AAV-mediated gene therapy for sialidosis

Abstract

Sialidosis (mucolipidosis I) is a glycoprotein storage disease, clinically characterized by a spectrum of systemic and neurological phenotypes. The primary cause of the disease is deficiency of the lysosomal sialidase NEU1, resulting in accumulation of sialylated glycoproteins/oligosaccharides in tissues and body fluids. Neu1^−/− mice recapitulate the severe, early-onset forms of the disease, affecting visceral organs, muscles, and the nervous system, with widespread lysosomal vacuolization evident in most cell types. Sialidosis is considered an orphan disorder with no therapy currently available. Here, we assessed the therapeutic potential of AAV-mediated gene therapy for the treatment of sialidosis. Neu1^−/− mice were co-injected with two scAAV2/8 vectors, expressing human NEU1 and its chaperone PPCA. Treated mice were phenotypically indistinguishable from their WT controls. NEU1 activity was restored to different extent in most tissues, including the brain, heart, muscle, and visceral organs. This resulted in diminished/absent lysosomal vacuolization in multiple cell types and reversal of sialyl-oligosacchariduria. Lastly, normalization of lysosomal exocytosis in the cerebrospinal fluids and serum of treated mice, coupled to diminished neuroinflammation, were measures of therapeutic efficacy. These findings point to AAV-mediated gene therapy as a suitable treatment for sialidosis and possibly other diseases, associated with low NEU1 expression.

Graphical abstract

d'Azzo and colleagues describe a comprehensive, pre-clinical study in the mouse model of the lysosomal storage disease sialidosis, using an AAV-mediated gene therapy approach. Sialidosis is an orphan disease caused by genetic deficiency of the lysosomal sialidase NEU1. No cure is currently available for affected individuals.

Introduction

Neuraminidase 1 (NEU1) is essential for the hydrolysis of sialo-glycoconjugates in lysosomes, which the enzyme initiates by cleaving terminal sialic acid residues from their glycan chains. In mammalian tissues, NEU1 is found in a high-molecular-weight complex with two other lysosomal enzymes, the serine carboxypeptidase protective protein/cathepsin A (PPCA) and the glycosidase, β-galactosidase (β-Gal), encoded by the CTSA and GLB1 genes, respectively. Interaction with PPCA is a prerequisite for the proper localization, stability, and activation of NEU1. In absence of a functional PPCA, NEU1 is enzymatically silent.^1,2

Congenital deficiency of NEU1 in humans affects lysosomal catabolism of sialo-glycoconjugates, resulting in the lysosomal storage disease (LSD), sialidosis, or mucolipidosis I.^3,4 Patients with sialidosis are classified into two main subtypes based on the age of onset and severity of their symptoms, and residual enzyme activity. Type I sialidosis is the attenuated form of the disease with onset during adolescence, while type II sialidosis is the severe, neuropathic form of the disease, which can be further classified into congenital or hydropic, infantile, and juvenile forms.^4,5 Type I patients have normal intellectual ability and are mostly asymptomatic during childhood or early adolescence. They are typically diagnosed in the second decade of life, when they develop symptoms, such as myoclonus, gait abnormalities, and progressive visual loss, associated with bilateral macular cherry-red spots.^4,5 Type II patients with the acute congenital form of the disease present with facial edema and inguinal hernias, and can develop hydrops fetalis, neonatal ascites, or both, leading to stillbirth or death shortly after birth. The infantile/juvenile type II patients have coarse facial features, hepatosplenomegaly, dysostosis multiplex, skeletal dysplasia, macular cherry-red spots, hearing loss, angiokeratoma, myoclonus, cardiomyopathy, and mental retardation.^4,6 Lysosomal accumulation of sialylated glycoproteins and oligosaccharides in tissues and body fluids (oligosacchariduria) are diagnostic of both disease subtypes.

Neu1 knockout mice (Neu1^−/−) closely resemble type II sialidosis.⁷ They present with signs of the disease at birth, which are associated with severe growth retardation and, in a percentage of newborns, death at weaning.⁷ The disease progresses into a severe systemic and neurodegenerative condition, leading to premature death (4–6 months). During the course of their lifespan, affected mice show dysmorphic features, progressive edema, and oligosacchariduria.⁷ This sialidosis model has been extensively used to study mechanisms of pathogenesis and to test preclinical therapeutic modalities. During these studies, Neu1 was identified as a central negative regulator of the ubiquitous process of lysosomal exocytosis that Neu1 restrains by desialylating the lysosomal-associated membrane protein 1 (Lamp1), a substrate of Neu1, extending its half-life.^8,9,10 Loss of Neu1 activity leads to Lamp1 accumulation, which is associated with an increased number of Lamp1⁺ lysosomes docked at the plasma membrane (PM), poised to fuse with the PM and release their contents extracellularly. Exacerbation of lysosomal exocytosis has since been established as a central pathogenic mechanism in the Neu1^−/− mouse model, which likely reflects known clinical manifestations of sialidosis patients.^{10,11,12,13,14,15} For instance, excessive lysosomal exocytosis in the Neu1^−/− connective tissue has been linked to the development of muscle fibrosis responsible for muscle fiber degeneration and muscle atrophy.^13,15 A similar paradigm applied to the brain of Neu1^−/− mice was shown to contribute to the progressive deposition of β-amyloid resembling Alzheimer’s disease pathology and leading to neurodegeneration.¹¹

Currently, there is no target treatment or cure available for type I or type II sialidosis. Patients receive supportive care and symptomatic relief, primarily for the management of myoclonus and seizures.^4,5 We have previously tested different therapeutic approaches in various Neu1-deficient mouse models, including enzyme replacement therapy,¹⁶ chaperone-mediated gene therapy with PPCA,¹⁷ and treatment with pharmacological and dietary compounds.¹⁸

These encouraging results have set the stage for the current preclinical study exploring the efficacy of an AAV-mediated gene therapy approach for future treatment of sialidosis. A cohort of Neu1^−/− mice was intravenously (i.v.) injected with two self-complementary, recombinant AAV2/8 (scAAV) vectors, expressing human NEU1 and PPCA proteins. Treatment restored NEU1 activity, albeit to different extent, in most tissues tested, including the brain, muscles, and multiple visceral organs. This was accompanied by an overall improvement of tissue morphology, prevention of generalized fibrosis, and reversal of neuroinflammation and sialyl-oligosacchariduria. These findings highlight the suitability of this therapeutic approach for treating sialidosis and other conditions in which low expression of NEU1 may be a contributing factor to the disease.

Results

Biodistribution of AAV vectors and immune response in injected mice reflect positive therapeutic outcome

To investigate whether the administration of a single combined dose of scAAV2/8-CMV-NEU1 (2 × 10¹² vg/mouse) and scAAV2/8-CMV-CTSA (1 × 10¹² vg/mouse) (scAAV-N/P) by i.v. injection could revert or prevent the development of typical pathological manifestations in the sialidosis mice, we co-injected a cohort of 1-month-old Neu1^−/− mice that were sacrificed at 3 months post injection. Treated mice appeared almost indistinguishable from their control littermates: their dysmorphic facial features and progressive edema of the eyelids and paws, two prominent phenotypes of the disease, were largely resolved (Figure 1A).

Figure 1.

Gross appearance of Neu1^−/−-treated mice and tissue biodistribution of recombinant AAV vectors

(A) Representative images of edema of the eyelids and disheveled coat of Neu1^−/− mice at age 1 and 4 months, features that were corrected in scAAV-N/P-treated mice. (B and C) Vector genome copies of scAAV2/8-CMV-NEU1 (B) and scAAV2/8-CMV-CTSA (C) in multiple organs of injected mice at 3 months post injection, measured per diploid genome. n = 4; ovary: n = 3. (D and E) scAAV2/8-CMV-NEU1 (D) and scAAV2/8-CMV-CTSA (E) vector genome copies in cerebellum, brain stem, cortex, and hippocampus, measured per diploid genome. n = 2. See also Figure S1.

We then evaluated vector biodistribution by measuring the copy number of recombinant scAAV2/8-CMV-NEU1 and scAAV2/8-CMV-CTSA in total genomic DNA extracted from multiple organs of treated mice, and calculated vector genome (vg) copies per diploid genome. We found that the highest NEU1 viral load (average of ∼43 copies per diploid genome) was detected in the liver, followed by the lung, kidney, spleen, tibialis anterior, and heart (Figure 1B). The other tissues tested had ∼20–40× lower amounts of viral genome compared with the liver, with the bone marrow (BM) containing the lowest copy number (Figure 1B). Testing the viral load of CTSA in the same tissues showed a different biodistribution pattern compared with NEU1, with the highest viral load in the liver and heart (average of ∼3.25 and ∼2.9 copies per diploid genome), followed by the lymph nodes, lung, kidney, peritoneum, spleen, and diaphragm (Figure 1C). The remaining tissues had a ∼25–100× lower amount of CTSA viral genome compared with the liver and heart, with the lowest amount found in the BM (average of ∼0.014 copies per diploid genome) (Figure 1C). Notably, we also detected a measurable vg copy number for both NEU1 and CTSA in different regions of the brain, such as the cerebellum, brain stem, cortex, and hippocampus (Figures 1D and 1E), indicating that both viruses had crossed the blood-brain barrier. In all tissue tested, however, the average copy number for the CTSA vector was considerably lower than that for the NEU1 vector, which could be in part due to the double dose of NEU1 vector injected (Figures 1B–1E). We also tested whether treated mice would mount an adaptive immune response toward the vector (AAV2/8) and/or the transgene product (NEU1 and PPCA). For this purpose, we tested serum samples from injected mice and controls at age 4 months using ELISA. No specific antibodies for human NEU1 or PPCA were detected in any of the samples tested (Figures S1A and S1B). However, we did measure antibodies against the viral capsid in injected samples (Figures S1C and S1D), which were probably due to the use of a double-stranded self-complementary AAV vector, rather than a single-stranded vector, as previously reported.¹⁹

Expression of therapeutic NEU1 and PPCA is detected in different brain regions

It is well established that the brain is a difficult tissue to treat using gene therapy following a systemic injection route due to the presence of the blood-brain barrier.²⁰ Therefore, we wanted to determine whether the expression of the transgene products in the aforementioned brain regions of treated mice would parallel the biodistribution of the two AAV vectors. We first tested the enzyme activities of NEU1 and PPCA in lysates of the cerebellum, brain stem, cortex, and hippocampus from scAAV-N/P-injected mice. Slightly increased NEU1 activity was measured in the brain stem, cortex, and cerebellum of the treated mice compared with the Neu1^−/− non-injected controls, although the values did not reach statistical significance (Figure 2A). Cathepsin A (CA) activity instead followed a different pattern. As we have previously reported in the fibroblasts of patients with sialidosis, NEU1 deficiency is associated with a variable increase in CA activity.¹⁸ We now confirmed this paradigm in the Neu1^−/− mice, where significantly increased activities for both CA and β-Gal, the third member of the lysosomal multienzyme complex, were measured in the different regions of the Neu1^−/− brain, highest in the hippocampus, compared with the WT controls (Figures 2B and 2C). Interestingly, the hippocampus, one of the early and most affected areas of the brain in the Neu^−/− mice, showed the most significant decrease (35%) in CA activity, whereas the other three regions were unchanged (Figure 2B). Instead, β-Gal activity was reduced in the four brain regions of injected mice (Figure 2C), confirming the dependency of β-Gal on the localization and activity levels of NEU1.²¹ Two other lysosomal enzymes, β-hexosaminidase (β-Hex) and α-mannosidase (α-Man), known to be influenced by the deficiency of Neu1, were also used as a proxy method to assess the efficacy of treatment. These enzyme activities were significantly increased in all regions of the brain of Neu^−/− mice compared with WT controls, but their values were again reduced in the same regions of the injected mice (Figures S2A and S2B). These data suggest that only a small increase in NEU1 activity in the brain of scAAV-N/P i.v.-injected mice was sufficient to improve the altered activities of CA and other lysosomal enzymes.

Figure 2.

Enzyme activities and expression of NEU1 and PPCA in different brain regions of scAAV-N/P-treated Neu1^−/− mice

(A–C) Enzyme activity levels for NEU1 (A), cathepsin A (CA) (B), and β-Gal (C) in the hippocampus (Hip), brain stem (BS), cortex (CX), and cerebellum (CB) of WT (n = 4), Neu1^−/− (n = 4) and scAAV-N/P-treated (n = 3) mice. (D and E) Co-IHC of NEU1 and PPCA in the CA3 region of treated mice. Scale bar, 50 μm. Right panels are zoomed-in images of boxed areas. Scale bar, 10 μm. Brown and purple puncta/staining depict lysosomal localization of PPCA and NEU1, respectively. (F and G) Co-IHC of NEU1 and PPCA in the Purkinje cells of the cerebellum. Scale bar, 50 μm. Right panels are zoomed-in images of boxed areas. Scale bar, 10 μm. Brown and purple puncta depict lysosomal localization of PPCA and NEU1, respectively. Values are expressed as mean ± SD. Statistical analyses were performed using Brown-Forsythe and Welch ANOVA. ∗ Indicates significance between WT and Neu1^−/− or scAAV-N/P-treated mice, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. ^# Indicates significance between Neu1^−/− and scAAV-N/P-treated mice, ^#p < 0.05. See also Figure S2.

Using either single immunohistochemistry (IHC) or combined IHC (co-IHC) and co-immunofluorescent (co-IF) staining with anti-PPCA and anti-NEU1 antibodies, we could detect expression of NEU1 and PPCA proteins in sparse neurons of the CA3 region of the hippocampus and in Purkinje cells of the cerebellum, but only a few of these cells co-expressed both proteins (Figures 2D–2G and S2C–S2G). As a measure of therapeutic efficacy, we next tested the pattern and level of Lamp1 expression in the brain of treated and untreated mice, compared with WT controls. IHC with Lamp1 antibody, followed by quantification of DAB⁺ areas (represented as percentage) showed a substantial increase in Lamp1 staining in neurons and glia of the Neu^−/− hippocampus and cerebellum (Figures S3A–S3D), which was significantly reduced to almost control levels in both regions of the brain of the injected mice (Figure S3A–S3D).

Reversal of histopathology and sustained expression of NEU1 and PPCA occur in the choroid plexus of scAAV-treated Neu1^−/− mice

Epithelial/endothelial cells and macrophages, including microglia, are primary targets of pathogenesis in sialidosis.^4,10,11 CP epithelial cells follow this pattern, showing signs of disease, i.e., numerous storage-filled lysosomes, early in the life of Neu1^−/− mice (Figure S3E). After scAAV-N/P treatment, these cells were fully corrected, as seen on H&E-stained brain sections, and the CP of the treated mice appeared morphologically indistinguishable from that of the WT mice (Figure S3E). Sustained expression of both NEU1 and PPCA was detected in virtually all cells of the CP, although only some of them showed strong co-localization of the two proteins by both co-IHC and co-IF with NEU1 and PPCA antibodies (Figures 3A–3C and S3F–S3H). Furthermore, IHC with Lamp1 antibody revealed normalization of Lamp1 levels (Figures 3D and 3E). This finding was particularly relevant because Lamp1 accumulation downstream of Neu1 deficiency is paralleled by excessive lysosomal exocytosis.^10,11 To measure the extent of lysosomal exocytosis in treated and untreated mice, we tested the activity of β-Hex in their cerebrospinal fluids (CSFs) and sera. We found that the increased β-Hex activity in both samples of the Neu^−/− mice was significantly decreased after treatment, confirming that the restoration of NEU1 activity in treated mice reduced Lamp1 levels and decreased lysosomal exocytosis (Figures 3F and 3G). We also tested the levels of Aβ-42 in the CSF, which is a component of the amyloidosis process that occurs in the Neu1^−/− mice.¹¹ A significant decrease in the levels of Aβ-42 was measured in the CSF of treated mice (Figure 3H). However, correction of this phenotype was apparently not sufficient to prevent the formation of amyloid deposits in scAAV-N/P-injected mice, likely due to the low NEU1 expression and activity in neurons (Figure S4A).

Figure 3.

Co-expression of NEU1 and PPCA in the CP of scAAV-N/P-treated mice reduces Lamp1 levels, lysosomal exocytosis in the CSF and serum, and brain inflammatory marker CD68

(A) Co-IHC of NEU1 and PPCA in the choroid plexus (CP) epithelial cells of treated mice. Scale bar, 25 μm. Lower panels are zoomed-in images of boxed areas. Scale bar, 10 μm. Brown and purple puncta/staining depict lysosomal localization of PPCA and NEU1, respectively. (B) Maximum intensity projections created from z-stacked co-IF staining of CP epithelium with anti-NEU1 (yellow) and anti-PPCA (red) antibodies. DAPI was used to stain the nuclei. Scale bar, 10 μm. (C) 3D reconstruction of 59 slice z-stacked (approx. 10 μm) region of interest from IF images of CP epithelial cells stained with anti-NEU1 (yellow) and anti-PPCA (red) antibodies and DAPI (blue). Scale bar, 5 μm. (D) Representative IHC images of the choroid plexus from WT, Neu1^−/−, and scAAV-N/P-treated mice, using anti-Lamp1 antibody. Scale bar, 25 μm. (E) Quantification of Lamp1⁺ area (%), shown in 3D, using 2–8 planes per tissue. WT (n = 8), Neu1^−/− (n = 5), and scAAV-N/P-treated (n = 7) mice. (F) β-Hex enzyme activity measured in the CSF from WT (n = 4), Neu1^−/− (n = 6), and scAAV-N/P-treated (n = 4) mice. (G) β-Hex enzyme activity measured in sera from WT (n = 7), Neu1^−/− (n = 5), and scAAV-N/P-treated (n = 7) mice. (H) Levels of Aβ-42 exocytosed into the CSF: WT (n = 4), Neu1^−/− (n = 6), and scAAV-N/P-treated (n = 4) mice. (I) Quantification of CD68⁺ area (%), shown in 3G, using 2–8 planes per tissue: WT (n = 10), Neu1^−/− (n = 6), and scAAV-N/P-treated (n = 11) mice. (J) Representative IHC images of the CP from WT, Neu1^−/−, and scAAV-N/P-treated mice using anti-CD68 antibody. Scale bar, 25 μm. Values are expressed as median ± quartiles for (E) and (I) or mean ± SD for (F–H). Statistical analyses were performed using one-way ANOVA or Brown-Forsythe and Welch ANOVA. ∗ Indicates significance between WT and Neu1^−/− or scAAV-N/P-treated mice, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. ^# Indicates significance between Neu1^−/− and scAAV-N/P-treated mice, ^#p < 0.05, ^##p < 0.01, ^###p < 0.001. See also Figure S3 and S4.

Reduction of microgliosis in the brain and choroid plexus of scAAV-treated Neu1^−/− mice

Microgliosis is a pathologic feature of the Neu1^−/− mice.⁷ To assess the status of microglia activation before and after AAV injection we performed IHC of brain sections from WT, Neu1^−/−, and scAAV-N/P mice with anti-CD68 antibody. CD68 is a widely used, specific marker of pro-inflammatory, activated monocytes, which include microglia. The Neu1^−/− mice showed an increase in microglial CD68 staining in the CA3 region of the hippocampus and the cerebellum compared with the WT controls, which did not have detectable CD68⁺ cells (Figures S4B–S4E). Mice treated with scAAV-N/P showed a significantly reduced CD68 staining at 3 months post injection (Figures S4B–S4E), as determined by quantification of DAB⁺ areas (represented as percentage). Similarly, the numerous CD68⁺ microglia that were evident in the CP of Neu1^−/− mice were no longer detectable in this structure of the treated mice (Figures 3I and 3J).

Together these results indicate that a single i.v. co-injection of scAAV-N/P leads to sufficient expression of NEU1 to partially restore lysosomal function and ameliorate some of the neuropathological and neuroinflammatory characteristics of the Neu1^−/− mice.

Restored NEU1 activity in treated Neu1^−/− mice reverses fibrosis in skeletal muscle and heart

Extensive and progressive expansion of the connective tissue in skeletal and heart muscles is the cause of muscle fiber degeneration, muscle atrophy,^13,15 and heart fibrosis in the Neu1^−/− mice. This phenotype has been linked to the effects of Neu1 deficiency on fibroblasts, which retain a partially differentiated and activated state, characterized by increased Lamp1 levels, and excessive exocytosis of lysosomal contents and exosome that propagate fibrotic signals to unaffected areas and bolster disease progression.¹³ For the evaluation of the muscle phenotype before and after treatment, we first tested the expression of the NEU1 and PPCA proteins in skeletal muscle and heart tissue sections by co-IHC. Sustained co-expression of NEU1 and PPCA was detected in most, if not all, cardiomyocytes of treated mice, but the level of expression of both proteins appeared to vary depending on the fiber type (Figures 4A and S5A). Similarly, in the gastrocnemius (GA) muscle, positive co-staining of NEU1 and PPCA was observed in most of the fibers, but again the levels of the two proteins differed in different fibers (Figures 4B and S5B).

Figure 4.

NEU1 and PPCA protein expression in heart and skeletal muscle corrects the activity of other lysosomal enzymes in Neu1^−/− mice treated with scAAV-N/P

(A and B) Representative co-IHC images of the heart (A) and gastrocnemius (GA) muscle (longitudinal) (B) from scAAV-N/P-treated mice using anti-human NEU1 and anti-human PPCA antibodies. Scale bar, 25 μm. Brown and purple puncta/staining depict lysosomal localization of PPCA and NEU1, respectively. (C–H) Enzyme activity levels measured in the heart and GA skeletal muscle from WT, Neu1^−/−, and scAAV-N/P-treated mice: NEU1 (C and D), CA (E and F), and β-Gal (G and H); n = 6. Values are expressed as mean ± SD. Statistical analyses were performed using the one-way ANOVA. ∗ Indicates significance between WT and Neu1^−/− or scAAV-N/P-treated mice, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. ^# Indicates significance between Neu1^−/− and scAAV-N/P-treated mice, ^###p < 0.001, ^####p < 0.0001. See also Figure S5.

These results were further corroborated by assaying NEU1 and PPCA enzyme activities in the heart and GA muscle of scAAV-N/P-injected mice (Figures 4C–4F). Very high NEU1 and CA activities were measured in the heart tissue of the treated mice (Figures 4C and 4E), probably due to the high transduction efficacy of cardiomyocytes.^22,23 In the GA muscle of treated mice, NEU1 activity was lower than in the heart but also robustly increased compared with untreated mice and WT controls (Figure 4D). As in other tissues, the increased CA activity in untreated Neu1^−/− mice was reduced in the GA muscle of injected mice (Figure 4F). In these tissues of the Neu1^−/− mice, β-Gal, β-Hex, and α-Man activities were significantly higher (∼5 to ∼25×) compared with the WT controls, but the levels of these enzymes were normalized after treatment (Figures 4G, 4H, and S5C–S5F).

Correction of Neu1 activity was accompanied by clear improvement of the fibrosis phenotype in both the heart and GA muscle of treated Neu1^−/− mice.^13,15 Extensive vacuolization of fibroblasts in the expanded connective tissue at the attachment site of the heart valves, including the mitral valve, and the GA muscle was prominent in the Neu1^−/− mice, but appeared to be fully corrected in the treated mice (Figures S6A and S6B).

Normalization of tissue morphology in scAAV-N/P-treated mice was confirmed by staining heart and GA muscle sections with Masson’s trichrome. Treatment prevented the extensive collagen deposition that was instead evident in both tissues of the untreated Neu1^−/− mice (Figures S6C and S6D). In addition, immunostaining of heart and GA muscle with anti-Lamp1 and anti-CD68 antibodies, followed by quantification of DAB⁺ areas (represented as percentage), showed a significant reduction in the number of Lamp1⁺ and CD68⁺ cells in treated mice compared with the Neu1^−/− mice (Figures 5A–5H and S6E).

Figure 5.

Reduced Lamp1 and CD68 levels in scAAV-N/P-treated mice

(A and B) Normalization of Lamp1 staining in the heart (A) and skeletal muscle (B). Scale bar, 50 μm. (C and D) Quantification of Lamp1⁺ areas (%), shown in (A) and (B), using 2–10 planes per tissue. Heart: WT (n = 7), Neu1^−/− (n = 7), and scAAV-N/P-treated (n = 6) mice. GA muscle: WT (n = 9), Neu1^−/− (n = 8), and scAAV-N/P-treated (n = 8) mice. (E and F) Reduction of the inflammatory marker CD68 in both muscle tissues; heart (E) and skeletal muscle (F). Scale bars, 100 and 50 μm. (G and H) Quantification of CD68⁺ areas (%), shown in (E) and (F), using 2–10 planes per tissue. Heart: WT (n = 10), Neu1^−/− (n = 8), and scAAV-N/P-treated (n = 8) mice. GA muscle: WT (n = 8), Neu1^−/− (n = 8), and scAAV-N/P-treated (n = 7) mice. Values are expressed as median ± quartiles. Statistical analyses were performed using the one-way ANOVA. ∗ Indicates significance between WT and Neu1^−/−, or scAAV-N/P-treated mice, ∗∗∗∗p < 0.0001. ^# Indicates significance between Neu1^−/− and scAAV-N/P-treated mice, ^####p < 0.0001. See also Figure S6.

Partial correction of kidney morphology in treated mice is accompanied by reversal of sialyl-oligosacchariduria

We next examined the effects of scAAV-N/P injection in the kidney, which is difficult to treat and one of the primary affected organ of sialidosis in children.⁴ Co-IHC of NEU1 and PPCA in kidney sections of mice injected with scAAV-N/P showed expression of both proteins in the proximal renal tubules, the glomeruli, and the endothelial cells of the blood vessels (Figures 6A and 6B). Co-localization of NEU1 and PPCA proteins was mostly seen in the glomeruli and endothelial cells of the blood vessels, while the proximal tubular epithelium expressed primarily NEU1 (Figures 6A and 6B). In agreement with the IHC results, we also measured a 2-fold increase in NEU1 activity in kidney tissue lysates from treated mice compared with Neu1^−/− samples (Figure 6C). The slightly increased CA activity in the Neu1^−/− kidney was reduced to WT levels after treatment (Figure 6D). Similarly, β-Gal activity, which was ∼3× higher than WT controls in untreated mice, was significantly reduced in injected mice (Figure 6E). Normalization of enzyme activities was also observed for β-Hex and α-Man, a finding consistent with the results obtained in other tissues (Figures S7A and S7B).

Figure 6.

Expression of therapeutic NEU1 and PPCA in the kidney of treated mice normalizes other enzyme activities, resolves sialyl-oligosacchariduria, and reduces Lamp1 and CD68 levels

(A and B) Representative co-IHC images of the kidney from scAAV-N/P-treated mice using anti-human NEU1 and anti-human PPCA antibodies. Brown and purple puncta depict lysosomal localization of PPCA and NEU1, respectively. Scale bars, 50 and 25 μm. (C–E) Enzyme activity levels measured in the kidneys of WT, Neu1^−/−, and scAAV-N/P-treated mice: NEU1 (C), CA (D), and β-Gal (E) (n = 6). (F) Total sialic acid levels measured in the urine of WT, Neu1^−/−, and Neu1^−/−scAAV-N/P mice at age 1, 2, and 4 months (n = 8). (G–J) Representative IHC images of the kidney from WT, Neu1^−/−, and scAAV-N/P-treated mice, stained with anti-Lamp1 (G) and anti-CD68 (I) antibodies. Scale bar, 50 μm. Quantification of Lamp1⁺ (H) and CD68⁺ (J) areas (%) shown in (G) and (I), using 2–8 planes per tissue. Lamp1: WT (n = 9), Neu1^−/− (n = 4), and scAAV-N/P-treated (n = 7) mice; CD68: WT (n = 12), Neu1^−/− (n = 7), and scAAV-N/P-treated (n = 10) mice. Values are expressed as mean ± SD for (C)–(F) or median ± quartiles for (H) and (J). Statistical analyses were performed using the one-way ANOVA. ∗ Indicates significance between WT and Neu1^−/−, or scAAV-N/P-treated mice, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗∗p < 0.0001. ^# Indicates significance between Neu1^−/− and scAAV-N/P-treated mice, ^#p < 0.05, ^##p < 0.01, ^###p < 0.001, ^####p < 0.0001. See also Figure S7.

Histopathology of kidney sections stained with H&E showed progressive vacuolization of the proximal tubular epithelium and the glomeruli in Neu1^−/− mice at age 1 and 4 months (Figures S7C and S7D). scAAV-N/P injection prevented disease progression, as demonstrated by the significant reduction of vacuolization in the tubular epithelium of treated mice compared with untreated Neu1^−/− mice (Figures S7C and S7D). As a further indication of partially restored kidney function in treated mice, we tested the levels of total sialic acids in urine samples collected from age-matched WT, Neu1^−/−, and injected mice. Increased sialyl-oligosacchariduria was measured in samples of Neu1^−/− mice at all ages (Figure 6F). This characteristic phenotype was fully resolved in scAAV-N/P-injected mice at 2 and 4 months after treatment (Figure 6F), despite tissue morphology still showing signs of the disease.

We also stained kidney sections from WT, Neu1^−/−, and scAAV-N/P-treated mice for Lamp1. In the WT kidney, Lamp1⁺ staining was only detected in the proximal tubules, whereas in the Neu1^−/− tissue both the proximal tubules and the glomeruli showed significantly increased Lamp1 staining, as quantified by DAB⁺ areas (represented as percentage) (Figures 6G and 6H). After scAAV-N/P treatment, Lamp1 levels were reduced by ∼50% compared with the levels in the Neu1^−/− tissue (Figures 6G and 6H). We also performed IHC of kidney sections from the same set of mice with anti-CD68 antibody, followed by quantification of DAB⁺ staining (Figures 6I and 6J). Unlike untreated mice that had a significant increase in the number of CD68⁺ cells, especially surrounding the glomeruli, no or few CD68⁺ cells were detected in the kidney of treated mice, indicating reversal of the inflammatory response also in this organ (Figures 6I and 6J).

Full correction of histopathology in the liver and spleen of treated Neu1^−/− mice

Similar positive outcome post treatment was observed in the liver and spleen of scAAV-N/P-injected mice compared with the untreated Neu1^−/− mice. Co-IHC of liver tissue from injected mice showed a very prominent co-localization of NEU1 and PPCA staining in hepatocytes, Kupfer cells, and sinusoidal endothelial cells (Figure 7A). Variation in the level of expression of both proteins was observed throughout the tissue. A similar co-IHC result was observed in the spleen of treated mice, which showed the highest number of splenocytes expressing or co-expressing NEU1 and PPCA in the red pulp (Figure 7B). Also in this tissue, variation was evident in the level of expression of the two proteins in different regions and cell types. Notably, treated mice did not develop splenomegaly (Figures S7E and S7F), a characteristic of Neu1^−/− mice that is linked to time-dependent extramedullary hematopoiesis.¹⁰ NEU1 activity in the liver of scAAV-N/P-injected mice was 5× higher than that in WT controls (Figure 7C), while in the spleen no significant changes in activity were measured after treatment (Figure 7D). CA activity, which was significantly increased in both liver and spleen of the Neu1^−/− mice (Figures 7E and 7F), was normalized after treatment in the spleen (Figure 7F) but remained high in the liver (Figure 7E). This could be due to the high viral load for both NEU1 and CTSA in the liver of treated mice and the sustained expression of the PPCA protein in the hepatocytes. In addition, the activities of β-Gal, β-Hex, and α-Man were restored to WT levels in both liver and spleen of scAAV-N/P-injected mice (Figures 7G, 7H, and S8A–S8D). At the histopathological levels, H&E-stained sections of the Neu1^−/− liver showed severe and widespread vacuolization of the hepatocytes, the epithelial cells (cholangiocytes) of the bile ducts, the endothelial cells of the sinusoids, and the Kupffer cells (Figure S9A). Liver histopathology was fully corrected in treated mice (Figure S9A). Similarly, in the Neu1^−/− spleen, the large areas of the red pulp showing extensive vacuolization of splenocytes appeared cleared of storage after treatment (Figure S9B).

Figure 7.

Co-expression of NEU1 and PPCA in liver and spleen of treated mice normalizes other enzyme activities and reduces Lamp1 and CD68 levels

(A and B) Representative co-IHC images of the liver (A) and spleen (B) from scAAV-N/P-treated mice using anti-human NEU1 and anti-human PPCA antibodies. Brown and purple puncta/staining depict lysosomal localization of PPCA and NEU1, respectively. Scale bar, 25 μm. Right panels are zoomed-in images of boxed areas. Scale bar, 10 μm. (C–H) Enzyme activity levels measured in the liver and spleen of WT, Neu1^−/−, and scAAV-N/P-treated mice: NEU1 (C and D), CA (E and F), and β-Gal (G and H) (n = 6). (I–L) Representative IHC images of the liver (I) and spleen (K) from WT, Neu1^−/−, and scAAV-N/P-treated mice stained with anti-Lamp1 antibody. Scale bar, 50 μm. Quantification of Lamp1⁺ areas (%), shown in (J) and (L), using 2–8 planes per tissue. WT (n = 9), Neu1^−/− (n = 4), and scAAV-N/P-treated (n = 8) mice. (M–P) Representative IHC images of the liver (M) and spleen (O) from WT, Neu1^−/−, and scAAV-N/P-treated mice stained with anti-CD68 antibody. Scale bar, 50 μm. Quantifications of CD68⁺ areas (%), shown in (N) and (P), using 2–8 planes per tissue. WT (n = 10), Neu1^−/− (n = 7), and scAAV-N/P-treated (n = 10) mice. Values are expressed as mean ± SD for (C–H) or median ± quartiles for (J), (L), (N), and (P). Statistical analyses were performed using the one-way ANOVA. ∗ Indicates significance between WT and Neu1^−/− or scAAV-N/P-treated mice, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. ^# Indicates significance between Neu1^−/− and scAAV-N/P-treated mice, ^#p < 0.05, ^###p < 0.001, ^####p < 0.0001. See also Figure S8 and S9.

As seen in other organs, the liver of untreated Neu1^−/− mice showed a significantly increased Lamp1⁺ immunostaining in the hepatocytes, and most prominently in cells surrounding the bile duct and hepatic artery (Figures 7I and 7J). In the spleen, intense Lamp1 immunostaining was observed in areas of the red pulp that contained the most affected cells (Figures 7K and 7L). After treatment, Lamp1 levels in both liver and spleen were comparable with those in WT tissues (Figures 7I–7L). Quantifications were made by evaluating the areas of positive DAB, represented as percentage.

Next, we tested whether the inflammatory marker CD68 would also be corrected. Indeed, quantification of DAB⁺ areas showed that the liver and spleen of scAAV-N/P-injected mice had a significantly reduced number of CD68⁺ cells, which was indicative of diminished or absent inflammation (Figures 7M–7P).

Prevention of generalized fibrosis in scAAV-treated Neu1^−/− mice

As a final assessment of the extent of correction of the disease phenotype in scAAV N/P-injected mice, we also performed Masson’s trichrome staining of the kidney, liver, and spleen and compared it with the staining of these tissues in untreated Neu1^−/− mice and WT controls. As described previously,¹³ tissue sections of the Neu1^−/− kidney, stained with Masson’s trichrome, showed the presence of numerous collagen-positive cells and increased connective tissue surrounding the glomeruli (Figure S9C). A similar staining pattern was observed in sections of the Neu1^−/− liver that revealed collagen-positive cells surrounding the bile ducts and the hepatic artery, as well as prominent collagen staining in the areas of connective tissue lining the central hepatic vein (Figure S9D). In the Neu1^−/− spleen the area with increased collagen was the red pulp (Figure S9E). At 3 months after treatment, no fibrosis was detected in the kidney, liver, and spleen that appeared morphologically identical to WT controls (Figures S9C–S9E). These results confirmed the occurrence of a pathogenic generalized fibrotic condition in the connective tissue of Neu1^−/− mice that can be used as a parameter to assess therapeutic efficacy after treatment.

Discussion

AAV-mediated gene therapy has emerged as a reliable in vivo gene transfer system for LSDs, given its efficacy of infection, targeting of both dividing and non-dividing cells, broad tropism, and safety. Upon successful infection, the genome of rAAV vectors inefficiently integrates into the host DNA and remains mostly episomal, facilitating expression of the encoded gene by the host’s transcription and translation machineries.²⁴ There are multiple serotypes of the virus with distinct tissue tropism, which have been further engineered to either target specific cell populations or ubiquitously infect multiple cell types providing a more systemic effect.²⁵ Furthermore, the use of cell-specific promoters restricts the expression of the transgene to only certain tissues or cells. The AAV serotype 2 was the first to be vectorized and several pseudo-types have been generated with cell- or tissue-specific tropism.^26,27 These vectors achieve stable transgene expression, and have been extensively employed in pre-clinical studies on various disease models, as well as in clinical trials.^{28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46} Several AAV-mediated gene therapy clinical trials, which have been approved by the Food and Drug Administration and the European Medicines Agency,²⁵ are currently available for different LSDs, including Gaucher disease, GM1- and GM2-gangliosidoses, Krabbe disease, several mucopolysaccharidoses (I, II, IIIA, B, C, D, VII), metachromatic leukodystrophy, Pompe disease, Batten disease (neuronal ceroid lipofuscinoses; CLN1-8), Fabry disease, and Danon disease.^{20,47,48,49,50,51}

Although long-term studies in patients are still ongoing, the positive results of AAV-mediated gene therapy in other LSDs so far, have encouraged the use of this approach also for the treatment of sialidosis type I and II, which currently has no cure. Previously, we used a recombinant AAV2/8 vector with expression of human PPCA restricted to the liver (scAAV2/8-LP1-PPCA) to treat Neu1^−/−;NEU1^V54M mice, a model of non-neuropathic type I sialidosis with residual NEU1 activity.¹⁷ High expression of PPCA in the liver led to efficient uptake of a circulating PPCA precursor by various systemic organs, as early as 1 month post injection.¹⁷ Increased levels of PPCA over the endogenous enzyme increased the activity of the mutant human NEU1 in most tissues, leading to reduced vacuolization, particularly in the kidney, and resolution of sialyl-oligosacchariduria. However, the circulating PPCA precursor was unable to cross the blood-brain barrier.¹⁷

In this study, systemic injection of Neu1^−/− mice with two AAV vectors expressing NEU1 and PPCA, under the control of a ubiquitous promoter, was designed with the idea that the localization, stability, and activity of therapeutic human NEU1 would be enhanced by co-expression of the human PPCA rather than the endogenous murine enzyme.^1,2,52 Several parameters confirm the validity of this therapeutic approach for treating sialidosis. In addition to improving the overall gross appearance of the mice, AAV treatment regressed the edema phenotype, corrected lysosomal vacuolization of cells in peripheral organs, and prevented sialyl-oligosacchariduria. Moreover, the severe fibrosis, affecting the heart, skeletal muscle, liver, kidney, and spleen of the Neu1^−/− mice, was also completely resolved, probably in part due to decreased lysosomal exocytosis from cells of the connective tissue. Remarkably, we also detected a measurable copy number of both viruses in different brain regions of i.v.-injected mice, which resulted in co-expression of NEU1 and PPCA in sparse neurons of the hippocampus and cerebellum, and decreased expression of a neuroinflammatory marker in microglia. Most relevant for the CNS phenotype linked to excessive lysosomal exocytosis downstream of Neu1 deficiency, has been the effect of treatment in the CP, a crucial supporting structure for the brain parenchyma and the major source of CSF production.^53,54 The sustained expression of NEU1 and PPCA in the CP epithelium and the reversal of the CP histopathology likely contribute to the decreased exocytosis of lysosomal contents, including Aβ-42, measured in the CSF of treated mice. Nevertheless, treatment did not resolve or prevent the progressive deposition of β-amyloid, a prominent phenotype of the Neu1^−/− mice. This partial correction could be attributed to the low levels of NEU1 expressed by neurons in the brain parenchyma after systemic i.v. injection, but further studies are needed to clarify this point. However, the outcome of this therapeutic approach in the brain and CP of treated mice suggests that it has the potential to ameliorate some of the neuropathological characteristics of sialidosis in patients.^4,5,55 A recent study by Hwu et al.,⁵⁶ which corroborates our results, describes the use of intracerebroventricular or facial vein injection of an AAV9 bicistronic vector, co-expressing human NEU1 and PPCA, in a different knockout mouse model of sialidosis.⁵⁶ This therapeutic approach was shown to reduce astrogliosis and Lamp1 accumulation in the nervous system and to improve motor function and survival of the treated mice. The latter study, together with our findings, support the notion that combining the routes of AAV injections could be effective in treating both type I and type II sialidosis.

We have reported earlier that Neu1^−/− mice develop a generalized fibrosis affecting multiple organs. In the muscle, this phenotype is associated with severe muscle atrophy, which could be the cause of the muscle weakness reported in some patients with sialidosis.^4,13,15 In the absence of NEU1, fibrosis is driven by excessive release of exosomes, loaded with profibrotic signaling molecules, by activated fibroblasts/myofibroblasts, leading to muscle degeneration, and the activation of an inflammatory response.^12,13,15 We now show that systemic i.v. injection of scAAV-N/P in Neu1^−/− mice normalizes the levels of Lamp1 in their connective tissue, which is paralleled by decreased exocytosis of lysosomal enzymes in the serum of these mice, and reduced collagen deposition. Therefore, we can infer that the reversal of the generalized fibrosis phenotype in Neu1^−/−-treated mice is due, at least in part, to the decreased exocytic activity of fibroblasts and the reduction of inflammation.¹³ This could be particularly relevant for patients with sialidosis type II who may develop a heart condition characterized by heart murmur,^6,57,58 dilated ventricles and atria, decreased ventricular function,^{6,58,59,60,61,62,63,64} and abnormality of the mitral valves^6,58,62,65 that may lead to heart failure. The restoration of connective tissue morphology in the heart of Neu1^−/− mice after treatment suggests that an AAV-mediated gene therapy approach may prevent heart involvement in type II sialidosis patients. Furthermore, in the adult human population downregulation of NEU1 may represent a risk factor for the initiation and/or progression of idiopathic forms of fibrosis.¹³ In these cases, increasing the endogenous levels of NEU1 may be beneficial to prevent or stop the progression of the disease, although more studies are needed to demonstrate the involvement of NEU1 in human forms of fibrosis.

As it is the case for other LSD patients, treatment of sialidosis should start as soon as the patient is diagnosed to achieve the maximum effect of halting or preventing disease progression and the development of debilitating neurological manifestations. However, type I patients are often misdiagnosed or diagnosed when some of their clinical symptoms are already apparent. Nevertheless, AAV-mediated gene therapy for this category of patients may still be beneficial even if started later in life, although therapeutic efficacy in these cases may depend on the type of mutations and severity of the symptoms at the time of treatment.

In conclusion, the overall success of these preclinical studies justifies the development of this therapeutic approach for the treatment of sialidosis, although future studies comparing the gene expression profiles of treated mice with those of WT and Neu1^−/− mice will allow us to fully address the functional implications of this type of therapy. Thus, additional long-term pre-clinical studies may be required prior to initiating a clinical trial for patients with this disease. Lastly, given the beneficial effects of repurposed drugs for ameliorating some of the clinical signs in type I sialidosis patients, a controlled diet and medication regimen should be taken into consideration in combination with target therapy, including gene therapy, for this disease.^{17,18,66,67,68,69}

Materials and methods

Animals

Animals were housed in a fully AAALAC (Assessment and Accreditation of Laboratory Animal Care)-accredited animal facility with controlled temperature (22°C), humidity, and lighting (alternating 12 h light/dark cycles). Food and water were provided ad libitum. All procedures in mice were performed according to animal protocols approved by the St. Jude Children’s Research Hospital Institutional Animal Care and Use Committee and National Institutes of Health guidelines (no. 235-100636-02/23; date of approval 02-22-2023). WT and Neu1^−/− mice aged 1, 2, and 4 months (FVB/NJ background)⁷ were included in this study. One-month-old mice (n = 11) were i.v. injected (total 100 μL/mouse) with a single combined dose of scAAV2/8-CMV-NEU1 (2 × 10¹² vg/mouse) and scAAV2/8-CMV-CTSA (1 × 10¹² vg/mouse) (scAAV2/8-N/P) and sacrificed at age 4 months.

Production and titration of the scAAV2/8-CMV-NEU1 and scAAV2/8-CMV-CTSA vectors

The production and titration of the scAAV2/8-CMV-NEU1 and scAAV2/8-CMV-CTSA constructs has been described previously.⁷⁰ The proteins are under the control of a constitutive cytomegalovirus early enhancer promotor that drives the expression of the 1,248 base pair human NEU1 and 1443 base pair human CTSA cDNAs, respectively. The scAAV vector particles were made using a good manufacturing practice (GMP) process, comparable transient-transfection procedure in the Children’s GMP, LLC facility on the St. Jude campus, as previously described for the hemophilia B vector.⁷¹

Biodistribution

Total genomic DNA was extracted from the cerebellum, cortex, brain stem, hippocampus, liver, lung, heart, kidney, spleen, ovary, diaphragm, jejunum, colon, peritoneum, and inguinal lymph nodes using the DNeasy Blood and Tissue Kit (QIAGEN, Germantown, MD), following the manufacturer’s guidelines. RT-qPCR was performed using SsoAdvanced universal SYBR green supermix (Bio-Rad, Hercules, CA), 100 or 250 ng genomic DNA, 10 μM primers (CMV promotor: 5′-CGG TGG GAG GTC TAT ATA A-3ʹ; PPCA: 5′-GCG GGC TCG GAT CAT GGT GGA AT-3′; NEU1: 5′-GTC GCT CCC CAG TCA TCT CTC CC-3′) and RNAse free water in a 25 μL reaction volume on a CFX96 opus 96 Real-Time PCR system (Bio-Rad). The copy number was calculated using the molecular weight of CMV-PPCA or CMV-NEU1 plasmids. The standard curve was generated for quantification of the viral genome. PPCA standard curve: 1.79 × 10⁸, 8.79 × 10⁷, 1.76 × 10⁷, 8.79 × 10⁶, 1.76 × 10⁶, 8.79 × 10⁵, 1.76 × 10⁵, 8.79 × 10⁴, 1.79 × 10⁴, 8.79 × 10³, 1.79 × 10³ copies. NEU1 standard curve: 1.84 × 10⁸, 9.19 × 10⁷, 1.84 × 10⁷, 9.19 × 10⁶, 1.84 × 10⁶, 9.19 × 10⁵, 1.84 × 10⁵, 9.19 × 10⁴, 1.84 × 10⁴, 9.19 × 10³, 1.84 × 10³ copies. All samples were tested in triplicate and a negative control (H₂O) was included to monitor sample contamination.

Detection of antibodies against AAV2/8 capsid and human NEU1 and PPCA

Transgene products

Antibody levels were tested in the sera of WT, Neu1^−/− mice, and treated Neu1^−/− mice via an enzyme-linked immunosorbent assay (ELISA). Serum collected at 3 months post injection was used for the ELISA. In brief, either disrupted or intact scAAV2/8-CMV-NEU1 and scAAV2/8-CMV-CTSA virus or purified NEU1 (Novus Biologicals, Centennial, CO) and PPCA proteins was diluted to 1 mg/mL in DPBS at 50 μL per well in a Nunc MaxiSorp ELISA plate (Thermo Scientific, Waltham, MA). When purified virus was used, virus concentration was estimated using NanoDrop lite Spectrophotometer (Thermo Scientific), prior to disruption. The virus stock was incubated in disruption buffer containing 5 mM Tris (pH 7.8), 60 mM KCl, and 0.05% Triton X-100 for 15 min at room temperature (RT). Virus was then diluted in DPBS to 1 mg/mL. A human-specific NEU1 or PPCA antibody was used as positive control for the ELISA against the two proteins and an anti-AAV8 antibody (Progen, Wayne, PA) was used as a positive control against the capsid. Plates were incubated at 4°C overnight to allow the protein/virus to bind. Following incubation, the plates were washed twice with DPBS/0.05% Tween 20 (washing buffer) and blocked with 10% FBS in DPBS (blocking buffer) for 30 min at RT. Plates were washed twice with washing buffer, and diluted samples in triplicate were placed in the wells for 2 h. Plates were then washed twice with washing buffer, and goat anti-mouse immunoglobulin G-horseradish peroxidase (IgG-HRP) (Southern Biotechnology, Birmingham, AL) was used as a secondary antibody diluted 1:5,000 in blocking buffer. Plates were incubated for 1 h at RT and washed three times with washing buffer, and TMB 2-Component Microwell Peroxidase Substrate (SeraCare Life Sciences, Gaithersburg, MD) was added to each well. Plates were allowed to develop for 15 min, and the reaction was stopped by adding an equal volume of 1 M phosphoric acid. Plates were then read in a Molecular Devices Spectramax M5 (Molecular Devices, San Jose, CA) at an OD of 450. Triplicates were averaged and plotted.

Immunohistochemical analyses

For IHC analysis, tissues were fixed in 10% buffered formalin and embedded in paraffin. The sections were cut (6 μm) and deparaffinized.

For NEU1, PPCA, and CD68 staining, antigen retrieval was performed by the pressure cooker method using citrate buffer (0.1 M citric acid and 0.1 M sodium citrated [pH 6.0] containing 0.05% Tween 20). After sections were blocked with 2.5% Normal Horse Serum (Vector Laboratories, Newark, CA), they were incubated overnight at RT with anti-human NEU1 (in-house), anti-human PPCA (in-house), or anti-mouse CD68 (Cell Signaling Technology, Danvers, MA) antibodies. Sections were rinsed in PBS and subsequently incubated with biotinylated secondary goat anti-rabbit antibody (Jackson ImmunoResearch Laboratory, West Grove, PA) for 1 h. Endogenous peroxidase was quenched by incubation the sections with 3% hydrogen peroxidase for 20 min. Sections were then developed using the ImmPRESS HRP Horse Anti-Rabbit IgG Polymer Detection Kit (Vector Laboratories) and counterstained with hematoxylin.

For Lamp1 staining, antigen retrieval was completed in the microwave using citrate buffer (0.1 M citric acid and 0.1 M sodium citrated [pH 6.0] containing 0.05% Tween 20). After sections were blocked with PBS containing 0.1% bovine serum albumin (BSA), 10% normal goat serum, and 0.5% Tween 20, they were incubated overnight at RT with anti-mouse Lamp1 (Cell Signaling Technology) antibody. Sections were rinsed in PBS and subsequently incubated with biotinylated secondary goat anti-rabbit antibody (Jackson ImmunoResearch Laboratory) for 1 h. Endogenous peroxidase was quenched by incubation the sections with 3% hydrogen peroxidase for 15 min. Sections were then developed using the ImmPRESS HRP Horse Anti-Rabbit IgG Polymer Detection Kit and counterstained with hematoxylin.

For APP staining, the Mouse on Mouse (M.O.M.) ImmPRESS HRP Polymer Kit (Vector Laboratories) was used. Antigen retrieval was performed by the pressure cooker method using citrate buffer (0.1 M citric acid and 0.1 M sodium citrated [pH 6.0] containing 0.05% Tween 20). Endogenous peroxidase was quenched by incubating the sections with 3% hydrogen peroxidase in methanol for 15 min. Slides were washed twice in PBS for 5 min and incubated for 1 h in the working solution of prepared M.O.M. mouse IgG blocking reagent. Sections were washed twice in PBS for 5 min and incubated for 5 min in M.O.M. normal horse serum, 2.5%. Sections were incubated overnight at RT with mouse anti-APP antibody for 30 min. The sections were washed twice in PBS and incubated for 10 min with M.O.M. ImmPRESS Reagent and washed again twice with PBS. Antibody detection was performed using the ABC Kit (Vector Laboratories) and diaminobenzidine substrate (Invitrogen, Carlsbad, CA) and counterstained with hematoxylin.

For histopathological examination, sections were cut (6 μm) and stained with a standard H&E method.

For co-IHC staining, sections were co-immunolabeled with anti-NEU1 and anti-PPCA antibodies using a Ventana Discovery Ultra autostainer (Roche, Indianapolis, IN). All reagents and kits used were obtained from Ventana Medical Systems (Roche). The following conditions were followed: heat-induced antigen retrieval was performed using Cell Conditioning Solution ULTRA CC1 for 32 min and rinsed with diluted reaction buffer (DRB). Sections were incubated with anti-NEU1 antibody (in-house) for 20 min, rinsed with DRB, and incubated with secondary antibody anti-Rb HRP DISCOVERY OmniMap for 16 min. After the slides were rinsed with DRB, the sections were incubated with a purple chromogen for 4 min using the DISCOVERY Purple kit. Endogenous peroxidase was quenched by incubating the sections with 3% hydrogen peroxidase for 32 min and rinsed with DRB. A second heat denaturation step was performed with Cell Conditioning Solution ULTRA CC2 for 8 min. The sections were incubated with anti-PPCA antibody (in-house) for 20 min, rinsed with DRB, and incubated with secondary antibody anti-Rb HRP DISCOVERY OmniMap for 16 min. The slides were rinsed with DRB and incubated with a brown chromogen for 8 min using the ChromoMAP DAB kit, followed by Hematoxylin II for 12 min and Bluing reagent for 4 min. The slides were rinsed with DRB and cover slipped with Liquid Coverslip.

Immunofluorescent analyses

All immunofluorescence stainings were performed on the Ventana Discovery Ultra autostainer (Roche). All reagents and kits used were obtained from Ventana Medical Systems (Roche). Tissue sections (6 μm) were deparaffinized and the U DISCOVERY 5-Plex IF procedure was followed. Slides were dewaxed and incubated with DISCOVERY inhibitor in cell conditioning solution (ULTRA CC2) for 64 min. The slides were washed with DRB and incubated with rabbit anti-NEU1 (in-house) for 64 min and washed again. The sections were incubated with DISCOVERY OmniMap anti-rb HRP for 16 min, washed, and incubated with a DISCOVERY Rhodamine 6G kit for visualization for 8 min. The slides were washed and incubated with rabbit anti-PPCA (in-house) for 64 min. The sections were washed and incubated with DISCOVERY OmniMap anti-rb HRP for 16 min and the protein was visualized using the DISCOVERY CY5 Kit (8 min). Discovery QD DAPI nuclear counterstain was applied, and the slides were mounted with Prolong Gold Antifade reagent (Thermo Fisher Scientific, Life Technologies, Waltham, MA).

Enzyme assays

The tissues were homogenized in 10× w/v ddH₂O and, when necessary, further diluted in Enzyme Dilution Buffer (50 mM Na-acetate [pH 5.0], 100 mM NaCl, 1 mg/mL BSA [w/v], 0.02% Na-azide [w/v]). Enzyme activities were measured with the appropriate fluorometric substrates as described previously.^18,66,72 Neu1 activity was measured against synthetic 4-methylumbelliferyl-α-D-N-acetylneuraminic acid (Sigma-Aldrich, St. Louis, MO) at pH 4.3, β-Gal activity against 4-methylumbelliferryl-β-D-galactopyranoside (Sigma-Aldrich), β-Hex activity against 4MU-N-acetyl-β-D-glucosaminide (Sigma-Aldrich), α-Man activity against 4MU-α-D-mannopyranoside (Sigma-Aldrich), and CA activity against the dipeptide Z-Phe-Ala (Sigma-Aldrich). All reactions were performed at 37°C for 1 h and stopped with carbonate stop buffer (0.5 M Na₂CO₃ with the pH set to 10.7 by adding 0.5 M NaHCO₃) for Neu1, β-Gal, β-Hex, and α-Man activities and the CA activity was stopped at 100°C for 5 min and 10 μL reaction mixture was read in 250 μL 50 mM Na-carbonate stop buffer (pH 9.5) containing 500 μL o-phtaldialdehyde (10 mg/mL) and 500 μL 2-mercaptoethanol (5 μL/mL) per 30 mL. All fluorescence was measured at λ_ex = 355/λ_em = 460 and interpolated to a standard curve adjusted for dilution, length of incubation, and normalized to protein concentration.

ELISA of Aβ-42

The amount of mouse Aβ-42 in the cerebrospinal fluid was determined by using mouse Aβ-42 enzyme-linked immunosorbent assay kit (Thermo Fisher Scientific; Invitrogen, Waltham, MA) following manufacturer’s instructions.

Masson trichrome staining

Masson trichrome staining was performed according to the manufacturer’s protocol (Polysciences., Warrington, PA). Tissues were fixed in 10% buffered formalin and embedded in paraffin were cut (6 μm), deparaffinized, and fixed in Mordant in Bouin’s solution for 1 h at 60°C. Sections were stained sequentially at RT with Weigert’s iron hematoxylin (10 min), Biebrich scarlet acid fuchsin (5 min), phosphotungstic/phosphomolybdic acid (10 min), and aniline blue (5 min). The sections were washed, dehydrated, and mounted with a xylene-based mounting medium.

Sialic acid assay

Urine was collected from individual control animals (WT and Neu1^−/−) at age 1, 2, and 4 months and from 2- and 4-month-old mice treated with scAAV-N/P. Urine samples were further diluted (1:5) in PBS when necessary. The total sialyloligosaccharide content in mouse urine was determined by the release of bound sialic acid using an Enzychrome Sialic Acid Assay Kit from BioAssay Systems (Hayward, CA). Total sialic acid was measured following acid hydrolysis for 1 h at 80°C. After the addition of neutralization buffer, samples were cooled at RT and incubated with an enzymatic master mix to utilize an enzyme-coupled reaction to oxidize any liberated sialic acid. A standard curve was generated from a 10 mM sialic acid stock solution that was serially diluted in ultrapure water. The fluorescence was read at λ_ex = 535 nm, λ_em = 595 nm and the data were interpolated to the standard curve to yield a concentration value in μM (nmol/mL).

Vacuolization area measurements

To measure the extent of vacuolization in kidney sections, 1- and 4-month-old WT and Neu1^−/− mice and 4-month-old mice treated with scAAV-N/P were stained for H&E (n = 8). A macro was developed in ImageJ to analyze six random areas per section of each mouse. A detection threshold was set for each genotype to analyze the total area occupied by vacuoles and batched to all images using the same parameters.

Quantification of Lamp1⁺ and CD68⁺ areas

Lamp1⁺ and CD68⁺ DAB staining was automatically segmented in IHC images using ilastik’s pixel classification workflow⁷³ (v.1.3.2), which uses a random forest classifier to segment pixels based on local image features. Pixels were designated as belonging to either background, cellular layers (e.g., nuclear pattern of the tissue), tissue folds, or positive DAB staining. Owing to differences in tissue and DAB⁺ morphologies, a separate ilastik classifier was developed for each tissue (choroid plexus, cerebellum, hippocampus, heart, skeletal muscle (GA), kidney, liver, and spleen) and DAB staining (Lamp1⁺ and CD68⁺). Each classifier was trained on 4–15 cropped images or "regions of interest" (ROIs) of various size pixels using ilastik’s graphical user interface. ROIs were selected to ensure that each pixel class was sufficiently well represented.

A custom MATLAB class was written to coordinate batch processing and perform image quantification for whole tissue sections (see data and code availability statement). Raw images were first converted from .czi to .tiff using the Bio-Formats MATLAB Toolbox (v.6.6.0). Next, the trained classifier was run headlessly over whole tissue sections in batch. The resultant segmented images were then filtered to exclude small objects attributable to noisy segmentation. Connected component analysis was then performed on the size-filtered segmentation and the area of each object was measured using MATLAB’s built-in region props method.

The raw data output was presented as Lamp1⁺ or CD68⁺ area per total tissue area (%) and statistical analyses was performed using GraphPad Prism 10 (GraphPad Software, San Diego, CA).

Statistical analysis

Statistical analyses were performed using GraphPad Prism 9 (GraphPad Software, San Diego, CA). Significant differences between study groups for all quantitative outcome measures were established using one-way ANOVA (n ≥ 6) or the Brown-Forsythe and Welch ANOVA (n < 6). Significance was defined as p < 0.05. The data are presented as mean ± SD or as median ± quartiles.

Data and code availability

All relevant data supporting the key findings of this study are available within the article and its supplemental information files or from the corresponding author upon reasonable request. The MATLAB class and an example script used to analyze Lamp1 and CD68 DAB positive areas can be found here: https://github.com/stjude/dab_detection.