Starch Binding Domain-containing Protein 1 Plays a Dominant Role in Glycogen Transport to Lysosomes in Liver*

A small portion of cellular glycogen is transported to and degraded in lysosomes by acid α-glucosidase (GAA) in mammals, but it is unclear why and how glycogen is transported to the lysosomes. Stbd1 has recently been proposed to participate in glycogen trafficking to lysosomes. However, our previous study demonstrated that knockdown of Stbd1 in GAA knock-out mice did not alter lysosomal glycogen storage in skeletal muscles. To further determine whether Stbd1 participates in glycogen transport to lysosomes, we generated GAA/Stbd1 double knock-out mice. In fasted double knock-out mice, glycogen accumulation in skeletal and cardiac muscles was not affected, but glycogen content in liver was reduced by nearly 73% at 3 months of age and by 60% at 13 months as compared with GAA knock-out mice, indicating that the transport of glycogen to lysosomes was suppressed in liver by the loss of Stbd1. Exogenous expression of human Stbd1 in double knock-out mice restored the liver lysosomal glycogen content to the level of GAA knock-out mice, as did a mutant lacking the Atg8 family interacting motif (AIM) and another mutant that contains only the N-terminal 24 hydrophobic segment and the C-terminal starch binding domain (CBM20) interlinked by an HA tag. Our results demonstrate that Stbd1 plays a dominant role in glycogen transport to lysosomes in liver and that the N-terminal transmembrane region and the C-terminal CBM20 domain are critical for this function.

A small portion of cellular glycogen is transported to and degraded in lysosomes by acid ␣-glucosidase (GAA) in mammals, but it is unclear why and how glycogen is transported to the lysosomes. Stbd1 has recently been proposed to participate in glycogen trafficking to lysosomes. However, our previous study demonstrated that knockdown of Stbd1 in GAA knockout mice did not alter lysosomal glycogen storage in skeletal muscles. To further determine whether Stbd1 participates in glycogen transport to lysosomes, we generated GAA/Stbd1 double knock-out mice. In fasted double knock-out mice, glycogen accumulation in skeletal and cardiac muscles was not affected, but glycogen content in liver was reduced by nearly 73% at 3 months of age and by 60% at 13 months as compared with GAA knock-out mice, indicating that the transport of glycogen to lysosomes was suppressed in liver by the loss of Stbd1. Exogenous expression of human Stbd1 in double knock-out mice restored the liver lysosomal glycogen content to the level of GAA knock-out mice, as did a mutant lacking the Atg8 family interacting motif (AIM) and another mutant that contains only the N-terminal 24 hydrophobic segment and the C-terminal starch binding domain (CBM20) interlinked by an HA tag. Our results demonstrate that Stbd1 plays a dominant role in glycogen transport to lysosomes in liver and that the N-terminal transmembrane region and the C-terminal CBM20 domain are critical for this function.
Liver and muscle are the two primary sites for glycogen metabolism in mammals. Glycogen is synthesized and mostly degraded in the cytoplasm, but a small percentage of glycogen is transported into the lysosomes and hydrolyzed by the enzyme acid ␣-glucosidase (GAA) 3 (1)(2)(3). Deficiency of GAA causes Pompe disease (glycogen storage disease type II), resulting in a progressive lysosomal glycogen accumulation and dysfunction in cardiac and skeletal muscle and other tissues (4 -6). Enzyme replacement therapy with recombinant human GAA is the only available therapy for this disorder to have remarkable success, but it also has significant limitations such as high titer antibody formation and difficulties in delivering the therapeutic protein to skeletal muscle tissues (7)(8)(9)(10)(11). A conceivable alternative way to treat Pompe disease would be to block the transport of cytoplasmic glycogen to the lysosomes. Thus, understanding how glycogen enters the lysosome is a critical step in finding new therapeutic targets for Pompe disease. It has been demonstrated that conventional autophagy (macroautophagy) plays an important role in the transport of glycogen to lysosomes in skeletal muscle (12). Specifically, knock-out of an autophagy key gene Atg7 resulted in a decrease of glycogen content by 50 -60% in skeletal muscles of GAA knock-out mice (12). This suggests that other routes exist for glycogen transport to lysosomes.
In recent studies, starch binding domain-containing protein 1 (Stbd1) has gained attention for its potential involvement in glycogen transport. The Stbd1 protein contains a putative N-terminal transmembrane region, an Atg8 interacting motif (AIM), and a C-terminal carbohydrate binding module family 20 (CBM20) (13)(14)(15). The observation that it binds to glycogen and interacts with the autophagy protein GABARAPL1 in cultured cells led to speculations that Stbd1 may function as a cargo-binding protein that delivers glycogen to lysosomes in a specific autophagic pathway (13,14). Contradictorily, our previous study undermined this proposition, as significant knockdown of Stbd1 in skeletal muscle of GAA knock-out (GKO) mice did not affect lysosomal glycogen storage in the tissue (16). In this study, we generated Stbd1/GAA double knock-out (dKO) mice to explore the role of Stbd1 in glycogen transport to lysosomes, as accumulated lysosomal glycogen is an easily measurable indicator of glycogen transport to this organelle.

Stbd1 Plays a Dominant Role in Transporting Glycogen to
Lysosomes in Liver, but Not in Heart and Skeletal Muscles-The lack of GAA in GKO and dKO mice enables us to monitor glycogen transport to lysosomes, because the glycogen cannot be degraded once inside lysosomes and glycogen build-up/storage can be readily observed/quantified. To assess whether the loss of Stbd1 affects glycogen transport to lysosomes, 3-monthold male GKO and dKO mice were subjected to overnight fasting and glycogen contents were measured in different tissues. Age-and gender-matched Stbd1 knock-out (SKO) mice and their littermate WT mice were used as controls.
As shown in Fig. 1A, both SKO and WT mice had very low levels of glycogen in heart, skeletal, and liver tissues. The GKO mice showed high levels of lysosomal glycogen storage in these tissues as seen in the GAA knock-out (6 neo /6 neo ) mice (17). There were no significant differences in glycogen contents in cardiac and skeletal muscles between the dKO mice and GKO mice. Surprisingly, liver glycogen content in the dKO mice (11.8 Ϯ 3.9 mol of glucose/g) was only 27% of that in the GKO mice (43.7 Ϯ 4.6 mol of glucose/g). Consistent with this result, periodic acid-Schiff (PAS) staining of liver sections showed strong glycogen staining in the GKO mice and markedly weaker staining in the dKO mice. No obvious PAS-positive glycogen was observed in the livers of WT and SKO mice (Fig. 1B). Our data indicate that Stbd1 plays a significant role in glycogen transport to the lysosomes in liver, but not in skeletal muscle and heart. The fact that liver glycogen level in dKO mice was still significantly higher than in WT (Fig. 1, A and B) indicates that a small amount of glycogen was still transported to lysosomes in the absence of Stbd1. To determine the impact of the Stbd1-independent pathway(s) on lysosomal glycogen accumulation over time, we examined liver glycogen levels in old dKO mice at 13 months of age. As shown in Fig. 1C, fasting liver glycogen content in the 13-month-old dKO mice (20.0 Ϯ 5.8 mol of glucose/g) was significantly higher than that of the 3-month-old dKO mice (11.8 Ϯ 3.9 mol of glucose/g), but it was still far below the level in age-matched GKO mice (50.6 Ϯ 8.1 mol of glucose/g). This demonstrates that other minor routes may have limited contribution to the transport of glycogen into lysosomes, and the Stbd1-mediated pathway plays a dominant role in this process.
To exclude the possibility that the reduced lysosomal glycogen accumulation in dKO mouse liver was due to suppression of glycogen synthesis in the absence of Stbd1, we compared liver glycogen contents in WT, GKO, and dKO mice at 3 months of age at fed state. No significant difference in liver glycogen contents was found among these genotypes (Fig. 1D). Glycogen level in GKO mice was seemingly higher, possibly due to an extra glycogen pool trapped in the lysosomes. We also examined the expression of the two key enzymes for glycogen synthesis, glycogen synthase 2 (GS2) and glycogen branching enzyme (GBE1). There was no difference in expression of either enzyme between GKO and dKO livers as determined by Western blotting (supplemental Fig. S1). Thus, Stbd1 knock-out does not affect liver glycogen synthesis.
The N-terminal Transmembrane Region and the C-terminal CBM20 Domain Are Critical for Glycogen Transport to Lysosomes in Liver-Stbd1 has an N-terminal hydrophobic transmembrane region, a putative leucine zipper, an identified AIM, and a CBM20 (13,14,18). To determine critical elements in Stbd1 for glycogen transport to lysosomes, we constructed a series of AAV vectors to express WT human Stbd1 and its mutants in dKO mice ( Fig. 2A). AAV transduction efficiencies for all constructs are comparable as real-time PCR results revealed similar viral copy numbers in livers of AAV-injected dKO mice (supplemental Fig. S2). Western blots demonstrated that the AAV vectors expressed GFP (negative control), Stbd1, and its mutants in the liver of injected dKO mice (Fig. 2B). Protein level of the ⌬N24 mutant was extremely low and only weakly detectable with longer exposure, possibly due to the instability of this mutant protein in liver cells (Fig. 2B).
Exogenous expression of human Stbd1 restored glycogen transport to lysosome in liver as indicated by the rebound of liver glycogen storage in the treated dKO mice to a similar level as GKO mice (Fig. 2C). This confirms that the much lower liver lysosomal glycogen storage in dKO mice was a direct result of Stbd1 knock-out.
To test whether the interaction between Stbd1 and GABARAPL1 is required for glycogen transport to lysosomes, we generated an AIM mutant, AIMm (W203A/V206A; Fig.  2A), that loses the GABARAPL1 binding ability (14). The AIMm mutant restored lysosomal glycogen storage in liver as potently as wild-type Stbd1 (Fig. 2C). This indicates that the Stbd1-GABARAPL1 interaction is not required for glycogen transport to lysosome in liver.
The ⌬N24 mutant failed to restore glycogen storage in liver (Fig. 2C). However, the extremely low expression level of this mutant prevents us from drawing a conclusion on the necessity of the N terminus in glycogen transport. It seems that this N-terminal transmembrane region is important for the stability of the protein in liver, because the same construct could express the protein in amounts comparable with other constructs in 293T cells (data not shown).
Lastly, we constructed a mutant (⌬26 -259) that contains only the N-terminal transmembrane region and the C-terminal CBM20 domain linked by an HA tag ( Fig. 2A). Surprisingly, this mutant functioned as potently as the full-length Stbd1 to restore lysosomal glycogen storage in the liver of dKO mice (Fig. 2C). This demonstrated that the intervening region (amino acids 26 -259) is dispensable for glycogen transport to lysosomes in liver.
Stbd1 Knock-out Does Not Block Autophagy Activation in Mouse Liver-Autophagy has been shown to be involved in glycogen trafficking to lysosomes in fast-twitch skeletal muscle (12). To check whether the loss of Stbd1 affects autophagy activation in liver, we checked the autophagy marker LC3 in dKO and GKO mice under both fed and fasted conditions. As shown in Fig. 2D, LC3-II level in fasted liver was dramatically higher than that in fed liver of both GKO and dKO mice, indicating the induction of autophagy by fasting in both genotypes. In other words, the activation of autophagy was not blocked by the Stbd1 knock-out. In addition, there was no obvious difference in expression of other autophagy-related genes such as P62, Beclin 1, and Atg8L1 in fasted livers between the dKO and GKO mice (supplemental Fig. S1). On the other hand, fasting-induced activation of autophagy in liver was not accompanied by ACCELERATED COMMUNICATION: Role of Stbd1 in Glycogen Transport AUGUST 5, 2016 • VOLUME 291 • NUMBER 32 an obvious change in the level of Stbd1 protein in GKO mice (Fig. 2D). Our results suggest that loss of Stbd1 does not affect conventional autophagy in liver.
Taken together, our data demonstrate that Stbd1 plays a dominant role in transporting glycogen to lysosomes in liver, and that the N-terminal transmembrane region and the C-terminal CBM20 domain are critical in this process.

Discussion
Cellular glycogen is mostly degraded in the cytoplasm, but a small portion is transported into the lysosomes and hydrolyzed by GAA (1-3). Understanding the mechanism of how glycogen is transported into lysosomes is a critical step for the exploration of new therapeutic targets for Pompe disease considering the significant limitations of enzyme replacement therapy (7)(8)(9). It has been demonstrated that autophagy plays a partial role in the transport of glycogen to lysosomes in skeletal muscle of GAA-KO mice (12), which suggests that other route(s) also exist for this process.
Recent studies led to a proposition that Stbd1 functions as a cargo receptor for delivery of glycogen to lysosomes in a specific autophagic pathway (termed "glycophagy"), and the interactions between Stbd1 and GABARAPL1 might be critical in this process (13,14). This provoked a great expectation that Stbd1 might be a new therapeutic target for Pompe disease. Contradictorily, our previous study demonstrated that significant knockdown of Stbd1 in skeletal muscle of GAA knock-out mice did not reduce lysosomal glycogen accumulation, suggesting that Stbd1 may not be involved in lysosomal glycogen trafficking at least in this tissue (16). In an effort to clarify this controversy, we generated Stbd1/GAA double knock-out mice. These mice serve as an ideal system for investigating the role of Stbd1 in transporting glycogen into lysosomes in various tissues because glycogen accumulation due to the lack of GAA is a reliable and quantifiable indicator of glycogen transport into lysosomes.
In agreement with our previous study (16), we found that loss of Stbd1 in dKO mice did not affect lysosomal glycogen accumulation in cardiac and skeletal muscles (Fig. 1A), indicating that Stbd1 may not be involved in glycogen transport to lysosomes in these tissues. However, strikingly, we found that lysosomal glycogen storage in the liver of dKO mice was 73% lower than that of the GKO mice, suggesting that Stbd1 is a major but not the sole mediator for glycogen transport into lysosomes.
Similar to our findings of Stbd1's function in liver, Raben et al. (12) demonstrated that autophagy plays a critical role in muscle in transporting glycogen to lysosomes. They reported that suppression of autophagy reduced lysosomal glycogen content by 50 -60% in skeletal muscles of GAA-KO mice. These results demonstrate that different tissues could have distinct major pathways for glycogen transport to lysosomes, and in the same tissue multiple routes exist. Because liver and muscle are very different in terms of the regulation of glycogen metabolism and the utilization of glycogen storage (3), it is not surprising that different pathways play major roles in transporting glycogen to lysosomes in the two tissues. Although our data clearly demonstrated that Stbd1 plays a dominant role in glyco-gen transport to lysosomes in liver, we cannot exclude the possibility that it may play a minor role in muscle.
Stbd1 knock-out did not block the activation of autophagy in liver, as a similar elevation of LC3-II abundance was observed in the GKO and dKO mice under fasting conditions (Fig. 2D), and had no effect on the expression of autophagy-related proteins P62, Beclin1, and GABARAPL1 (supplemental Fig. S1). On the other hand, fasting-induced activation of autophagy in GKO liver was not accompanied by apparent change in the level of Stbd1 protein (Fig. 2D). These results indicate that the reduced lysosomal glycogen accumulation in dKO mouse liver was not due to blockage of the conventional autophagy. Whether the conventional autophagy contributed to the transport of the small amount of lysosomal glycogen in the dKO liver, in the absence of Stbd1, remained undetermined.
The Stbd1-GABARAPL1 interaction is not required for Stbd1-mediated glycogen transport to lysosome because the Stbd1 mutant with the GABARAPL1 binding motif mutation (AIMm) was able to replenish the lysosomal glycogen content in the liver of dKO mice (Fig. 2C). Furthermore, the mutant missing the entire intervening region (⌬26 -259) also successfully restored the liver lysosomal glycogen content in dKO mice (Fig. 2C), demonstrating that only the N-terminal hydrophobic transmembrane region and the C-terminal CBM20 are critical for carrying out this function. Our data support the notion that Stbd1 may function as an anchor to bring glycogen to an intracellular membrane (13). GABARAPL1 is implicated in several different cellular processes including autophagy and is the most highly expressed gene in the central nervous system among the GABARAP family members (19 -21). This Stbd1-GABARAPL1 interaction may play important roles in other tissues, especially the central nervous system. This is the first study to demonstrate the function of Stbd1 in mice. Our results demonstrate that Stbd1 dominates glycogen transport to lysosomes in liver, but not in muscles. It would be interesting to further investigate the mechanisms by which Stbd1 participates in this process. As Stbd1 is also highly expressed in skeletal muscle and heart, it would be interesting to uncover the function of this protein in these tissues. Our study suggests that the mechanism of glycogen transport to lysosomes in liver is distinct from that in muscles, and thus different treatment schemes might be developed for different target tissues in Pompe disease.
Construction and Packaging of AAV Vectors for Expression of Wild-type and Mutant Stbd1-The scAAV-miR26a-eGFP vector was a gift from Joshua Mendell (Addgene plasmid 21894) (23). The pri-miR26a sequence was removed to generate the AAV-GFP vector, which served as backbone and negative control vector. Human Stbd1 cDNA and its mutants were cloned into the AAV-GFP vector to replace GFP. The Stbd1 mutants were generated as described by Jiang et al. (13,14) unless described otherwise. All constructs were verified by DNA sequencing. The resulting AAV vectors were packaged as AAV2/9 in HEK 293T cells using standard phosphate-mediated transfection and purified using iodixanol gradient ultracentrifugation (24 -26).
Animal Injection and Tissue Collection-AAV vectors were intravenously (tail vein) injected into 2-month-old male dKO mice at a dose of 5 ϫ 10 11 viral genomes per mouse. Tissues were collected 30 days later following overnight fasting. All animal procedures were done in accordance with Duke University Institutional Animal Care and Use Committee-approved guidelines.
Tissue Glycogen Content Measurement and PAS Staining-Tissue glycogen content was quantified using an enzymatic method as described previously (27). PAS staining of glycogen was performed on formalin-fixed tissues at the Research Histology Laboratory at Duke University.
Statistical Analysis-Values in the graphs are presented as mean Ϯ S.D. Statistical significance between two groups was determined by unpaired, equal valence, two-tailed t test.
Author Contributions-T. S. and H. Y. conducted most of the experiments, analyzed the results, and wrote most of the paper. C. Y. managed the mouse colonies. P. S. K. and B. S. conceived the idea for the work. All authors reviewed the results and approved the final version of the manuscript.