Enhanced lysosomal exocytosis and altered growth factor signaling are associated with cartilage pathology in a model of mucopolysaccharidosis type IVA
Jen-Jie Lee, Po-Nien Lu, Lynn Dukes-Rimsky, Chelsi Jeter, Maxwell B. Colonna, Andrzej B. Poplawski, Gavin Arno, Jenna Hallman, Christina Underwood, Amrita Basu, Laura Pollard, Ryan J. Weiss, Richard Steet, Heather Flanagan-Steet
TL;DR
This study shows that lysosomal exocytosis and disrupted growth factor signaling contribute to cartilage problems in a zebrafish model of a genetic disorder.
Contribution
The study reveals a new connection between lysosomal exocytosis and altered growth factor signaling in mucopolysaccharidosis type IVA.
Findings
Loss of galns increases lysosomal exocytosis in developing cartilage of mutant zebrafish.
Increased exocytosis is linked to reduced cathepsin activity and lower TGFβ and BMP signaling.
Altered glycosaminoglycan levels are observed in intracellular and extracellular compartments.
Abstract
Optimal lysosomal function is essential for early tissue development. This is evidenced by the large number of inherited disorders, collectively called the lysosomal storage disorders (LSDs), caused by lysosomal dysfunction. Although it is clear that macromolecular accumulation adversely impacts tissue development, the breadth of downstream pathways contributing to pathology has yet to be elucidated. Multiple studies indicate that mechanisms beyond lysosomal storage also profoundly influence early tissue formation. Of these, abnormal growth factor signaling has been linked to pathology in several different LSDs. Recent work in a zebrafish model of sialidosis demonstrated that mislocalizing lysosomal cathepsins by increased exocytosis disrupts the TGFβ-related signaling pathways that control skeletal formation. Here, we show that loss of N-acetyl galactosamine-6-sulfatase (galns) also…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8- —National Institute of Neurological Disorders and Strokehttp://dx.doi.org/10.13039/100000065
- —National Institute of General Medical Scienceshttp://dx.doi.org/10.13039/100000057
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsLysosomal Storage Disorders Research · Fish biology, ecology, and behavior · Calcium signaling and nucleotide metabolism
INTRODUCTION
Lysosomes represent a heterogeneous population of organelles, the diverse functionality of which is reflected in their unique enzymatic contents and positional flexibility (Ballabio and Bonifacino, 2020). Several lysosomal functions, including membrane repair and bone remodeling, depend on the ability of the lysosome to traffic within the cell and fuse with other organelles and membranes. The importance of strictly regulating lysosomal functions is underscored by the growing number of rare disorders caused by its dysfunction. The bulk of these disorders, collectively termed the lysosomal storage disorders (LSDs), stem from loss-of-function variants in the hydrolytic enzymes or structural proteins that support macromolecule degradation and membrane fusion. Although the neurological and skeletal systems are commonly affected in most LSDs, there is remarkable diversity in the tissues impacted among these disorders. This is evident within the mucopolysaccharidoses (MPSs), as some disorders primarily affect the central nervous system (e.g. MPSIIIA) and others more substantially affect the skeletal system (e.g. MPSIVA) (Muenzer, 2011; Shapiro and Eisengart, 2021; Leal et al., 2023). The prevailing view of this phenotypic specificity has been that it is driven by accumulation of specific macromolecules in affected tissues. Much attention has focused on the impact of macromolecular storage within affected tissues, but the full breadth of downstream pathways causing pathology is not fully defined. Many recent studies support the existence of mechanisms other than storage that profoundly influence early tissue formation, including impaired mitochondrial function, oxidative stress, inflammation and abnormal growth factor signaling (Fiorenza et al., 2018).
Disrupted growth factor signaling has been linked to pathology in multiple LSDs, including Gaucher, MPSII, mucolipidosis II (MLII) and sialidosis (Costa et al., 2020; Zancan et al., 2015; Bellesso et al., 2018; Flanagan-Steet et al., 2016, 2018; van de Vlekkert et al., 2019; Lee et al., 2024). Studies of cartilage development in zebrafish models of MLII and sialidosis suggest that growth factor disruptions can precede macromolecular storage, implicating other aspects of lysosomal dysfunction as early drivers of tissue pathology (Flanagan-Steet et al., 2016; Flanagan-Steet et al., 2018; Lee et al., 2024). In sialidosis, defects in the neuraminidase NEU1 cause lysosomes to fuse with plasma membrane (Yogalingam et al., 2008). Unregulated lysosomal exocytosis causes proteolytic enzymes and undigested molecules to be expelled outside cells. In MLII, lysosomal dysfunction stems from defects in the carbohydrate-dependent, mannose-6-phosphate (M6P)-targeting pathway (Kollmann et al., 2010; Kornfeld and Mellman, 1989; Reitman and Kornfeld, 1981; Tiede et al., 2005). Failure to generate M6P residues on the N-glycans of soluble acid hydrolases impairs their trafficking to the lysosome, causing multiple proteases to be secreted outside cells. This enzyme secretion is compounded by Neu1 mislocalization, which also enhances lysosomal exocytosis (Lee et al., 2024). Studies in both models show that cathepsin proteases are active in the extracellular space, where they were can inappropriately act on multiple substrates, including the TGFβ-related growth factors that control skeletal development.
Multiple factors influence the enzymatic function of cathepsin proteases. Studies in vitro and in vivo show that the processing, stability and substrate specificity of these proteases is regulated by the type of glycosaminoglycans (GAGs) present in the extracellular space (Li et al., 2002, 2000; Wilson et al., 2009; Wilson and Bromme, 2010; Zhang et al., 2024). In particular, chondroitin sulfate proteoglycans are not only necessary to process pro cathepsin K; they also stimulate its activity. In contrast, heparan sulfate proteoglycans have been shown to inhibit several cathepsin activities. This is particularly important in the context of LSDs, as excessive GAG accumulation and altered growth factor signaling can synergize to alter the landscape of intracellular and extracellular GAGs present in different tissues. The ability of GAGs to differentially regulate cathepsin function may contribute to variability in bone and cartilage phenotypes among the LSDs. This is supported by studies in MLII zebrafish, in which increased extracellular chondroitin-4-sulfate was shown to stimulate Cathepsin K (Ctsk) activity (Flanagan-Steet et al., 2018). In this case, extracellular Ctsk promoted TGFβ and inhibited BMP signaling, maintaining chondrocytes in an immature developmental state (Flanagan-Steet et al., 2016; Petrey et al., 2012). In contrast, studies in MPSI suggest that excessive buildup of heparan sulfate and dermatan sulfate likely inhibits CTSK, impairing osteoclast activity and decreasing cartilage resorption (Wilson et al., 2009; Wilson and Bromme, 2010).
Like other MPS disorders, skeletal disease is prominent in MPSIVA. Patients exhibit skeletal dysplasia and early cartilage deterioration (Hendriksz et al., 2013; Sawamoto et al., 2020). Studies in mouse and rat models show that these phenotypes are associated with an increase in hypertrophic chondrocytes and reduced ossification (Leal et al., 2023; Bertolin et al., 2021). Given the phenotypic similarities between MLII, sialidosis and MPSIVA, we asked whether enhanced lysosomal exocytosis and altered cathepsin activity contributes to skeletal pathology in zebrafish with N-acetyl galactosamine-6-sulfatase (galns) deficiency. We show that zebrafish bearing a loss-of-function mutation in galns exhibit dysmorphic cartilages with abnormal chondrocytes and reduced bone mineralization. Using a combination of biochemical and microscopic approaches, we demonstrate that lysosomal exocytosis and cathepsin secretion is increased in galns-deficient chondrocytes. Abnormal cathepsin activity was associated with reduced TGFβ and BMP signaling and differences in the level of heparan sulfate and chondroitin sulfate GAGs present in mutant chondrocytes. These findings implicate a broader and more complex role for enhanced lysosomal exocytosis in skeletal pathology in the LSDs, suggesting that the repertoire of intracellular and extracellular GAGs can profoundly influence the phenotypic outcome of skeletal disease.
RESULTS
Mutation of an essential splice site generates a loss-of-function allele in zebrafish galns
The Sanger Center's Targeting Induced Local Lesions in Genomes (TILLING) screen identified a c.539+1G>A (sa31602) mutation predicted to disrupt an essential splice site in exon 5 of zebrafish galns (Fig. 1A). F1 embryos carrying the c.539+1G>A mutation were generated by in vitro fertilization. Using next-generation sequencing, multiple F1 adults bearing the galns variant were identified and outcrossed with wild-type TLAB for six generations. Reverse transcription PCR (RT-PCR) analysis of 8 days post-fertilization (dpf) larvae identified galns transcripts of similar size in wild-type (+/+) and homozygous mutant (m/m) larvae (Fig. 1B).
*Mutation of an essential splice site generates a null allele of galns. (A) Schematic of galns gene structure with the exon 5–intron 5 boundary and G/A mutation highlighted, denoted by an arrow. Arrowheads indicate the position of the oligonucleotide primers used for RT-PCR and sequence analysis. (B) RT-PCR analysis of galns transcript in 8 days post-fertilization (dpf) wild-type (+/+) and mutant (m/m) larvae. (C) Long-read sequence analysis performed using Oxford Nanopore Technologies identified 5 bp on intron 5 (I5) inserted into the transcript in galns mutant larvae. The predicted impact of this insertion on the amino acid sequence (as shown) includes a shift in frame and introduction of an early stop codon. Sequencing was performed on mRNA-isolated samples containing 30 larvae. (D) Galns enzyme assays performed on larvae 6 dpf detected little to no activity in galns mutants (m/m). n=3 biological replicates of samples containing 25 larvae per genotype. Error bars=s.e.m. Two-tailed paired Student's t-test, ***P<0.001, ****P<0.0001. (E) A high-resolution melt curve assay was developed to genotype animals for experiments and line propagation. (F) The progeny from multiple pairwise heterozygous crosses were analyzed daily from 0 to 18 dpf. All progeny were genotyped as they died. The number of progeny per genotype that died at each timepoint is indicated on the graph, with the majority of galns mutants (m/m) dying between 9 and 14 dpf. (G) All progeny of three independent pairwise heterozygous crosses were genotyped at 30 and 50 dpf. The numbers of animals per genotype were identified at each timepoint. n=3 biological replicates with a total of 25-50 animals per experiment. Error bars=s.e.m. Dunnett's multiple comparisons test, *P<0.05, **P<0.01. (H) Zebrabox automated behavioral tracking identified a progressive motility defect in galnsm/m mutant larvae. Representative images of traced swim paths of larvae 6 and 8 dpf. Green traces represent periods of slow-speed swimming; red traces represent fast-speed swimming. (I) Graphs comparing total distance swam and the number of swim events initiated 4-10 dpf in wild-type (+/+) and mutant (m/m) larvae. Each dot represents one animal. n=30 larvae from three biological matings. Error bars=s.e.m. *P<0.05, **P<0.01, ***P<0.001, ***P<0.0001.
Using Oxford Nanopore Technologies long-read sequencing to analyze RT-PCR amplicons, we showed that the first 5 bp of intron 5 are inserted between the exon 5 and 6 coding sequences in 100% of mutant transcripts (Fig. 1C). Together with enzyme activity assays, which detected very little Galns activity in mutant larvae (Fig. 1D), the data suggest that the c.539+1G>A (sa31602) mutation generates a null allele. Notably, multiple pathogenic or likely pathogenic spice site variants are reported in ClinVar, including one affecting the same 5′ donor site as found in the galns mutant zebrafish (see Fig. S1A). To facilitate ease of sample generation for all subsequent experiments, we developed a high-resolution melt curve assay that was used to genotype adults and larvae throughout the remainder of the study (Fig. 1E) (Klaver et al., 2021).
galnsm/m larvae exhibit increased lethality and abnormal swim behaviors
By 8 dpf, galns^m/m^ larvae were slightly smaller than wild-type clutch mates and had increased pigmentation and visibly smaller eyes (Fig. S1B,C). We observed that 83% of mutant larvae analyzed also exhibited a reduction in the size of the optic lens opening (Fig. S1D). By10 dpf, increased lethality was evident, with the majority of mutant larvae dying by 18 dpf (Fig. 1F). Genotypic analyses of progeny from multiple heterozygous matings indicated that a small number of galns^m/m^ larvae survive up to 30 dpf, but all mutants die by 50 dpf (Fig. 1G). Using an automated behavioral tracking system, we characterized larval swimming and found that increased larval lethality corresponded with defects in elicited and spontaneous swim behaviors (Fig. 1H). Traced paths of individual swim events revealed differences in the number of swim events initiated by mutant larvae and in the overall distance swam (Fig. 1I). Reduced motility was evident as early as 4 dpf, with significant differences in each of these parameters apparent by 6 dpf. Analyses of the traced paths and all measured parameters indicated that mutant larvae were deficient in all behaviors, including movement at slow and fast speeds.
Craniofacial cartilage development is disrupted in galnsm/m mutant larvae
Consistent with skeletal anomalies characteristic of patients with MPSIVA (Muenzer, 2011; Leal et al., 2022), in zebrafish, galns deficiency is associated with abnormal craniofacial cartilage and vertebral skeletal development. Alcian Blue staining of larvae at 7 dpf showed that galns^m/m^ mutants had shorter craniofacial cartilages than those of wild-type larvae that did not extend as far past the eyes (Fig. 2A). Measurement of the Meckel's and ceratohyal cartilages indicated that both structures were 15-20% shorter in the galns^m/m^ mutants than those in the wild-type larvae. Although the overall area of the face was smaller, no differences were noted in the articulating angle between the ceratohyal structures. This suggested that the mutant jaw was smaller than the wild-type jaw but not overtly dysmorphic (Fig. 2B). Confocal analysis of wild-type and mutant larvae expressing fli1a:EGFP, which labels neural crest-derived craniofacial chondrocytes (Lawson and Weinstein, 2002), showed that cellular morphology and organization were also altered (Fig. 2C). By 7 dpf, the majority of wild-type cartilages contained chondrocytes that were elongated and present in a single row. Consistent with their reduced length, galns^m/m^ mutant cartilages contained several regions with immature round cells that were present in multiple layers (Fig. 2C,D). Alterations in cell shape were quantified by calculating the ratio of the short and long axes, which became smaller as chondrocytes elongated (Fig. 2D) (Petrey et al., 2012). As is typical in LSDs, the majority of galns^m/m^ mutant chondrocytes were also highly vacuolated (Fig. 2E, yellow arrows). This was true across all areas of Meckel's cartilage, including within cells comprising the central and lateral portions of the structure (Fig. 2F). Similar defects in chondrocyte shape and organization were noted in a prior study involving morpholino-based reduction of galns expression (Lee et al., 2024). These abnormalities also mirror those previously described in an MPSIVA rat model expressing the common human R386C variant (Bertolin et al., 2021).
*galnsm/m mutant larvae have craniofacial cartilage and bone development. (A) Alcian Blue staining of galns wild-type (+/+) and mutant (m/m) larvae showing that mutant cartilages are shorter and slightly dysmorphic. Lines demarcating the edge of the eyes (black line), the tip of the nose (black dashed lines) and the tip of Meckel's cartilage show that wild-type cartilage (red dashed lines) extends much further past the eyes than the mutant cartilage. Red arrows highlight cartilages affected by Galns defiency. Scale bar: 20 µm. (B) Schematics highlighting the parameters measured, including the lengths of the Meckel's (M; blue) and ceratohyal (CH; green) cartilages, and angle between the ceratohyal structures (red, palatoquadrate). The distance Meckel's cartilage protrudes past the eyes and area of the head was also measured. With the exception of the ceratohyal angle, all measurements were reduced in mutant larvae. n=14-18 larvae per genotype per timepoint from three independent biological crosses analyzed. Error bars=s.e.m. Two-tailed paired Student's t-test, *P<0.05, **P<0.01, ****P<0.0001. (C) Confocal analyses of 7 dpf fli1a:EGFP-positive cells from the lateral and central portions of Meckel's cartilage identified alterations in the shape and organization of mutant chondrocytes. Scale bar: 10 µm. (D) Graphs show quantitative measurements of cell shape and organization in the central and lateral portions of Meckel's cartilage. n=20-25 larvae from three biological matings. Error bars=s.e.m. Two-tailed paired Student's t-test, ****P<0.0001. (E) Confocal images showing that mutant chondrocytes are highly vacuolated. Yellow arrows highlight vacuoles. Scale bar: 10 µm. (F) Analysis of vacuolation shows that nearly 100% of mutant chondrocytes contain vacuoles, whereas very few wild-type cells are vacuolated. n=23 larvae per genotype from three independent biological crosses were analyzed. Error bars=s.e.m. Two-tailed paired Student's t-test, ****P<0.0001. (G) Alizarin Red staining of larvae at 7 and 10 dpf identified multiple regions in galnsm/m mutants with decrease mineralization. The boxed area surrounding Meckel's cartilage represents region of interest (ROI)1, and the boxed area surrounding the ceratohyal represents ROI2. Scale bar: 20 µm. (H) The fluorescence intensity of Alizarin Red staining was measured in each of the ROIs. Data from n=12 larvae per genotype from three independent biological crosses were analyzed and graphed. Data from 7 dpf (left column) and 10 dpf (right column) are presented. Error bars=s.e.m. Two-tailed paired Student's t-test, *P<0.05, **P<0.01, ***P<0.001. (I) Alizarin Red staining of larval vertebrae also shows reduced mineralization in mutant vertebrae. Asterisks highlight regions of difference at 7 dpf. Bars highlight the reduction in the diameter of mineralized structures in 10 dpf mutants. (J-L) Graphs show quantitative measurements of the number of mineralized vertebrae in 7 dpf (left) and 10 dpf (right) larvae (J), the intensity of Alizarin Red (AR) fluorescence in 7 dpf larvae (K) and the diameter of Alizarin Red-positive vertebrae in 10 dpf larvae (L). n=21 larvae from three biological replicates. Error bars=s.e.m. Two-tailed paired Student's t-test, *P<0.05, ***P<0.001, ***P<0.01.
Altered calcification of long bone growth plates was also noted in MPSIVA GALNS-deficient rats (Bertolin et al., 2021). Like mammalian long bones, the zebrafish craniofacial bones largely develop by a process of endochondral ossification. To ask whether cartilage maturation and mineralization is altered in the zebrafish galns^m/m^ mutants, 7 and 10 dpf larvae were stained with the calcium-binding dye Alizarin Red, and the degree of calcification was quantitated (Fig. 2G,H). Reduced calcification was noted in multiple facial regions, including in the bony veils surrounding the Meckel's and ceratohyal cartilages and in the cleithrum that supports the pectoral fin. Disruptions in the character and shape of calcified area were also noted. This was particularly clear in the wild-type ceratohyal cartilage, in which a thin sheet of Alizarin Red staining was detected between the two collars [see Fig. 2G, region of interest (ROI)2]. In galns^m/m^ mutant larvae, staining was primarily noted along the edges, with little calcification evident between the collars. Parallel analysis of the developing spinal column showed that the number of Alizarin Red-positive vertebrae was also reduced in 7 dpf mutants compared to that in wild-type larvae (Fig. 2I,J). Although the number of calcified vertebrae increased as the mutants aged (Fig. 2I-L), the mineralized area remained significantly lower than that in wild-type larvae. In light of these data, it is currently unclear whether impaired swim behavior stems from defects in the vertebral skeleton that impede larval movement or whether galns^m/m^ mutants also have previously unrecognized neurological deficiencies.
Loss of galns impacts the activity of other enzymes in the lysosomal multi-enzyme complex
Several studies suggest that the Galns enzyme can function as part of the lysosomal multi-enzyme complex (LMC) (Pshezhetsky and Potier, 1996; Pshezhetsky and Ashmarina, 2001). The core LMC includes the neuraminidase Neu1, the Neu1 chaperone protective protein cathepsin A (PPCA) and β-galactosidase (Fig. 3A). To ask whether any of the phenotypes noted in galns^m/m^ mutant larvae correlate with alterations in other enzymes in the LMC, we analyzed each of their activities in larvae at 6, 8 and 10 dpf (Fig. 3B-E). As noted in our initial analyses (see Fig. 1D), Galns activity was largely undetected in mutant larvae at each of these timepoints (Fig. 3B). Although β-galactosidase levels were indistinguishable from those in wild-type larvae at 6-8 dpf, they were reduced at 10 dpf in mutants (Fig. 3C). In contrast, reduced neuraminidase activity was apparent at 6, 8 and 10 dpf (Fig. 3D), but no differences in the neuraminidase chaperone PPCA were noted (Fig. 3E). These data suggested that Neu1 likely reaches the lysosome, where its activity may be compromised owing to other aspects of lysosomal dysfunction. Notably, in our prior analyses of galns morphants, β-galactosidase and neuraminidase levels were unaffected. These differences likely stem from the fact that, unlike galns^m/m^ mutants, the morphants maintain 40% residual activity (Lee et al., 2024). To ask whether macromolecular storage and compromised lysosomal function contribute to reduced Neu1 activity in mutants, we adapted a non-reducing end (NRE) liquid chromatography–mass spectrometry-based method to examine whether undigested GAGs accumulate in mutant lysosomes (Fig. 3F; Fig. S2A,B) (Klionsky et al., 2021; Basu et al., 2025). NRE analysis of 8 dpf larvae identified a 2.5-fold increase in disease-specific N-acetyl-galactosamine-6-sulfate (GalNAc-6S) NRE species, indicating the presence of undigested chondroitin/dermatan sulfate in galns^m/m^ mutants (Fig. 3F). Importantly, these NRE species have also been detected in samples from patients with MPSIVA (Lawrence et al., 2020). Although NRE analyses do not show where these GAGs may be accumulating, the data support the possibility that macromolecule storage contributes to early skeletal pathology in MPSIVA. This is different than previously noted in zebrafish models of MLII and sialidosis, in which phenotypes preceded detectable levels of lysosomal storage (Flanagan-Steet et al., 2009, 2016; Lee et al., 2024).
*The activities of other enzymes in the lysosomal multi-enzyme complex are altered in galnsm/m mutant larvae. (A) Schematic illustrating the composition of the lysosomal multi-enzyme complex (LMC). βGAL, β-galactosidase; GALNS, N-acetylgalactosamine-6-sulfatase; NEU1, neuraminidase; PPCA, protective protein cathepsin A. Created in BioRender by Flanagan-Steet, H. (2026). https://BioRender.com/mijhv3d. This figure was sublicensed under CC-BY 4.0 terms. The illustration was adapted from a similar cartoon presented in Lee et al. (2024) under the terms of the CC-BY-NC license. (B-E) The enzyme activities of Galns (B), Neu1 (C), PPCA (D) and β-galactosidase (E) were measured in whole galns+/+ wild-type and galnsm/m mutant larvae at 6, 8 and 10 dpf. (F) Non-reducing end (NRE) (Klionsky et al., 2021) analysis of whole wild-type and mutant larval samples via HILIC-Q-TOF-MS revealed significant accumulation of N-acetylgalactosamine-6-sulfate (GalNAc-6S) from chondroitin sulfate/dermatan sulfate glycosaminoglycans (GAGs). (G) Neuraminidase (sialidase) activity was assayed in brain isolated from 10 dpf larvae. All analyses were performed on three biological replicates of samples containing 35 larvae per genotype per timepoint. Error bars=s.e.m. Two-tailed paired Student's t-test, *P<0.05, **P<0.01, ***P<0.001, ***P<0.0001.
A recent study by Xu et al. (2025) also identified a reduction in neuraminidase activity in the brains of MPSIVA mice (Xu et al., 2025). To ask whether reduced levels of neuraminidase correlated with the motility deficits, we isolated the brains of 10 dpf larvae. However, unlike noted in mice, sialidase activity levels were not appreciably different in the brains of wild-type and galns-deficient zebrafish (Fig. 3G). It is unclear whether the activity of other acidic sialidases (i.e. Neu3 or Neu4) may be masking differences in the zebrafish brain or whether the reduced neuraminidase activity noted in mice develops as the animals age. Zebrafish do express several other acidic neuraminidases, including neu3.1-neu13.5 and neu4. No alterations in transcript expression were noted for any of the neuraminidase enzymes. Although only Neu4 has been localized to the lysosome, we cannot rule out the possibility that it compensates for Neu1 in zebrafish (Seyrantepe et al., 2008, 2004; Miyagi and Yamamoto, 2022; Paolini et al., 2017; Cirillo et al., 2016).
Lysosomal exocytosis is increased in the cartilage of galnsm/m mutant larvae
Studies in Neu1 knockout mice showed that loss of NEU1 activity causes the lysosomal integral membrane protein LAMP1 to become hypersialylated (Yogalingam et al., 2008). Hypersialylation increases the half-life of LAMP1, promoting exocytosis of lysosomes positioned proximal to the membrane. Prior studies in zebrafish models of MLII (gnptab-deficient) and sialidosis (neu1-deficient), two other LSDs also characterized by skeletal dysplasia, showed that Neu1-mediated increases in lysosomal exocytosis contribute to skeletal pathology (Lee et al., 2024). In light of this and the noted reduction in neuraminidase activity, we asked whether Lamp1 abundance or lysosomal exocytosis are also increased in galns^m/m^ mutant larvae. To do this, we utilized a transgenic line that expresses a Lamp1-mCherry fusion protein under the control of a heat shock promoter [Tg(hsp701:Lamp1-mCherry)] (Fig. 4A) (Ellis et al., 2013). Wild-type and mutant transgenic zebrafish were heat shocked at 4 dpf and harvested from 0.5 to 24 h post-heat shock (phs). Western blot analyses of Lamp1-mCherry showed that although the same amount of protein was initially produced, by 3 h phs, Lamp1 abundance had increased in galns^m/m^ mutant zebrafish relative to that in wild-type zebrafish (Fig. 4B,C; Fig. S3A,B). Levels of Lamp1-mCherry subsequently declined until 24 h phs, with mutants ultimately containing ∼30% less Lamp1-mCherry than wild-type zebrafish (Fig. 4D,E and Fig. S3A,B). To ask whether these fluctuations in Lamp1-mCherry abundance correlated with increased lysosomal exocytosis, we microscopically assessed Lamp1 location in the chondrocytes of fli1a:EGFP; hsp70:Lamp1-mCherry double-positive animals (Fig. 4F). Live confocal analyses showed that wild-type cartilages contained lower-intensity mCherry-positive puncta that were centrally located. In mutant chondrocytes, these intracellular puncta were larger and more intensely stained and, in some cases, appeared swollen (see ROI2, blue arrows, Fig. 4F). Although low-level membrane-localized mCherry signal was evident in wild-type cells (see ROI1, yellow arrows, Fig. 4F), the intensity of membrane-localized Lamp1 was much lower than that noted in mutant cells (Fig. 4G). Quantitative analyses showed that a similar number of wild-type and mutant cells exhibited moderate-intensity surface-localized Lamp1-mCherry. However, twice as many mutant chondrocytes exhibited high-intensity surface-localized mCherry, supporting the likelihood that lysosomal exocytosis was enhanced in a subset of galns^m/m^ mutant chondrocytes.
*Increased lysosomal exocytosis alters cathepsin activity in galnsm/m mutant larvae. (A) Schematic illustrating the hsp701:Lamp1-mCherry transgene and experimental workflow for analysis of Lamp1 abundance and subcellular location in developing larvae. (B,C) Western blots probed for mCherry in the whole animal indicate that the level of Lamp1-mCherry is increased 3 h post-heat shock (phs) but steadily declines until 24 h phs. (D,E) Lamp1-mCherry abundance at 0.5-10 h phs (D) and 0.5-24 h phs (E). The level present at each timepoint was normalized to the amount of Lamp1 present at 0.5 h phs. n=3 biological replicates of samples containing ten embryos per genotype per timepoint. Error bars=s.e.m. Two-tailed paired Student's t-test, **P<0.01, ****P<0.0001. (F) Confocal images of live 4 dpf galns+/+ wild-type and galnsm/m mutant larvae expressing the fli1a:EGFP (green) and hsp701:Lamp1-mCherry transgenes (red). Analysis of Lamp1-mCherry 8 h phs identified several differences in Lamp1-mCherry abundance, character and location. Yellow arrows highlight cell surface signal in red-only panels in an ROI from example 1. Blue arrows highlight large swollen Lamp1-mCherry-positive lysosomes in an ROI from example 2. Scale bar: 10 µm. (G) Schematic illustrating scored parameters, including the number of cells exhibiting high versus moderate to no cell surface Lamp1-mCherry and measurement of fluorescent intensity within ROIs throughout cartilage (Klionsky et al., 2021). Graphs show that the abundance of cell surface-localized Lamp1-mCherry and overall Lamp1-mCherry signal are both increased in mutant larvae. n=15 larvae scored per genotype. Error bars=s.e.m. Two-tailed paired Student's t-test, *P<0.05, **P<0.01. (H) Schematic showing the general mechanism of labeling active cysteine cathepsin with the BMV109 activity-based probe. Created in BioRender by Flanagan-Steet, H. (2026). https://BioRender.com/ooh869k. This figure was sublicensed under CC-BY 4.0 terms. Confocal analysis of sections of larval cartilage stained with wheat germ agglutinin (WGA; blue) and the BMV109 activity-based probed (red) shows that cathepsin activity is only present within chondrocytes of 4 dpf galns+/+ wild-type larvae. However, in galnsm/m mutants, activity is noted at the cell surface (indicative of exocytosis, white arrows) and outside the cell (see panel 1, in which the surface is denoted with a dashed line and extracellular activity with yellow arrows). Scale bar: 10 µm. (I) Schematic illustrating the procedure for labeling cathepsin activity in live larvae. ABP, activity-based probe. In-gel analyses of cathepsin activity in larval lysates show multiple alterations in Ctsk and Ctsl activity at 2-6 dpf. (J-L) Graphs show quantitation of individual cathepsins at 2-3 dpf (J), 3-4 dpf (K) and 5-6 dpf (L). n=3 biological replicates of samples containing ten embryos per genotype per labeling timepoint. Error bars=s.e.m. Two-tailed paired Student's t-test, *P<0.05, **P<0.01. (M,N) Quantitative RT-PCR (qRT-PCR) analysis showed increased expression of ctsk, ctsla and ctslb transcripts at 4 and 8 dpf in mutant larvae. n=5 biological replicates of samples containing ten embryos per genotype per labeling timepoint. Error bars=s.e.m. Two-tailed paired Student's t-test, *P<0.05, ***P<0.001. (O) Western blot of Ctsk showing that its inactive pro form is increased in mutant larvae, correlating with the noted reduction in Ctsk activity. Three different examples are shown. (P) Graphs show densitometry analysis of pro Ctsk. n=3 biological replicates of samples containing ten embryos per genotype per labeling timepoint. Error bars=s.e.m. Two-tailed paired Student's t-test, *P<0.01.
Lysosomal exocytosis is often analyzed by measuring the levels of extracellular lysosomal enzymes, but isolating extracellular pools of enzyme from larvae is not experimentally feasible. Therefore, as an additional measure of exocytosis, wild-type and galns^m/m^ mutant cartilage sections were stained with the BMV019 activity-based probe (ABP) (Fig. 4H). BMV109 contains a Cy5 moiety, the fluorescence of which is quenched until the probe covalently binds an active cysteine cathepsin (Oresic Bender et al., 2015; Verdoes et al., 2012). This makes the ABP a powerful means to stably monitor the level and location of lysosomal cathepsin activity. Local cathepsin activity was assessed by staining fli1a:EGFP-positive cartilage sections with the BMV109 ABP and wheat germ agglutinin (WGA), a lectin that binds terminal N-acetyl-glucosamine (GlcNAc) residues on protein-bound oligosaccharides. Because these glycoproteins are enriched at the cell surface and in the extracellular matrix, WGA was used to define regions of intracellular versus extracellular cathepsin activity. Confocal analyses of BMV109-stained fli1a:EGFP-positive cartilage sections identified multiple regions in mutant chondrocytes with surface-localized BMV109 reactivity (Fig. 4H, white arrows). Notably, BMV109-labeled cathepsin activity was also evident outside mutant chondrocytes (Fig. 4H, yellow arrows). In wild-type chondrocytes, cathepsin activity was exclusively found within the cell. Together with increased levels of surface localized Lamp1, these data suggest that loss of galns increases exocytosis in developing cartilage.
Increased exocytosis is associated with alterations in the activity of cathepsin proteases
Previous studies in zebrafish models of MLII and sialidosis showed that mislocalizing cathepsin proteases outside the cell through a carbohydrate-dependent mechanism increases their activity (Flanagan-Steet et al., 2018; Lee et al., 2024). In MLII, cathepsin mislocalization primarily stems from loss of the M6P modification that targets these enzymes to the lysosome (Flanagan-Steet et al., 2018; Petrey et al., 2012), with enhanced exocytosis exacerbating their extracellular release (Lee et al., 2024). In sialidosis, cathepsin mislocalization occurs because LAMP hypersialylation increases lysosomal exocytosis (Yogalingam et al., 2008). To ask whether increased exocytosis is also associated with altered cathepsin activity in galns^m/m^ mutants, the BMV109 ABP was injected into the hearts of wild-type and galns^m/m^ mutant animals between 2 and 5 dpf, and cathepsin activity was analyzed after 16 h of labeling (Fig. 4I). Using previously validated molecular mass identities, the activities of multiple cysteine cathepsins were fluorescently analyzed within SDS-PAGE gels (Flanagan-Steet et al., 2018). In-gel analyses of larval lysates identified multiple changes in cathepsin activity, with increased Ctsk activity initially noted between 2 and 3 dpf (Fig. 4I,J). However, by 5 dpf, the activities of both Ctsk and Ctsl were reduced in the galns^m/m^ mutants compared to those in wild-type larvae (Fig. 4I-L). To determine whether the initial increase in cathepsin activity was associated with lysosomal proliferation, we assayed the transcript expression of multiple cathepsin proteases. Quantitative RT-PCR (qRT-PCR) analysis showed that ctsb, ctsk, ctsl and ctss transcripts were increased in mutant larvae, compared to those in wild-type larvae, at 4 and 8 dpf, with no differences noted in ctsd (Fig. 4M,N; Fig. S3D,E). However, because the activities of Ctsk and Ctsl both declined at 5 dpf (despite increased transcript expression), we suspected that additional post-translational mechanisms impact their proteolytic activity in galns mutants. Consistent with this possibility, western blot analyses of 4 dpf larvae showed that the inactive pro form of Ctsk was increased in galns^m/m^ mutants compared to that in wild-type larvae (Fig. 4O,P). This was accompanied by decreased abundance of its intermediate forms (Fig. S3F).
Global content of a subset of GAGs is altered in galnsm/m mutants
Studies in multiple systems showed that Ctsk processing and activity depend on interaction with chondroitin sulfate GAGs (Flanagan-Steet et al., 2018; Li et al., 2002, 2000). In fact, it was shown in the cartilage of MLII zebrafish that altering the level of certain chondroitin sulfate GAGs not only directly influences Ctsk stability and activity, but interaction with chondroitin sulfate is essential for its pro protein cleavage (Flanagan-Steet et al., 2018). Using mass spectrometry, we examined the level of chondroitin sulfate, heparin sulfate, dermatan sulfate and keratan sulfate in wild-type and galns^m/m^ mutant larvae at 6 and 10 dpf. Analyses of total GAG content identified small (albeit insignificant) decreases in the abundance of both chondroitin sulfate and dermatan sulfate at 6 dpf, with no differences detected in either heparan sulfate or keratan sulfate (Fig. 5A,B). By 10 dpf, the global levels of chondroitin sulfate and dermatan sulfate appeared to normalize with decreases in heparan sulfate emerging. Because these analyses reflect the total content of GAG present in the entire animal, we asked whether either chondroitin sulfate or heparan sulfate GAGs are more substantially altered locally within galns^m/m^ mutant cartilages. For this, fli1a:EGFP-positive larval sections were immunohistochemically stained with pan-reactive antibodies recognizing either chondroitin sulfate or heparan sulfate (Fig. 5C,D) (Flanagan-Steet et al., 2016, 2018), two major classes of GAGs expressed in cartilage at these timepoints. Confocal analysis of immunohistochemically stained cartilage confirmed that heparan sulfate was locally reduced in galns^m/m^ mutants compared to that in wild-type larvae at 4 dpf. This was most evident intracellularly, with no overlap between WGA and heparan sulfate noted, suggesting that heparan sulfate biosynthesis is compromised in mutant chondrocytes. Chondroitin sulfate levels were also visibly altered in mutant larvae. Not only was substantially less staining noted on the surface of mutant chondrocytes, but there was also an increase in the number of intracellular chondroitin-positive puncta. This phenotype, which was particularly prominent in the ceratohyal cartilage, likely represents lysosomal remodeling of extracellular matrix. Unlike wild-type chondrocytes, which contained smaller centrally located puncta, in mutant chondrocytes, the chondroitin-positive puncta were large and positioned proximally to the membrane. Importantly, these findings were also consistent with the detected increase in GalNAc-6S NRE. Because Ctsk activity is inhibited by heparan sulfate but stimulated by chondroitin sulfate, these findings are consistent with the short-term increase noted in cathepsin activity followed by its progressive inactivity.
The content of multiple GAGs is altered globally and locally reduced in chondrocytes of galnsm/m mutant larvae. (A,B) Mass spectrometry-based analyses of global GAG content in whole wild-type and galnsm/m mutant larvae at 6 dpf (A) and 10 dpf (B) show slight, albeit insignificant, differences in chondroitin sulfate (CS) and dermatan sulfate (DS) at 6 dpf that are, by 10 dpf, matched by changes in heparan sulfate (HS) and keratan sulfate (KS). (C) Confocal images of immunohistochemically analyzed cartilage sections from fli1a:EGFP-positive larvae show that heparan sulfate levels (red) are locally decreased in mutant larvae as early as 4 dpf. WGA, blue. Scale bar: 10 µm. (D) Similar analyses of chondroitin sulfate (red) also showed local alterations in mutant cartilages, including lower levels of cell surface chondroitin sulfate that are matched by increased intracellular staining. This was evident in both the ceratohyal and Meckel's cartilages. Arrowheads show regions of cell surface CS staining in wild-type larvae, but intracellular CS in mutant larvae. Scale bar: 10 µm.
TGFβ-related growth factor signaling is altered in galnsm/m mutant cartilage
Prior studies identified increased exocytosis and cathepsin-mediated alterations in the TGFβ-related pathway as drivers of cartilage pathology in zebrafish models of MLII and sialidosis (Flanagan-Steet et al., 2016, 2018; Lee et al., 2024; Petrey et al., 2012). The alterations in galns^m/m^ mutant cartilage and bone are also consistent with disruptions in these pathways. To address this possibility, we leveraged two transgenic lines (Tg:BRE:dmKO and Tg:SBE:nmCherry) that express fluorophores in response to either BMP or TGFβ signaling, respectively (Collery and Link, 2011). Comparison of the destabilized monomeric Kusabira Orange (dmKO) fluorophore in fli1a:EGFP-labeled chondrocytes of Meckel's cartilage suggested that BMP signaling is reduced in galns^m/m^ mutant cartilage compared to that in wild-type cartilage (Fig. 6A). This was evidenced by fewer cells with dmKO signal, as well as a 50% reduction in overall fluorescent intensity (Fig. 6B). Lower levels of dmKO fluorescence were also noted in the perichondrial cells that encase Meckel's cartilage and in the nasal placodes anterior to Meckel's cartilage (Fig. 6A, yellow and blue arrows). Notably the reduction in perichondrial BMP signaling was consistent with decreased ossification, which was noted in these same regions (see Fig. 2G). Similar analyses using the SBE:nmCherry reporter also identified a reduction in TGFβ signaling (Fig. 6C-E). This was particularly evident in the central portion of the tip of Meckel's cartilage and also in the ceratohyal cartilage. Because the SBE reporter expresses a nuclear-localized version of mCherry, signal was scored as the percentage of nuclei with positive staining. Like noted with the BMP reporter line, 25-30% fewer galns^m/m^ mutant chondrocytes than wild-type chondrocytes exhibited nuclear mCherry, which also corresponded to a significant reduction in fluorescence intensity (Fig. 6D,E). qRT-PCR analyses of larvae at 4 and 8 dpf indicated that the reduction in signaling was also associated with a reduction in the transcript expression of TGFβ-regulated col2a1a and BMP-regulated col10a1a (Fig. 6F,G). These findings provide important insight into MPSIVA skeletal pathology, as expression of each of these matrix proteins is essential during different aspects of cartilage and bone maturation. Type II collagen is typically expressed throughout early chondrogenesis, with type X collagen expression corresponding to later-stage aspects of chondrocyte hypertrophy and bone mineralization (Goldring et al., 2006). The finding that chondrocyte shape, organization and mineralization were all altered suggests that multiple processes related to cartilage and bone development are disrupted in galns^m/m^ mutant larvae.
*TGFβ-related growth factor signaling is reduced in chondrocytes of galnsm/m mutant larvae. (A) Live confocal images of Meckel's cartilage fli1a:EGFP;BRE:dmKO-positive galns+/+ (wild-type) and galnsm/m (mutant) larvae show reduced levels of BMP signaling in mutant chondrocytes and in perichondrial cells (yellow arrows). Right column shows that signaling is also decreased in mutant nasal placodes. M, Meckel's cartilage. Scale bars: 10 µm. (B) Graphs show that the percentage of cells with positive dmKO signal and fluorescence intensity are reduced in mutant cells. n=19 larvae from three independent matings. Error bars=s.e.m. Two-tailed paired Student's t-test, **P<0.01. (C) Live confocal images of the Meckel's and ceratohyal cartilages in fli1a:EGFP; SBE:nmCherry (nuclear)-positive galns+/+ (wild-type) and galnsm/m (mutants) larvae show that TGFβ signaling is also reduced in mutant cells. Yellow arrows highlight regions with mCherry signal, in chondrocytes and perichondrium. In right panels, blue arrows highlight reduced signal in perichondrium. Scale bar: 10 µm. (D,E) Graphs show that the percentage of cells with positive nuclear mCherry signal and fluorescence intensity are reduced in the Meckel's (D) and ceratohyal (E) cartilages in mutant larvae. n=13 larvae from three independent matings. Error bars=s.e.m. Two-tailed paired Student's t-test. *P<0.05, **P<0.01. (F,G) qRT-PCR indicated that the transcript expression levels of the BMP-regulated col10a1a (F) and TGFβ-regulated col2a1a (G) are reduced in galnsm/m mutant larvae at 4 and 8 dpf. n=5 biological replicates of samples containing 20 embryos per genotype per timepoint. Error bars=s.e.m. Two-tailed paired Student's t-test, **P<0.01, ***P<0.001, ****P<0.0001. (H) Live confocal images of fli1a:EGFP;BRE:dmKO-positive galns+/+ (wild-type) and galnsm/m (mutants) larvae treated at 3-4 dpf with either 0.025% DMSO or 25 µM E64d suggest that limited cathepsin inhibition does increase BMP signaling. Scale bar: 10 µm. (I) Graphs showing the percent of cells in larvae of each genotype and treatment condition with dmKO signal or fluorescence intensity in cartilage. n=17 larvae per genotype per treatment condition. Error bars=s.e.m. One-way ANOVA with Dunnett's multi-comparison correction, *P<0.05, **P<0.01, ***P<0.001. (J,K) Graphs show that the cell shape (J) and the percentage of cells intercalated (K) both improve in galnsm/m mutant larvae treated with E64d. Data from wild-type larvae are presented in Fig. S3. n=17 larvae per genotype per treatment condition from three independent matings. Error bars=s.e.m. One-way ANOVA with Dunnett’s multi-comparison correction, ****P<0.001. (L) Live confocal images of fli1a:EGFP-positive larvae show that E64d treatment does not improve vacuolation in galnsm/m mutants. Yellow arrows highlight vacuoles. (M) Graph shows that E64d treatment does not improve vacuolation. Error bars=s.e.m. Two-tailed paired Student's t-test, P<0.05. Data from wild-type larvae are presented in Fig. S3.
Global analyses of cysteine cathepsin activity using the BMV109 ABP indicated that Ctsk activity was increased in galns^m/m^ mutant larvae compared to that in wild-type larvae between 2 and 3 dpf (see Fig. 4). In light of this and findings in vitro showing that Ctsk can inactivate BMP growth factors (Flanagan-Steet et al., 2016), we asked whether this short-term increase in proteolytic activity contributes to the loss of BMP signaling. To do this, embryos expressing the fli1a:EGFP and BRE:dmKO transgenes were treated at 60 hpf with pan-reactive cysteine cathepsin inhibitor E64d (also known as aloxistatin). Analysis of dmKO levels in wild-type and mutant cartilages showed that short-term cathepsin inhibition does increase the level of BMP signaling (Fig. 6H,I). Increased dmKO levels corresponded with improved shape and organization of mutant chondrocytes (Fig. 6J; Fig. S4A,B) but had no effect on cellular vacuolation (Fig. 6K,L; Fig. S4A,B). In fact, in several cases, cathepsin inhibition exacerbated vacuolation, perhaps because E64d treatment concomitantly also reduced intracellular cathepsin activities. Importantly, later-stage application of E64 did not improve BMP signaling or cartilage phenotypes in galns^m/m^ mutant larvae. This is consistent with the finding that, after 3 dpf, Ctsk and Ctsl activities decreased in galns^m/m^ mutant larvae relative to those in wild-type larvae (see Fig. 4K,L).
Pharmacological inhibition of lysosomal exocytosis improves cartilage pathology in galnsm/m mutants
We previously showed that the pharmacological agent vacuolin-1, which inhibits Ca^+2^-dependent fusion of lysosomes with the plasma membrane, reduces lysosomal exocytosis in developing cartilage (Lee et al., 2024). To more directly assess whether increased lysosomal exocytosis contributes to cartilage pathology, 3 dpf larvae were treated with vacuolin-1. Confocal analysis of treated and untreated fli1a:EGFP-positive larvae showed that inhibiting exocytosis improves multiple aspects of cartilage pathology in galns^m/m^ mutants (Fig. 7A-D). Most notably, vacuolin treatment significantly improved the shape of mutant chondrocytes and reduced the level of vacuolation (Fig. 7C,D). Using the BRE:dmKO reporter line, we further showed that these improvements corresponded to an increase in BMP signaling (Fig. 7E,F). Similar studies with the SBE:mCherry transgenic reporter indicated that the level of TGFβ signaling was also increased when lysosomal exocytosis was inhibited (Fig. 7G-I). These findings suggest that loss of Galns increases lysosomal exocytosis, disrupting TGFβ-related signaling and impairing chondrocyte differentiation. Because vacuolin treatment may also affect other forms of vesicle fusion, we cannot rule out the possibility that defects in intracellular trafficking also contribute to these phenotypes in galns^m/m^ larvae.
*Pharmacological inhibition of lysosomal exocytosis improves cartilage pathology in galnsm/m mutant cartilage. (A,B) Confocal images of fli1a:EGFP;BRE:dmKO-positive larvae show that treatment with 35 nM vacuolin-1 improves chondrocyte shape and increases BMP signaling within the central (A) and lateral (B) portions of Meckel's cartilage in galnsm/m mutants. Scale bars: 20 µm. (C,D) Graphs show quantitative measurements of cell shape and the level of vacuolation in chondrocytes within the central (C) and lateral (D) regions. Error bars=s.e.m. One-way ANOVA with Dunnett's multi-comparison correction, *P<0.05, **P<0.01, ***P<0.00, ****P<0.0001. (E,F) Graphs show the level of fluorescent intensity in dmKO-positive cells within the central (E) and lateral (F) regions. dmKO was normalized to the local EGFP level in the cell. Error bars=s.e.m. One-way ANOVA with Dunnett's multi-comparison correction, *P<0.05, **P<0.01, ****P<0.0001. (G) Confocal images of fli1a:EGFP;SBE:mCherry-positive larvae show that treatment with 35 nM vacuolin-1 increases TGFβ signaling. (H) Graphs show the level of fluorescent intensity in mCherry-positive cells within the central (left) and lateral (right) regions. mCherry was normalized to the local EGFP level in the cell. Error bars=s.e.m. One-way ANOVA with Dunnett's multi-comparison correction. *P<0.05, **P<0.01. (I) Graph shows the percentage of cells with a high versus medium level of fluorescent signal in treated and untreated larvae. Error bars=s.e.m. One-way ANOVA with Dunnett's multi-comparison correction, P<0.05.
RNA sequencing of 8 dpf larvae supports additional alterations in MAPK signaling, and lens and vascular development
In addition to their impact on cathepsin activity, the chondroitin sulfate and heparan sulfate GAGs play important roles in establishing growth factor gradients, regulating growth factor bioavailability, and promoting interactions between several growth factors and their receptors (Raman et al., 2005). To ask whether the loss of Galns and the associated changes in extracellular GAG levels impairs other pathways important for early development, we sequenced total RNA isolated from 8 dpf wild-type and galns^m/m^ mutant larvae (Fig. 8). Analyses of the differentially expressed genes identified alterations in transcripts involved in the regulation of mitogen activated protein kinases (MAPKs), lens fiber cell differentiation and multiple aspects of blood vessel function (Fig. 8A-C). Gene Ontology (GO) analysis suggested that the top processes altered at this timepoint in galns^m/m^ mutants relate to ‘inhibition of MAPK activity’ (Fig. 8B,C). In particular, the transcript expression levels of four different dual specificity phosphatases (dusp1, dusp2, dusp4 and dusp6) were significantly reduced in galns^m/m^ mutant larvae relative to those in wild-type larvae (Fig. 8C,D). The Dusps regulate the ability of the MAPKs JNK, ERK and p38 to modulate cellular proliferation and differentiation, to promote immune system development, and to mediate cellular responses to stress and immunological insult (Patterson et al., 2009). Alterations in expression of the Dusp genes is particularly compelling given the role of these MAPKs as downstream effectors of FGF signaling, which depends on the presence of heparan sulfate (Huang and Tan, 2012). Analysis of GAG content showed that heparan sulfate is globally (albeit not significantly) reduced in galns^m/m^ mutant larvae and also locally in mutant cartilages (see Fig. 5). To ask whether disruptions in FGF signaling are relevant to phenotypes in galns^m/m^ mutants, we used the transgenic reporter Tg(dusp6:EGFP) (Bellesso et al., 2018) to assay its general activity at 8 dpf. Live confocal analyses of wild-type and mutant larvae expressing the dusp6 transgene did identify regions in the mutant heads in which EGFP levels were reduced (Fig. S5). Consistent with the RNA-sequencing findings, reduced Dusp6 activity was particularly evident in blood vessel endothelia and in the nasal placodes. Slight differences were also noted in the craniofacial cartilages of a subset of mutant larvae, so we cannot rule out the possibility that alterations in FGF signaling also contribute to aspects of skeletal pathology and galns^m/m^ mutant larvae.
RNA-sequencing analyses reveal changes in transcript expression levels of multiple genes associated with MAPK signaling and vascular and ocular development. (A) Volcano plot highlights numerous genes with increased or decreased transcript expression in galnsm/m mutant larvae. FDR, false discovery rate. (B) Gene Ontology (GO) analyses suggest that the most significant changes in transcript expression occur in genes associated with inhibition of MAPK activity, lens fiber differentiation and multiple aspects of blood vessel regulation. (C) Table shows genes for which transcript changes were associated with key GO categories implicated in mutant larvae. DE, differentially expressed. (D) Heat maps show the degree of change in specific genes for the identified GO categories. The P-value associated with their change is listed.
Further analyses of transcript abundance suggested that several other processes are also impaired in galns^m/m^ mutants, including processes related to glycosyl metabolism and lactate catabolism, extracellular matrix organization and cellular stress responses (Fig. 8B-D). The high incidence of alterations in genes related to lens development was also compelling given the fact that patients with MPSIVA often experience corneal clouding, optic atrophy and retinal degeneration (Del Longo et al., 2018). Additionally, as described above, galns^m/m^ mutant larvae have smaller eyes than those of wild-type larvae, with small abnormal lens openings (Fig. S1D,E). It is currently unclear whether this stems from impaired lens formation, increased retinal tissue or corneal clouding. It should also be noted that, despite qRT-PCR-confirmed differences in cathepsin transcript abundances (see Fig. 4M,N; Fig. S3), the cathepsin proteases were not well represented in the RNA-sequencing dataset. Further, although the expression of ctsb was seemingly reduced, qRT-PCR studies indicate that ctsb expression was actually increased in mutant larvae compared to that in wild-type larvae. These differences most likely stem from the limited number of ctsb reads present in the transcriptomic dataset.
DISCUSSION
We present evidence from a powerful zebrafish model for the complex regulation of lysosomal exocytosis in MPSIVA, demonstrating the involvement of GAG- and cathepsin-mediated pathogenesis in this disorder. Previous studies in zebrafish models of mucolipidosis II and sialidosis showed that increased lysosomal exocytosis enhances the activity of multiple cathepsin proteases (Flanagan-Steet et al., 2016, 2018; Lee et al., 2024). In both cases, unregulated exocytosis was found to disrupt the TGFβ-related pathways that regulate cartilage development. Analysis of galns mutant zebrafish showed that lysosomal exocytosis is also increased in MPSIVA chondrocytes, albeit to a lesser degree. As noted in MLII and sialidosis, enhanced lysosomal exocytosis caused cathepsin proteases to be inappropriately released from galns mutant chondrocytes, where they disrupt TGFβ-related signaling. Despite these similarities, the impact exocytosis has on protease activity and growth factor signaling differs among the models. In MLII and sialidosis, a sustained increase in cathepsin activity was noted across all developmental timepoints tested. However, in galns mutants, although short-term increases were noted at early developmental stages, in older larvae, activities of Ctsk and Ctsl are both reduced. In galns mutants, the loss of cathepsin activity was associated with reductions in both TGFβ and BMP signaling, findings consistent with the more attenuated skeletal phenotypes characteristic of MPSIVA.
The mechanisms that underlie increased exocytosis in the galns mutant are not currently clear. Based on the finding that neuraminidase activity is only modestly reduced, we do not believe that the primary carbohydrate-dependent mechanisms that drive hydrolytic enzyme release in MLII and sialidosis explain exocytosis in galns mutants. Furthermore, M6P-dependent lysosomal sorting should not be affected by loss of Galns. Because a substantial increase in disease-specific chondroitin/dermatan sulfate NRE species was detected in galns mutants, it is possible that GAG accumulation triggers exocytosis in a subset of chondrocytes. Importantly, GAG accumulation has been shown to modulate the lysosomal ion channels and membrane proteins involved in this process (Orts and Arcisio-Miranda, 2022; Leal et al., 2022). Investigating whether exocytosis is enhanced in all tissues in MPSIVA or restricted to cartilage and bone, and whether this correlates with GAG storage will be an important area of focus for future studies.
In galns mutants, the combined reductions in proteolytic capacity and TGFβ signaling also correspond to altered abundance of cell surface GAGs. Mass spectrometry-based analysis of whole animals suggested that the abundance of chondroitin sulfate and dermatan sulfate is decreased at 6 dpf, with less heparan sulfate apparent by 10 dpf. Although these differences were not statistically significant, the consistent trend suggested that more substantial changes occur within the cartilage and bone. Immunohistochemical analysis confirmed that both chondroitin sulfate and heparan sulfate are, in fact, reduced in chondrocytes as early as 4 dpf. Further NRE analysis confirmed that lower levels of surface-localized chondroitin are matched by increased intracellular accumulation, implicating the presence of lysosomal storage. This is a noteworthy finding that distinguishes MPSIVA pathogenesis from that of MLII and sialidosis, for which storage was not detected at these early stages in developing zebrafish cartilage. Differences in the content of intracellular and extracellular GAGs provides a plausible explanation for why cathepsin activities are ultimately reduced in MPSIVA but increased in both MLII and sialidosis. In MLII, increased abundance of extracellular chondroitin sulfate was shown to promote the ability of CTSK to activate the latent TGFβ growth factors present in cartilage matrices (Flanagan-Steet et al., 2018). In galns mutants, reduced levels of heparan sulfate and chondroitin sulfate correlate with less cathepsin activity and lower levels of TGFβ signaling. Together, these findings are consistent with in vitro data showing that chondroitin sulfates, namely chondroitin-4-O-sulfate, promote Ctsk processing and activity, whereas heparan sulfate inhibits it. Our collective studies support the model that local GAG content influences cathepsin function, growth factor availability and cellular signaling during cartilage and bone development. We propose in LSDs that the phenotypic consequences of this mechanism likely depend on which GAGs accumulate, and whether enhanced exocytosis or enzyme secretion causes cathepsin proteases to be mislocalized outside cells.
MATERIALS AND METHODS
Zebrafish strains and husbandry
Animals were maintained according to standard protocols. The galns mutant line (sa31602) was generated by the Sanger Center’s TILLING screen and ultimately acquired from the Zebrafish International Resources Center (ZIRC). Primers used to amplify a region surrounding the galns c.539+1G>A (accession NM_001080641) mutation are listed in Table S1. The galns^+/m^ were identified by high-resolution melt analysis of genomic DNA (gDNA) extracted from embryos or fin tissue using an Extracta DNA Prep for PCR kit (95091-025, Quanta Biosciences). Genotyping of adult age and larval samples was performed by high-resolution melt curve analyses using Precision Melt Supermix (1725112, Bio-Rad) (Klaver et al., 2021). galns^+/m^ animals were outcrossed with the TLAB wild-type line for five generations prior to initiating experimental analyses. Stock lines were continually outcrossed throughout the study. The Tg(fli1a:EGFP)y1 was also acquired from ZIRC (Lawson and Weinstein, 2002). The Tg(hsp70:lamp1-mCherry) line was provided by Dr Michel Bagnat (Duke University) (Ellis et al., 2013). The Tg(SBE:nlsmCherry) TGFβ reporter lines and the Tg(dusp6:EGFP) reporter line were provided by Dr Enrico Moro (Bellesso et al., 2018). The Tg(BRE:dmKO) line was provided by Dr Brian Link (Collery and Link, 2011). Embryo and larval staging were performed according to established criteria (Kimmel et al., 1995). For imaging experiments, 0.003% 1-phenyl 2-thiourea (PTU) was added to embryo medium to block pigmentation.
RT-PCR and qRT-PCR analyses
Total RNA was isolated from samples containing 20 larvae using TRIzol™ Reagent (15596018, Invitrogen™, Thermo Fisher Scientific) according to the manufacturer's instructions. Contaminating gDNA was removed by DNAse digestion with a TURBO DNA-free™ Kit (31 AM1907, Invitrogen™, Thermo Fisher Scientific). Complementary DNA (cDNA) was synthesized from 500 ng RNA using qScript cDNA SuperMix (95048-025, Quanta Biosciences). gDNA contamination was checked in control samples lacking reverse transcription. ribosomal protein L4 (rpl4) was used as a normalization control as described previously (Flanagan-Steet et al., 2016). qRT-PCRs were performed using PerfeCTa SYBR Green FastMix (31 95072-250, Quanta Biosciences) on a CFX96 Real Time System (C1000, Bio-Rad). The sequences of all primers used are listed in Table S1. Data were analyzed using CFX Maestro 1.1 (version 4.1.2433.1219) software.
RNA-sequencing and data analyses
Fifteen 8 dpf larvae per genotype per sample were harvested into 1 ml Trizol Reagent (15596018, Invitrogen™, Thermo Fisher Scientific) and shipped on dry ice to the Azenta Genewiz Laboratories (Plainfield, NJ, USA) for library generation, poly A enrichment, RNA sequencing and data analysis. RNA sequencing was performed to a read depth of 50 million reads per sample. Sequence reads were trimmed using Trimmomatic v.0.36. The trimmed reads were mapped to the Danio rerio GRCz10.89 reference genome available on ENSEMBL using the STAR aligner v.2.5.2b. From the resulting BAM files, unique gene hit counts were calculated by using featureCounts from the Subread package v.1.5.2. Only unique reads that fell within exon regions were counted. The gene hit counts table was used for downstream analysis. Raw sequencing output files and the final gene hit count table are available at Gene Expression Omnibus (GEO; GSE300957). Differential expression analysis between mutant and wild-type samples was performed using DESeq2 (Love et al., 2014). The Wald test was used to generate P-values and log_2_ fold changes. All genes with false discovery rate<0.05 were used for pathway enrichment analysis. Identifiers for significant genes (either upregulated or downregulated) were fed to FishEnrichr (Chen et al., 2013; Kuleshov et al., 2016). Enrichment results for the GO Biological Process ontology were presented, as well as the significant genes that were found to overlap in enriched pathways.
Oxford Nanopore Technologies long-read sequence analysis of galns transcripts
Total RNA used for transcript analysis was isolated from pools of 15 genotyped larvae as described above. cDNA was generated and PCR amplification was performed as described above. Resultant amplicons were sequenced using a method adapted from Jurkute et al. (2022). 10 µl PCR reaction was purified (Ampure XP, Beckman Coulter Inc.). 2 fmol amplicon was used for end preparation (Ultra II end-prep kit, NEB) and native barcoding (SQK-NBD114.24, Oxford Nanopore Technologies). Barcoded samples were pooled for adapter ligation and sequenced overnight using a Flongle flowcell and MinION sequencer. Base calling was performed using the super-high accuracy model within the MinKnow GUI (v24-.6.14). After sequencing, the FASTQ read data were aligned to the zebrafish reference genome (GRCz11) using minimap2 (v2.28), processed with SAMtools (v1.21) to BAM format and visualized using the Integrative Genomics Viewer.
Zebrabox motility assay
Larvae were placed one per well into 12-well plates (Cellstar, Greiner-Bio One) containing 2 ml embryo medium. Locomotor activity was monitored from 5 to 12 dpf using the Zebrabox System (ViewPoint Inc.). During the analyses, plates were placed into the sound deprivation component of the Zebrabox system and desensitized for 15 min. Swim path and behavior were subsequently recorded for 10 min. All data were exported from the Zebrabox system software and analyzed using GraphPad Prism software (Version 8.1.0).
NRE analysis of lysosomal GAG storage
8 dpf larvae were homogenized and lysed in 0.5% CHAPS lysis buffer (50 mM HEPES, 120 mM NaCl, 2 mM EDTA, pH 7.4) containing a protease inhibitor cocktail (Roche). 50 µl cell lysate was set aside for protein quantification via bicinchoninic acid (BCA) assay. Homogenates were diluted 1:10 in a wash buffer (50 mM sodium acetate, 200 mM NaCl, 0.1% Triton X-100, pH 6.0) and incubated with Pronase (0.4 mg/ml; Sigma-Aldrich) overnight at 37°C with mild agitation. The product was centrifuged (4000 g, 20 min) then passed through a DEAE-Sephacel (Cytiva) column equilibrated in 50 mM sodium acetate buffer (pH 6.0), containing 200 mM NaCl, and passed through a PD-10 desalting column (Cytiva). The purified GAGs were subsequently enzymatically depolymerized with chondroitinase ABC (Sigma-Aldrich) and differentially mass labeled by reductive amination with [^12^C_6_] aniline, as described (Basu et al., 2025). Each sample was then mixed with [^13^C_6_] aniline-tagged GalNAc-6S (Biosynth; 20 pmol). The injection volume was 2.0 µl. Samples were analyzed by hydrophilic interaction liquid chromatography coupled with quadrupole time-of-flight mass spectrometry (HILIC-Q-TOF-MS). NREs were quantified using the isotopically labeled internal standard and normalizing to total protein, as measured by BCA assay. Fig. S2 details the method development, retention time and mass spectrum data for GalNAc-6S NRE species detected in zebrafish samples.
GAG analyses
Samples containing 20 larvae per sample per genotype were harvested at 6 and 10 dpf. Samples were lysed by sonication in Tris buffer containing 0.5% NP40. Total protein content was determined using a Micro-BCA protein assay kit (31 23235, Thermo Fisher Scientific). Total GAGs were isolated using a previously described standard methanolysis protocol (Zhang et al., 2023). Mass spectrometry-based analyses of chondroitin sulfate, dermatan sulfate, heparin sulfate and keratan sulfate were performed as previously described (Martell et al., 2011).
SDA-PAGE and western blotting
Samples containing 15 larvae per genotype were harvested at timepoints indicated; animals less than 5 dpf were manually deyolked prior to harvesting. Samples were lysed in a Tris-HCl buffer (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.5 mM EDTA, 0.5% Triton X-100) containing protease inhibitor cocktail (31 A32965, Thermo Fisher Scientific). Samples were homogenized on ice by probe sonication and protein concentration determined using the Micro-BCA protein assay kit (31 23235, Thermo Fisher Scientific). Equal amounts of protein were loaded on SDS-PAGE gels. The equivalence of protein loaded was assayed by Ponceau S staining, where the protein content of each lane was assessed by pixel densitometry using ImageJ software. This value was used as a normalizing factor for densitometry-based quantitation of protein abundance following western blot analysis. Primary antibodies utilized included an anti-mCherry antibody (1:1000; ab167453, Abcam) and an anti-Ctsk antibody (1:500; ab19027, Abcam). Appropriate horseradish peroxidase-labeled secondary antibodies were used, and blots were analyzed using a Bio-Rad MP Chemidoc system.
Analyses of lysosomal enzyme activity
Lysosomal enzyme activities were analyzed as previously described (Lee et al., 2024). Briefly, samples containing 20 larvae were harvested at the timepoints indicated and lysed in a Tris-HCl buffer containing protease inhibitors. Samples were homogenized by sonication, and 20 µl homogenized sample was incubated with 80 µl substrate (0.69 mM 4-MU-N-acetyl-α-D-neuraminic acid, pH 4.4) for 30 min at 37°C. The reaction was stopped with 3.9 ml 2-amino-2-methyl-1-propanol (0.1 M, pH 10.3) before reading fluorescence. 4-Methylumbelliferone released by the reaction was calculated using a standard curve. Enzyme activity was reported as nmol/h/mg protein. PPCA activity was measured as previously described (Lee et al., 2024). Briefly, 15 μg larval lysate was diluted in enzyme dilution buffer with or without 1.5 mM N-carbobenzoxy-L- phenylalanyl-L-alanine (31 SCP0262, Sigma-Aldrich) substrate. A dilution series of L-alanine (31 W381829, Sigma-Aldrich) was used as a standard to assign measured unit (nmol/min/mg).
Histochemical and immunohistochemical analysis
Alcian Blue staining was performed as previously described (Flanagan-Steet et al., 2009). Stained animals were photographed on an Olympus SZ16 stereoscope outfitted with an Olympus DP73 Camera. Cartilage structures were measured using Adobe Photoshop (CS6, Version 13.0). To account for differences in embryonic size, all measurements were normalized to the distance between the eyes.
For immunohistochemical analyses, larvae were fixed at 4°C for 4-6 h with 4% paraformaldehyde and processed for O.C.T. embedding and sectioning as previously described (Flanagan-Steet et al., 2018). 20 µm sections were cut on a Leica cryostat. Primary antibodies used for staining included an anti-GFP antibody (1:100; 31 A-21311, Thermo Fisher Scientific), an anti-heparan sulfate (1:100; clone F58-10E4, Amsbio) and an anti-chondroitin sulfate antibody (1:100; MA1-83055, Invitrogen, Thermo Fisher Scientific). Secondary antibodies utilized included anti-mouse Alexa 568 (1:400; 31A-11031, Invitrogen) and anti-GFPAlexa488 (1:200; 31FAB42401G, R&D Systems). In some cases, sections were also treated with Alexa 350-conjugated WGA (1:150; 31 W11263, Invitrogen, Thermo Fisher Scientific). Images were acquired on an Olympus FV3000 laser-scanning confocal microscope. Images were processed for publication using ImageJ (Java 64-bit 1.52K) and Adobe Photoshop (CS6, Version 13.0).
Pharmacological treatments
Cysteine cathepsins were inhibited by adding 25 nM E64d (31 HY-100229, MedChemExpress) added to the embryo media of 3 dpf larvae for 24 h. E64d was resuspended in DMSO, with a final concentration of 0.25% DMSO in treatment wells. Control samples were treated in parallel with 0.025% DMSO. To modulate lysosomal exocytosis, embryos were treated with 35 nM vacuolin-1 (C4084, APEXBIO) by injecting 2 nl of a drug from a 5 mM stock into the pericardial space of embryos at 3 dpf.
Heat shock and lamp1 transgene analysis
For analyses of Lamp-mCherry half-life and cellular location, transgene expression was induced by heat shock. Briefly, embryos were exposed to embryo medium pre-warmed to 39°C. Animals were incubated for 30 min in a 39°C water bath. For western blot analyses, ten 4 dpf embryos per sample were harvested at 30 min, 3 h, 5 h, 10 h and 24 h phs, and lysates were prepared as described above. For live confocal microscopy, larvae were mounted as previously described 5 h phs, and imaging was performed at room temperature (Lee et al., 2024).
BMV109 analyses
Analyses of cathepsin activity with the BMV109 ABP were performed as previously described (Flanagan-Steet et al., 2018). Briefly, 1 nl of a 10 µM solution of the BMV109 ABP was microinjected pericardially into embryos at the timepoints indicated. 25 embryos per condition were harvested 16 h post-injection and lysed by brief sonication in citrate buffer (50 mM citrate buffer pH5.5, 5 mM DTT, 0.5% CHAPS, 0.75% Triton X-100). Samples were centrifuged for 15 min at 15,000 g, and the supernatant was collected. Protein concentration was determined via a Micro-BCA assay, and Cy5 fluorescence was analyzed in SDS-PAGE gels on the Bio-Rad ChemiDoc system.
Sex as a biological variable
Sex as a biological variable is not a consideration for experiments involving embryo- or larval-stage zebrafish (1-10 dpf). Zebrafish sex does not depend on a sex chromosome and is not established until 2-3 months of age.
Statistical analyses
All experiments involving analyses of phenotypic rescue were performed by a person unaware of the experimental condition while acquiring and analyzing the measurements. All results are expressed as mean±s.e.m. Statistical analyses were performed using GraphPad Prism (Version 7.0a) software. For comparison of two paired groups, a two-tailed paired Student's t-test was used. For other parametric data, a one-way ANOVA was performed, followed by Dunnett's multiple comparisons test. P<0.05 was considered significant.
Study approval
Handling and euthanasia of fish for all experiments complied with the Greenwood Genetic Center (GGC) policies, as approved by the GGC Institutional Animal Care and Use Committee (permit #A2022 01-003-Y3).
Supplementary Material
10.1242/dmm.052582_sup1Supplementary information
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Ballabio, A. and Bonifacino, J. S. (2020). Lysosomes as dynamic regulators of cell and organismal homeostasis. Nat. Rev. Mol. Cell Biol. 21, 101-118. 10.1038/s 41580-019-0185-431768005 · doi ↗ · pubmed ↗
- 2Basu, A., Archer-Hartmann, S., Chopra, P., Taherzadeh Ghahfarrokhi, M., Dong, X., Patel, N. G., Zhang, Y., Choudhury, B., Funato, K., Yellajoshyula, D. et al. (2025). Quantitative HILIC-Q-TOF-MS analysis of glycosaminoglycans and non-reducing end carbohydrate biomarkers via glycan reductive isotopic labeling. Anal. Chem. 97, 17490-17500. 10.1021/acs.analchem.5c 0233840767435 PMC 12368834 · doi ↗ · pubmed ↗
- 3Bellesso, S., Salvalaio, M., Lualdi, S., Tognon, E., Costa, R., Braghetta, P., Giraudo, C., Stramare, R., Rigon, L., Filocamo, M. et al. (2018). FGF signaling deregulation is associated with early developmental skeletal defects in animal models for mucopolysaccharidosis type II (MPSII). Hum. Mol. Genet. 27, 2262-2275. 10.1093/hmg/ddy 13129648648 · doi ↗ · pubmed ↗
- 4Bertolin, J., Sánchez, V., Ribera, A., Jaén, M. L., Garcia, M., Pujol, A., Sánchez, X., Muñoz, S., Marcó, S., Pérez, J. et al. (2021). Treatment of skeletal and non-skeletal alterations of Mucopolysaccharidosis type IVA by AAV-mediated gene therapy. Nat. Commun. 12, 5343. 10.1038/s 41467-021-25697-y 34504088 PMC 8429698 · doi ↗ · pubmed ↗
- 5Chen, E. Y., Tan, C. M., Kou, Y., Duan, Q., Wang, Z., Meirelles, G. V., Clark, N. R. and Ma'ayan, A. (2013). Enrichr: interactive and collaborative HTML 5 gene list enrichment analysis tool. BMC Bioinformatics 14, 128. 10.1186/1471-2105-14-12823586463 PMC 3637064 · doi ↗ · pubmed ↗
- 6Cirillo, F., Ghiroldi, A., Fania, C., Piccoli, M., Torretta, E., Tettamanti, G., Gelfi, C. and Anastasia, L. (2016). NEU 3 sialidase protein interactors in the plasma membrane and in the endosomes. J. Biol. Chem. 291, 10615-10624. 10.1074/jbc.M 116.71951826987901 PMC 4865910 · doi ↗ · pubmed ↗
- 7Collery, R. F. and Link, B. A. (2011). Dynamic smad-mediated BMP signaling revealed through transgenic zebrafish. Dev. Dyn. 240, 712-722. 10.1002/dvdy.2256721337469 PMC 3072245 · doi ↗ · pubmed ↗
- 8Costa, R., Bellesso, S., Lualdi, S., Manzoli, R., Pistorio, V., Filocamo, M. and Moro, E. (2020). A transcriptional and post-transcriptional dysregulation of Dishevelled 1 and 2 underlies the Wnt signaling impairment in type I Gaucher disease experimental models. Hum. Mol. Genet. 29, 274-285. 10.1093/hmg/ddz 29331816052 · doi ↗ · pubmed ↗