, there is significant variability in storage among them. Future investigation to elucidate the mechanistic differences between these two vectors include earlier study endpoints, RNA quantification, and/or single-cell analysis.

There are innumerable studies evaluating the impact of ERT and AAV-GAA in the Pompe mouse, however there are surprisingly few studies which provide a head-to-head comparison of these two therapeutic modalities^32,49. In Falk et al. 1e11 vg of AAV9-DES-GAA was delivered intravenously to 3-month-old mice which were evaluated at 6 months old³². In this study they found that AAV9-treated mice had improved cardiac function, respiratory function during eupnea, and ex vivo diaphragm strength which may be attributed to increased glycogen clearance in the heart and diaphragm. These mice were compared to mice treated with and 20mg/kg of ERT every other week for 3 months, which had some cardiac function and ex vivo diaphragm strength improvement, but little respiratory function improvements and minimal GAA activity leading to correlative glycogen clearance in the diaphragm. In this study, Gaa^-/- mice had reduced frequency due to increased expiratory time, which were normalized in AAV-treated mice, but with no significantly differences in tidal volume or minute ventilation. In our study, Gaa^-/- mice had elevated respiratory rates, which persisted across all treatment groups (vs WT: p<0.06). In our study, tidal volume was modestly affected with statistical deficits occurring only in the AAVcc47-treated mice compared to WT. Minute ventilation, a composite value of frequency and tidal volume, however revealed significant deficits in both untreated Gaa^-/- and AAVcc47-treated mice compared to WT mice, while AAV9- and ERT-treated mice showed improvements. In both untreated Gaa^-/- and AAVcc47-treated mice, deficits to both frequency and tidal volume contribute to deficits in minute ventilation, as where the frequency and tidal volume are more balanced out in the AAV9- and ERT-treated mice to improve minute ventilation. We also observed some expiratory-related deficits that were slightly improved with AAV treatment but not ERT.

In Lee et al. yfAAV9-SynI-GAA was administered to P0 Pompe mouse pups via a targeted intracerebroventricular injection⁴⁹. These mice were compared to Pompe mice, intravenously treated with 20 mg/kg rhGAA every 2 weeks starting at 3 months. As expected, ERT-treated mice had elevated GAA activity and reduced glycogen in limb muscle, but no impact in the brain or spinal cord. These findings are reversed in the AAV-treated mice due to the localized administration. Three months after treatment initiation, both ERT- and AAV-treated mice had improved minute ventilation during a hypercapnic respiratory challenge, a measure of overall maximal respiratory capacity, compared to untreated Pompe mice. Grip strength and inverted screen were also evaluated with limited improvements in all treated mice compared to untreated Pompe and WT mice.

Here, we add to the limited preclinical direct comparisons of ERT and gene therapy. A notable difference in our study is the dose for both ERT and gene therapy, which more closely mimics current clinical dosing. We confirm that AAV administration results in robust GAA expression in the medulla and cervical spinal, with concentrated expression in motor neurons, compared to a lack of detectable GAA in these areas from ERT-treated mice. AAV9-GAA, AAVcc47-GAA, and ERT were able to significantly clear glycogen in the heart with no significant differences among the three treatments. In the diaphragm and TA muscles there was significantly superior glycogen clearance in AAV-treated mice compared to ERT-treated mice. These findings in the diaphragm echo the findings in a previous study comparing AAV9 and ERT, with the exception that here we find some glycogen clearance as where they found no glycogen clearance³². The neuromuscular junction (NMJ), the site where motor neurons terminate at the muscle sarcolemma, is disrupted in the Pompe mouse diaphragm. Previous analysis of NMJs in the Pompe mouse TA following localized AAV-GAA treatment show maintained innervation. Here, we also find that at 7 months post-injection, AAV-GAA treated mice have improved innervation, while ERT-treated mice do not. These findings are not surprising given the lack of GAA in motor neurons and persistent glycogen in the diaphragm.

In contrast to the diaphragm and TA however, ERT-treated mice had increased glycogen clearance in the bladder compared to AAV-treated mice. Across other tissues including the tongue, lungs, and esophagus, there was some significant and non-significant decrease in glycogen accumulation in treated mice, none of which reached WT levels. Together, glycogen clearance through the range of tissues in AAV9- and ERT-treated mice resulted in improved maximal respiratory capacity, motor coordination, and limb strength.

Treatment of IOPD and LOPD patients

The impact of gene therapy and enzyme replacement therapy is influenced by the degree of pathology at treatment initiation as well as cross-reactive immunologic material (CRIM) status. All patients who are CRIM-negative have mutations resulting in no residual GAA production and all present as IOPD. Patients who are CRIM-positive have mutations that do produce some GAA protein, which may or may not be functional, and thus may present as either IOPD or LOPD. CRIM-negative and some CRIM-positive patients develop high sustained antibody titers (HSAT) against therapeutic GAA. Immune tolerance induction (ITI) and immunosuppression regimens are used to prevent HSAT. Here, we utilize transient, low-dose methotrexate in the ERT-treated mice to prevent adverse immune responses due to repeated GAA administration. To date, only 5 clinical trials evaluating AAV gene therapy in Pompe disease have been initiated in the United States. Of these studies, only one study included CRIM-negative, IOPD patients. One barrier to developing effective gene therapies for IOPD patients is the anticipated need for re-dosing later in life, which will require immune tolerance to both GAA and the AAV capsid. Preliminary data demonstrate that the modifications to AAV9 present in AAVcc47 do not lead to substantial changes in neutralizing antibody evasion, however a large scale non-human primate study is necessary to confirm this in vitro studies³¹. It is possible that if there is enough difference between AAV9 and AAVcc47, AAVcc47 could be used as a redosing vector subsequent to AAV9.

Impact of ERT and AAV gene therapy on smooth muscle and lung pathology

Without intervention children with Pompe disease had significantly reduced life expectancies; with the advent of ERT survival in these children has been prolonged. However, this has resulted in the unmasking of pathologies not previously observed. Throughout the last several years our group has described these novel pathologies including the impact of GAA deficiency in smooth muscle and alveolar epithelia in Pompe patients and in the Pompe mouse^6–8,40. Glycogen accumulation in the smooth muscle of the trachealis, esophagus, bladder, and vasculature results in trachea-bronchomalacia, gastroesophageal reflex and feeding difficulties, urinary incontinence, and dilative arteriopathy, respectively, all of which result in decreased quality-of-life and may be life-threatening. Many of these pathologies were not resolved in patients receiving initial dosing of ERT – 20mg/kg every other week with little to no immune management. Glycogen clearance in lung epithelium has also not previously been investigated in mice or patients receiving ERT and only one study has previously evaluated the impact of AAV-GAA on the airways³⁶. This previous study demonstrated a lack of glycogen clearance in the trachealis muscle and a subsequent lack of improved airway function. In the present study, we observed moderately elevated GAA activity and reduced glycogen clearance in the esophagi and bladders of AAV- and ERT-treated mice.

In addition, the upper airway complications presented in the genioglossus and trachealis, the alveolar sac, specifically alveolar type 1 and type 2 (AT1 and AT2) cells have considerable pathology. AT1 and AT2 cells have key functions within the alveoli including providing a gas exchange surface, producing surfactant, and serving as progenitor cells. Further assessment of AAV vector transduction profile at the single-cell level within the lung in different animal models may shed light into potential approaches to develop corrective gene therapy dosing regimens to address respiratory pathology in Pompe disease. Surfactant is a critical protein complex responsible for decreasing surface tension in the alveoli prevent collapse and atelectasis. Progressing Pompe patients often require non-invasive positive pressure ventilation (NIPPV) or invasive mechanical ventilation (IMV) to help expand the lungs and prevent such¹³. Here we observe that GAA activity is elevated in AAV- and ERT-treated lungs, resulting in ~50% glycogen clearance. Zooming in on GAA in AT2 cells specifically, ERT yielded an increase of 4.7x more GAA than untreated Gaa^-/- mice, and AAV-GAA yielded an increase of 6.3 – 7.7x more GAA than untreated Gaa^-/- mice. This increase of GAA resulted in significant decreases in LAMP1 aggregation indicating a reduction of the quantity and/or size of lysosomes accumulated.

In conclusion, these findings demonstrate that systemic GAA replacement results in glycogen clearance across the respiratory motor unit and in the alveoli which improves subsequent cellular pathology and ultimately preserves respiratory function.

HIGHLIGHTS:

• GAA restoration via AAV9-GAA, AAVcc47-GAA gene therapy, and enzyme replacement therapy clears moderate to significant levels of glycogen in respiratory skeletal muscle and the lungs.

• AAV-GAA gene therapy but not enzyme replacement therapy leads to GAA restoration in respiratory motor neurons and results in improved neuromuscular junction integrity.

• AAV-GAA gene therapy and enzyme replacement therapy restore GAA in alveoli to improved alveolar cellular architecture.

• AAVcc47-GAA was more effective than AAV9-GAA in targeting the respiratory muscles.