Lysosomal glucose sensing and glycophagy in metabolism

Abstract

Lysosomes are cellular organelles that function to catabolize both extra- and intracellular cargo, act as a platform for nutrient sensing, and represent a core signaling node integrating bioenergetic cues to changes in cellular metabolism. Although lysosomal amino acid and lipid sensing in metabolism has been well characterized, lysosomal glucose sensing and the role of lysosomes in glucose metabolism is unrefined. This review will highlight the role of the lysosome in glucose metabolism with a focus on lysosomal glucose and glycogen sensing, glycophagy, and lysosomal glucose transport and how these processes impact autophagy and energy metabolism. Additionally, the role of lysosomal glucose metabolism in genetic and metabolic diseases will be briefly discussed.

Lysosomes in cellular metabolism

Lysosomes are catabolic organelles containing glycosidases, lipases, nucleases, phosphatases, phospholipases, proteases, and sulfatases within their acidic lumen that digest both extracellular and intracellular macromolecules received by endocytosis or autophagocytosis[1]. The resulting metabolites including amino acids, glucose, lipids, sterols, and micronutrients liberated through digestion are recycled and used for energy metabolism, proliferation, or differentiation[2], while free fatty acids are effluxed outside of the cell[3]. A single phospholipid bilayer encompasses the lysosomal lumen which prevents the numerous digestive enzymes from damaging surrounding healthy organelles and forms a barrier for digested metabolites and micronutrients that must be selectively transported outside the lumen. Imbedded within the lysosomal phospholipid membrane are numerous proteins involved in the selective import and/or export of molecules, acidification of the lumen, and lysosomal fusion with other organelles. Given its role in generating nutrients, it is not surprising that the lysosome has an essential role in nutrient sensing, preserving cellular homeostasis, and survival[4]. Although lysosomal amino acid[2,5–9] and lipid[10,11] sensing and metabolism is well described, there is much less work characterizing lysosomal glucose and glycogen sensing and metabolism. Recent developments in this emerging field [12–20] have provided more insight into how lysosomes sense glucose availability[13], glycogen autophagy (i.e. glycophagy)[12,15–18], lysosomal glucose transport[14,20], and the metabolic fate of glucose liberated via glycophagy[19]. The goal of this review is to highlight these new developments to provide an up-to-date review and interpretation of the current knowledge base regarding lysosomal glucose metabolism. Furthermore, the roles for lysosomal glucose handling in health and disease will be covered.

Mechanisms of Lysosomal Glucose and Glycogen Sensing

Two key nutrient sensing networks relay metabolic information to the lysosome. The first is the mechanistic target of rapamycin complex 1 (mTORC1) network, which is recruited to the lysosomal surface prior to its activation and is involved in cell growth and inhibiting autophagy[21]. The second network involves the adenosine monophosphate-activated protein kinase (AMPK) network, which senses the cellular energy state and activates autophagy during energy deficit conditions[22]. These signaling nodes are a significant part of the lysosome nutrient sensing complex and coordinate energy metabolism and autophagy in response to cellular needs.

mTORC1 inhibits autophagy by phosphorylating and inactivating the autophagy-initiating UNC-5 like autophagy activating kinase (ULK1) at Ser757, autophagy related gene 13 (ATG13)[23–26], autophagy/beclin 1 regulator 1 (AMBRA1)[27], and autophagy related gene 14 long isoform (ATG14L)[28]. mTORC1 regulates autophagy transcriptionally by phosphorylating transcription factor EB (TFEB) at Ser211, which retains TFEB in the cytosol and prevents it from entering the nucleus where it drives the expression of autophagy and lysosomal genes[29]. mTORC1, localized at the lysosomal surface, senses intracellular glucose availability [13,30–34] through numerous mechanisms to mediate autophagy and lysosomal glucose sensing (Figure 1). For example, in mouse embryonic fibroblasts Rag GTPases were shown to signal glucose availability to mTORC1 [30] (Figure1A). More recent discoveries revealed that glycolytic enzymes or metabolites signal glucose availability through mTORC1. The enzyme 6-phosphofructo-2- kinase/fructose-2,6-biphospatase (PFKFB3), which is a known glycolysis regulator, promotes mTORC1 lysosomal localization and activation in cancer cells[31] (Figure1B). In cardiomyocyte and noncardiomyocyte cells the glycolytic intermediate glucose-6-phosphate (G6P) prevents the glycolytic enzyme hexokinase 2, the predominant isoform in the heart, from binding to and inhibiting mTORC1[32] (Figure1C). Using HEK293T cells lacking AMPK the glycolytic intermediate dihydroxyacetone phosphate directly activated mTORC1 to signal glucose availability to the lysosome[13] (Figure1D). Further downstream, the enzyme glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was shown to function as a glucose sensor in MEF cells by interacting with the enzyme Ras homolog enriched in brain (RHEB) and preventing it from binding and activating mTORC1 on the lysosomal surface. Glyceraldehyde-3-phosphate (G3P), the glycolytic intermediate that binds GAPDH, destabilizes this GAPDH-RHEB interaction, suggesting G3P mediates the effects of low or high glucose levels on mTORC1 activation[33] (Figure1E). Each of these studies showed a specific mechanism in specific cell types ranging from MEFs, HEK293T cells, U2OS cells, neonatal rat ventricular myocytes, and rat heart tissue suggesting that cell type likely impacts lysosomal glucose sensing through mTORC1.

Figure 1. Lysosomal Glucose Sensing Via mTORC1.

A) High glucose availability activates Rag GTPases which promote mTORC1 activity and suppress autophagy. B) The 6-Phosphofructo-2-Kinase/Fructose-2,6-biphosphatase 3 (PFKFB3) protein promotes mammalian or mechanistic target of rapamycin (mTORC1) activity which suppresses autophagy. C) The glycolytic metabolite glucose-6-phosphate prevents the glycolytic enzyme hexokinase (HK) from binding and inhibiting mTORC1. D) The glycolytic metabolite dihydroxyacetone phosphate activates mTORC1. E) The glycolytic metabolite glyceraldehyde 3-phosphate binds to the glycolytic enzyme glyceraldehyde 3-phosphate dehydrogenase (GAPDH) to prevent GAPDH from binding and inactivating the mTORC1 activator Ras homolog enriched in brain (RHEB). Figure created with biorender.com.

AMPK antagonizes mTORC1 to regulate autophagy at the protein level by phosphorylating tuberous sclerosis complex 2 TSC2[35] and Raptor[36] to inhibit mTORC1 in addition to phosphorylating ULK1 at Ser317 and Ser777, phosphorylating the pro-autophagy protein Beclin 1, phosphorylating and inhibiting the autophagy inhibiting protein phosphatidylinositol 3-kinase catalytic subunit type 3 (VPS34)[26,37], and by forming a lysosomal complex consisting of v-ATPase, calcium transporter transient receptor potential channel subfamily V (TRPV), Ragulator, Axin, Liver Kinase B1 (LKB1), and AMPK leading to autophagy induction[38–40]. AMPK regulates lysosome biogenesis transcriptionally by phosphorylating the transcription factor FOXO3 at six different sites to control its transcriptional activity [41]. Unlike mTORC1 which phosphorylates TFEB at Ser211 to retain it in the cytosol, AMPK phosphorylates TFEB at Ser466, 467, and 469 to increase TFEB transcriptional activity which promotes autophagy related gene expression[42]. There are several distinct mechanisms by which AMPK signals glucose availability to the lysosome (Figure 2). With high glucose availability the glycolytic intermediate fructose-1,6-bisphosphate binds aldolase, which blocks aldolase from interacting with TRPV and v-ATPase thereby disrupting the TRPV, Ragulator, Axin, LKB1, and AMPK lysosomal protein complex[38] to inhibit autophagy (Figure 2A). Moreover, AMPK can signal glucose availability to the lysosome through non-direct mechanisms by phosphorylating TSC2[35] and Raptor[36] proteins which inactivate RHEB to prevent mTORC1 activation[43] in MEF or HEK293T cells (Figure 2B). Additionally, AMPK activation phosphorylates GAPDH, stimulating GAPDH nuclear translocation where it displaces deleted in breast cancer 1 (DBC1) binding to Sirtuin 1 (SIRT1) alleviating DBC1-mediated inhibition and allowing SIRT1 to promote autophagy and mitochondrial metabolism in MEF cells[44] (Figure 2C). Highly branched cytosolic glycogen molecules bind to the β subunit of AMPK to inactivate it which may represent a mechanism by which AMPK signals glycogen availability to the lysosome[45]. It is not clear if lysosomal glycogen can signal through an inside-out mechanism to mTORC1 or AMPK, or if only cytosolic glycogen is involved in this process. Collectively, these different glucose sensing mechanisms likely co-exist to better fine tune autophagy and energy metabolism through mTORC1 and AMPK signaling. An important challenge moving forward is to identify cell specific regulation of lysosomal glucose sensing through these networks.

Figure 2. Lysosomal Glucose Sensing Via AMPK.

1) The glycolytic intermediate fructose-1,6-phosphate prevents the glycolytic enzyme aldolase (ALDO) from binding to v-ATPase which promotes adenosine monophosphate activated protein kinase (AMPK) activity and autophagy. 2) AMPK phosphorylates the tuberous sclerosis 2 (TSC2) and Raptor proteins which subsequently inactivate the mTORC1 activator RHEB. 3) AMPK phosphorylates the glycolytic enzyme GAPDH which promotes GAPDH nuclear translocation where it alleviates deleted in breast cancer 1 (DBC1) inhibition of Sirtuin 1 (SIRT1) which promotes autophagy and mitochondrial metabolism. Figure created with biorender.com.

Mechanisms of Glycophagy, Its Regulation, and Role in Pompe Disease

Glycogen is catabolized to glucose in the cytosol and lysosomes. In the cytosol, glycogenolysis involves the coordinated activity of glycogen debranching enzyme and glycogen phosphorylase, where glycogen debranching enzyme removes α−1–6 glucose linkages from glycogen while glycogen phosphorylase removes α−1–4 glucose linkages. Glucose-1 phosphate generated from this pathway is converted to glucose-6-phosphate via phosphoglucomutase and shuttled into glycolysis or the pentose phosphate pathway[46]. The other mechanism of glycogen catabolism is glycogen autophagy (i.e. glycophagy). Glycophagy is a selective form of autophagy where glycogen is targeted to the lysosome and degraded by the enzyme alpha-acid glucosidase (GAA). GAA degrades both α−1–4 and α−1–6 branch points of glycogen to free glucose, but with higher affinity for α−1–4 branch points[47].

Proteins involved in glycophagy from the NCBI[48] and the STRING database[49] are depicted in Figure 3. GAA, STBD1, and GABARAPL1 had the same interaction flow in both humans and mice. We next investigated published reports to confirm these findings. Peter Roach’s group first demonstrated STBD1 and GABARAPL1 may participate in glycophagy. Overexpressing human STBD1 in COSM9 cells revealed STBD1 co-localized with GABARAPL1 (autophagy protein)[50]. In this initial publication they also found STBD1 co-localized with LAMP1 (endosomal/lysosome marker), findings later corrected as the authors admitted the wrong antibody for LAMP1 was used, and in follow-up experiments STBD1 did not co-localize with LAMP2 (lysosome marker)[51]. Nevertheless, the authors concluded that since STBD1 is a membrane bound protein[52] they hypothesized it anchors glycogen to membranes to affect its localization and intracellular trafficking to lysosomes[50]. A subsequent study identified the ATG8 interacting motif in STBD1 required for its interaction with GABARAPL1[53]. These findings appear to translate to hepatocytes, as double knockout of both GAA and STBD1 in the liver suppresses whole cell glycogen levels, although lysosomal glycogen was not measured which makes it difficult to conclude that STBD1 is directly involved in glycophagy [54]. A proteomics analysis of lysosomes isolated from mouse liver revealed STBD1 co-localized with lysosomes[14]. However, these data do not rule out the possibility that other STBD1-independent mechanisms are involved. Indeed, there are inconsistencies in the literature with some[53], but not all data[54], suggesting that STBD1 interacts with the autophagy protein, GABARAPL1. In HEK293T cells STBD1 does not co-localize with lysosomes, suggesting glycogen is transported to lysosomes via another STBD1-independent mechanism[20]. In MEFs, TSC deficiency increases glycogen content, in part, due to defective autophagy, suggesting macroautophagy is involved[55]. In skeletal muscle, there is evidence that macroautophagy is the major route for glycogen transport to lysosomes. For example, overexpression of STBD1 in C2C12 skeletal muscle cells does not alter the co-localization of LAMP1 and glycogen, suggesting no alteration in lysosomal glycogen content[56]. In mice, double knockout of GAA and STBD1 in skeletal muscle does not alter whole cell glycogen levels [57], although it is not clear if lysosomal glycogen content was altered. Since cytoplasmic glycogen content is disturbed in skeletal muscle with GAA knockout[15,58], measurements of whole cell glycogen content could mask the effects of double knockout of STBD1 and GAA on lysosomal glycogen. For example, skeletal muscle from a mouse model of Pompe disease display enhanced glycogen synthesis and reduced glycogen phosphorylase activity in the absence of AMP[58]. Consistent with this, a proteomics analysis on skeletal muscle from mice or humans with Pompe disease revealed increased abundance of proteins involved in glycogen synthesis, suggesting lysosomal glycogen accumulation increases cytoplasmic glycogen[15]. Therefore, based on the available data it is not clear if STBD1 is the major route of glycogen transport to lysosomes in skeletal muscle. There is stronger evidence that macroautophagy is the major route of glycogen transport to lysosomes in skeletal muscle. For example, double knockout of GAA and ATG7[59], a key autophagy gene, reduces whole cell and lysosomal glycogen levels in skeletal muscle, an effect that was enhanced with ERT therapy. In support of this idea, patients with Danon disease have defective macroautophagy as they lack the lysosomal enzyme LAMP2b, which is required for fusion of autophagosomes with lysosomes. These patients have a massive accumulation of glycogen (as well as numerous other cargo) within skeletal muscle despite normal levels of GAA[60]. This observation indicates that fusion of autophagosomes containing glycogen with lysosomes containing GAA may be a major route of glycogen transport to lysosomes in skeletal muscle. These patients have only a mild liver phenotype suggesting that macroautophagy may not be the primary means of glycogen transport to lysosomes in hepatocytes or other compensatory mechanisms may exist[61]. Consistent with these findings, chloroquine treatment in D. melanogaster (fruit fly) blocked autophagosome-lysosome fusion and increased autophagosome glycogen content in skeletal muscle, but not the larvae fat body (which in this model is analogous to the vertebrate liver and adipose tissue)[62]. Moreover, during fasting chloroquine treatment significantly suppressed glycogen loss in skeletal muscle. Collectively, these data provide strong evidence that macroautophagy is the primary route of glycogen delivery to lysosomes in skeletal muscle, but do not rule out a role for STBD1 which may be important for glycogen macroautophagy. Nevertheless, future work should aim to better understand how glycogen gets shuttled to the lysosome for glycophagy in different tissues and under different metabolic conditions (i.e. fasting vs. feeding). It will also be important to determine if glycogen transport to lysosomes can occur through microautophagy (glycomicroautophagy) or chaperone mediated autophagy.

Figure 3. Protein-protein interaction (PPI) network of suspected proteins involved in glycophagy using STRING database.

The upper left network is the PPI network in humans, while the upper right image is for mice. Figure created with biorender.com and STRING database.

Most work examining glycophagy in metabolism has been in the context of Pompe disease (glycogen storage disease type II). The excessive lysosomal glycogen levels impair lysosomal nutrient sensing in response to starvation and re-feeding in skeletal muscle by reducing mTORC1 activity and hyperactivating LKB1, AMPK, TSC1/2 complex, and v-ATPase[63,64]. The mechanism underlying these effects are not completely clear, but elevations in cytosolic calcium levels, and not alterations in lysosomal calcium handling, may have a role[65] as the L-type calcium channel blockers reduce calcium levels and alleviate some defects observed in Pompe disease[65].

Pompe disease also alters mitochondrial metabolism in skeletal muscle, but this effect may depend on the degree and length of loss. For instance, muscle biopsy samples from infants with Pompe disease revealed a reduction in respiratory chain enzyme activity, although mitochondrial content was not altered[66]. Consistent with this, in a mouse model of Pompe disease, distended mitochondria with aberrant cristae formation, abnormal mitochondrial dynamics, respiration, and calcium homeostasis were observed[65]. Moreover, in a three-dimensional tissue-engineered human skeletal muscle model of Pompe disease an RNA sequencing analysis revealed a transcriptional signature consistent with impaired oxidative metabolism and reduced ATP biosynthesis[66]. Work from our lab shows that a partial, short-term knockdown of glycophagy in muscle cells has largely the opposite effect on mitochondrial metabolism as that observed in Pompe disease. Not only did an RNA sequencing analysis reveal a metabolic signature consistent with increased mitochondrial metabolism, but functional assays revealed greater mitochondrial content, respiration rates, and fatty acid oxidation rates in C2C12 cells[67]. These differences could be due to the model system used, the duration/degree of knockout in each model, or both and warrants further investigation.

The dogma surrounding glycophagy is that it is a housekeeping process to degrade poorly branched glycogen molecules that are not ideal substrates for glycogen phosphorylase and debranching enzyme[67]. However, there is work showing glycophagy is dynamically regulated and likely has a bigger role in metabolism (Figure 5). The cyclic adenosine monophosphate (cAMP)/protein kinase A signaling node increases glycogen levels within lysosomes, whereas mTORC1 signaling reduces lysosomal glycogen levels in the liver[68,69]. The GAA gene has been identified as a TFEB direct target in the CLEAR network, suggesting GAA is upregulated during nutrient deficient conditions when autophagy is upregulated and mTORC1 is downregulated[70]. GAA is regulated through the Notch-1/Hes-1 network, a signaling node involved in numerous cell processes ranging from metabolism, inflammation, differentiation, proliferation, and apoptosis. The GAA gene contains an intronic repressor element that binds Hes-1, a downstream target in the Notch-1 signaling cascade while constitutive overexpression of Notch-1 and Hes-1 represses GAA gene abundance in hepatocytes[71]. Curiously, Hes-1 acts to increase GAA expression in fibroblasts[72], suggesting GAA regulation differs across tissues.

Figure 5. Regulation and Role of Glycophagy in Metabolism.

A) 1. In chicken granulosa cells follicle-stimulating hormone promotes glycophagy through PI3K / AKT signaling. 2. In hepatocytes mTORC1 activation suppresses glycophagy. 3–4. In hepatocytes an increase in cAMP/Protein Kinase A signaling either via glucagon (3) or norepinephrine (4) signaling promotes glycophagy through unknown mechanisms. 5. In hepatocytes and fibroblasts Notch signaling alters GAA transcription through unknown mechanisms. 6. In hepatocytes high glucose availability increases OCT / AKT1 / FOXO1 signaling which reduces GABARAPL1 transcription. B) The potential fate of glucose liberated via glycophagy. Figure created with biorender.com.

In various cell types a variety of stimuli alter markers of glycophagy. In chicken granulosa cells, follicle-stimulating hormone promoted glycophagy through PI3K/AKT signaling[17]. In hepatocytes, high-glucose availability increased O-GlcNAc transferase, AKT1, and FOXO1 signaling to reduce GABARAPL1 promoter activity[16]. In cervical cancer cells, ciclopirox olamine (anticancer agent) increased glycophagosome formation and STBD1 and GABARAPL1 co-localization, suggesting increased glycophagy[18].

Recent work from our lab shows skeletal muscle contraction increased GAA activity during contraction but 24 and 48 h post contraction GAA activity decreased[67]. Moreover, 24 h of glucose restriction reduced GAA activity but trehalose, Notch signaling, or cAMP signaling increased GAA activity, indicating the regulation of skeletal muscle GAA activity is dynamic. Further work is needed to clarify and precisely delineate how glycophagy is regulated.

Mechanisms of Lysosomal Glucose Transport and the Fate of Glucose Liberated Via Glycophagy

The lysosomal lumen is surrounded by a phospholipid bilayer, thus glucose liberated through glycophagy needs to be transported out of the lysosomal lumen. Evidence suggests the transport proteins used in this process may differ depending on cell type. A proteomics analysis of lysosomes from HEK293T cells (human kidney cells) identified that GLUT1 and GLUT3 co-localized with lysosomes, but only when mTORC1 was inhibited with the drug Torin 1[20]. GLUT8 and protein spinster homolog 1 (SPNS1) co-localized with lysosomes under full media, Torin 1, or starvation conditions, suggesting these transporters always reside on lysosomes, independent of media conditions[20]. GLUT8 has glucose transport activity[73] but global knockout of GLUT8 does not alter embryonic development, postnatal growth patterns, or glycemic control but does increase hippocampal cell proliferation[74], P-wave duration in the heart[74], and impairs energy metabolism in sperm[75]. In blastocysts[76], but not adipocytes[77], GLUT8 will translocate to the plasma membrane in response to insulin, suggesting GLUT8 functions differently in various cells or tissues. Given that GLUT8 deficient mice do not display a drastic phenotype in most tissues, other glucose transporters may be able to compensate. SPNS1 is a lysosomal enzyme possessing carbohydrate transport activity[78]. Knocking out SPNS1 prevents mTORC1 reactivation in response to nutrient surplus and leads to enlarged dysfunctional lysosomes containing massive amounts of carbohydrate[78], as indicated by PAS staining. It is not known if this is glycogen, free glucose, or glycosylated proteins as PAS staining is non-specific, but these data suggest that SPNS1 may be important for lysosomal carbohydrate metabolism. In liver from mice, a proteomics analysis on isolated lysosomes revealed that GLUT1, GLUT2, GLUT8, and SPNS1 proteins co-localized with lysosomes[14]. Whether all these glucose transporters co-localize with lysosomes under different conditions, such as feeding or exercise, has yet to be determined. It is also not clear if in other cell types these same or other glucose transporters co-localize with lysosomes to facilitate glucose transport of glucose liberated via glycophagy.

Glucose is metabolized through a variety of different cellular processes including glycolysis, glycogen synthesis, lipogenesis, the pentose phosphate pathway, or glycosylation (Figure 5), but the metabolic fate of glucose liberated via glycophagy is enigmatic and may differ between tissue types. Whether the glucose transported out of lysosomes remains a free glucose monomer or is selectively phosphorylated to glucose-6 phosphate, glucose-1 phosphate, or converted to UDP glucose or another metabolite remains undefined at this time. In the lungs of mice, there is evidence that glucose liberated via glycophagy is used for N-linked glycosylation in pulmonary fibrosis[19]. Work from our lab using skeletal muscle indicates glucose liberated via glycophagy is shuttled toward glycolysis as partial, short-term loss of glycophagy in skeletal muscle cells reduced glycolytic capacity[67]. In brown adipocytes from developing embryo’s, there is some evidence that glucose generated via glycophagy may be shuttled toward lipogenesis as inhibition of autophagy or glycogen synthase prevented the formation of lipid droplets[79]. In line with this work, our lab has shown that acute knockdown of GAA reduces triglyceride content in skeletal muscle[67], although case studies have shown that longer-term knockout of GAA in Pompe disease results in an increased amount of neutral lipid staining in skeletal muscle[80,81]. Thus, the tissue, extent, and duration of glycophagy loss likely has a major influence on intracellular lipid levels.

Lysosomal Glucose Metabolism in Various Metabolic Diseases

There is evidence that lysosomal glucose sensing and glycophagy are modulated in cancer. For example, a hallmark feature of liver cancer is disrupted lysosomal glucose sensing as AMPK activity is suppressed while mTORC1 signaling is elevated to restrict catabolism while promoting glycolysis, anabolism, and cancer growth[82–88]. HCC may also leverage glycophagy to support cell proliferation as indicated by an increase in GAA mRNA expression in mouse HCC tumors, compared to adjacent non-tumor tissue[89]. Consistent with this, previous studies have shown GAA is upregulated in various cancer cell lines during metabolic stress[90]. In humans with type 2 diabetes the GAA inhibitor acarbose, which inhibits intestinal and colon GAA, reduced colon cancer risk[91]. Moreover, genetically silencing GAA in the pancreas improved the anti-tumor effect of the pancreatic cancer drug gemcitabine[92]. Whole-body knockout of GAA reduced lung fibrosis burden induced by bleomycin[19], suggesting glycophagy is required for fibrosis progression, an initiating step in cancer induction.

Lysosomal glucose sensing through AMPK and mTORC1 is also disrupted in the setting of obesity and hyperglycemia in liver, skeletal muscle, β-cells, and the kidney. In liver and skeletal muscle, chronic AMPK inactivation[93] coupled with increased mTORC1 signaling[94,95] promotes insulin resistance[93,95,96]. This mechanism may occur through S6K overactivation, a downstream mTORC1 target, which will reduce insulin receptor substrate 1 and 2 (IRS 1 and 2) abundance when overactivated[97]. Similarly, studies in β-cells of mouse models of type 2 diabetes indicate mTORC1 is upregulated with hyperglycemia[98,99]. Genetically activating mTORC1 increases β-cell mass and function, while over time the impact of chronic mTORC1 activation is controversial with one study showing it promotes β-cell failure[98] while another study showed sustained improvements in β-cell function[100]. Nevertheless, chronic, prolonged mTORC1 activation reduces AMPK activity, which is important for stimulating autophagy and mitophagy, while lack of autophagy leads to β-cell dysfunction and apoptosis to promote type 2 diabetes[101]. Hyperglycemia also enhances mTORC1 signaling in the kidney to promote diabetic nephropathy, while genetically suppressing mTORC1 activity in mice increases kidney function and prevents diabetic nephropathy[102]. Consistent with elevated mTORC1 signaling, autophagy is dysregulated in the diabetic kidney[103,104], while activating AMPK in diabetic nephropathy improves renal function[105].

Glycophagy and lysosomal glycogen content may have a role in glucose tolerance and skeletal muscle glucose uptake. Whole body GAA knockout animals display enhanced whole-body glucose tolerance, an effect associated with elevated skeletal muscle GLUT1 and GLUT4 levels[106]. An insulin tolerance test did not reveal changes in whole body insulin sensitivity. In skeletal muscle, insulin stimulation did not alter GLUT1 and GLUT4 protein content. However, basal protein abundance of these enzymes was similar in quantity to the protein abundance of insulin stimulated wild type muscle, suggesting that GAA knockout promotes maximal GLUT1 and GLUT4 abundance that cannot increase further in response to insulin. These data suggest that inhibiting glycophagy may have metabolic health benefits in some contexts. Consistent with this, acarbose and miglitol, two drugs that inhibit GAA and glycophagy, when taken orally improve glycemic control and insulin sensitivity by preventing the degradation and absorption of complex carbohydrates in the intestine[107,108]. Work from our lab has shown that treating skeletal muscle cells with acarbose for 24 h or genetically reducing GAA activity increases insulin action and mitochondrial metabolism, indicating acute inhibition of glycophagy specifically in skeletal muscle may have health benefits[67].

The β-cells of the pancreas also accumulate glycogen in response to hyperglycemia and this is associated with impaired glucose metabolism, impaired insulin secretion, and apoptosis[109]. In Pompe disease glycogen accumulation has been reported in pancreatic islets, although it is not clear if this occurs specifically in β-cells[110]. High levels of autophagy occur in β-cells in pre-diabetes, but during the progression to type 2 diabetes autophagy decreases, and this contributes to β-cell failure. Thus, it is possible that reduced glycophagy could be one mechanism of β-cell dysfunction during progression to type 2 diabetes.

In response to hyperglycemia the kidneys accumulate glycogen in the renal tubules[111]. It is not clear if this is associated with dysregulated glycophagy or if other enzymes of glycogen metabolism are involved. In late onset Pompe disease there are reports of renal artery fibromuscular dysplasia[112] and acute renal failure[113] thus lack of kidney glycophagy can be pathological. Impaired autophagy occurs in kidney disease[103,104], and given glycophagy is a part of autophagy suggests impaired glycophagy may contribute.

Conclusions and Perspectives

The role of the lysosome in glucose metabolism under healthy and pathological states is gaining momentum, and this review highlights important advances in the field as well as future directions and unanswered questions. Lysosomes can sense glucose availability through modulations in mTORC1 signaling via Rag GTPases, PFKFB3, hexokinase 2, dihydroxyacetone phosphate, and GAPDH. Lysosomes sense glucose availability through AMPK signaling via the glycolytic enzyme’s aldolase and GAPDH. AMPK also indirectly signals glucose availability by inhibiting mTORC1 signaling. Evidence suggests the major route of glycogen transport to lysosomes in the liver includes the cargo binding protein STBD1. However, in skeletal muscle the role of STBD1 is not clear, and evidence suggests macroautophagy is the major route of glycogen transport to lysosomes. Glucose liberated via glycophagy may be actively transported out of lysosomes in kidney cells through GLUT1, 3, or 8 transporters and/or SPNS1 carbohydrate transporters, while in the liver there is evidence that GLUT1, 2, or 8 transporters and/or SPNS1 are involved. The fate of glucose liberated through glycophagy varies by cell type and in skeletal muscle may be glycolysis, in brown adipocytes is lipogenesis, and in the lungs is N-linked glycosylation. A dysregulation of lysosomal glucose sensing occurs in metabolic diseases where mTORC1 is overactive while AMPK signaling is often attenuated. It is not clear if glycophagy becomes dysregulated to contribute to the pathogenesis of metabolic disease, although there is evidence it may be upregulated in cancer. Although some progress has been made, much needs to be learned to deepen our knowledge of lysosomal glucose metabolism in health and disease (See Outstanding Questions Box).

Outstanding Questions.

Do lysosomes sense glucose and glycogen availability through an inside out mechanism? Lysosomes sense cytosolic glucose and glycogen availability, but glucose and glycogen are also within lysosomes, and it is unknown if lysosomes sense their availability.

What are the mechanisms of glycogen transport to lysosomes? In skeletal muscle glycophagy appears to be a part of macroautophagy. In the liver the cargo binding protein STBD1 may be the primary route of glycogen transport to lysosomes. However, the precise steps by which glycogen gets targeted to the lysosome in various cell types is not well defined.

How is glycophagy regulated? In skeletal muscle contraction, cAMP signaling, or Notch signaling regulates glycophagy, while in the liver Notch signaling, mTORC1 signaling, and cAMP signaling regulates glycophagy. It is not clear how glycophagy is regulated in other cell types or if glycophagy is a housekeeping process that does not dynamically change.

What is the metabolic fate of glucose liberated via glycophagy? Glucose liberated via glycophagy is a free glucose monomer, but it is not clear if this glucose is phosphorylated upon exiting the lysosome so it can be shuttled toward glycolysis, pentose phosphate pathway, lipogenesis, glycosylation, or other metabolic pathways.

What controls lysosomal glucose import or export? Lysosomes contain numerous glucose transporters, but it is not clear how glucose uptake or efflux outside the lysosome is regulated.

Does glycophagy have a role in chronic disease? Autophagy is typically dysregulated in many chronic diseases, but it is unknown if a dysregulation in glycophagy occurs to promote chronic disease development.

Figure 4. Glycophagy in Liver and Skeletal Muscle.

A) Evidence suggests that in the liver a major mechanism of glycogen transport to lysosomes occurs via the cargo binding protein starch binding domain containing protein 1 (STBD1) (Steps 1–2). STBD1 may interact with GABARAPL1 to facilitate glycogen delivery and translocation into the lysosome, although in the liver evidence to support this is mixed so a question mark is present (Step 3). After being translocated into the lysosome, the enzyme α-acid glucosidase (GAA) can catalyze the digestion of glycogen into glucose monomers (Step 3). The glucose monomers generated by glycophagy are transported out of lysosomes through glucose transporter solute carrier family 2 facilitated glucose transporter member 1, 2, or 8 (GLUT 1, 2, or 8) and/or the SPNS1 sugar transporter, which have been identified through proteomics analysis of isolated lysosomes. B) Evidence in skeletal muscle suggests that macroautophagy is the major route of glycogen delivery to lysosomes, while the role of STBD1 and GABARAPL1 is not clear. Glycogen molecules, along with other cytosolic cargo, are initially engulfed by autophagosomes (Steps 1–2), and then after fusing with lysosomes (Step 3) the GAA enzyme will have access to glycogen to degrade it into glucose molecules (Step 4). Glucose liberated via glycophagy is shuttled out of lysosomes via SPIN1 and/or other yet to be identified glucose transporters and subsequently utilized in metabolism (Step 5). Figure created with biorender.com.

Highlights.

Lysosomes sense glucose availability through modulations in mTORC1 signaling via Rag GTPases, PFKFB3, HK2, dihydroxyacetone phosphate, and GAPDH.

Lysosomes sense glucose availability through AMPK signaling via the enzyme’s aldolase and GAPDH.

The mechanisms of glycogen transport to lysosomes are not clear. In the liver there is evidence of an STBD1-dependent mechanism of glycogen transport to lysosomes, while in skeletal muscle there is some evidence of an STBD1-independent mechanism through macroautophagy.

Lysosomes isolated from mouse liver contain GLUT1, 2, and 8 in addition to the SPNS1 glucose transporter, while lysosomes isolated from kidney cells contain GLUT8 and SPNS1, in addition to GLUT1 and 3 when mTORC1 is inhibited.

Dysregulated lysosomal glucose sensing and glycophagy occurs in metabolic diseases such as cancer, insulin resistance, and hyperglycemia.

Glossary

Adenosine monophosphate-activated protein kinase

AMPK is a regulatory kinase acting in cellular energy homeostasis that modulates autophagy, glucose, and lipid metabolism in response to energetic demands

Autophagy

catabolic, self-degradative process where the cargo (cytosolic macronutrients and dysfunctional organelles) is sequestered within double-membrane vesicles called autophagosomes which fuse with lysosomes for degradation into micronutrients that can be recycled and used in cellular metabolism. Also referred to as macroautophagy

Chaperone-mediated autophagy

soluble proteins are translocated across the lysosomal membrane for degradation, a process that does not require membrane rearrangement like macroautophagy and microautophagy

Danon Disease

rare genetic condition with X-linked dominant inheritance pattern and more notorious clinical characteristics in males. Considered a lysosomal storage disorder associated with heart and skeletal muscle weakness, and intellectual disability

Glycogen

stored form of excess intracellular glucose which consists of multiple branches of glucose monomers

Glycophagy

autophagic degradation of glycogen by the lysosomal enzyme alpha-acid glucosidase

Mechanistic target of rapamycin complex 1

mTORC1 is a kinase that regulates cell growth and metabolism through signals from nutrients, growth factors, stress, and cellular energy levels

Microautophagy

direct engulfment of cytoplasmic cargo through membrane invagination and protrusion of the lysosome membrane for subsequent degradation

Pompe Disease

genetic disease due to a lack of the alpha-acid glucosidase protein which subsequently results in massive amounts of glycogen accumulation, particularly in cardiac muscle, skeletal muscle, and the nervous system