Present address The Rockefeller University, New York, NY, USA
Determinants of the nucleocytoplasmic shuttling of muscle glycogen synthase
Article first published online: 14 JUN 2005
Volume 272, Issue 12, pages 3197–3213, June 2005
How to Cite
Cid, E., Cifuentes, D., Baqué, S., Ferrer, J. C. and Guinovart, J. J. (2005), Determinants of the nucleocytoplasmic shuttling of muscle glycogen synthase. FEBS Journal, 272: 3197–3213. doi: 10.1111/j.1742-4658.2005.04738.x
- Issue published online: 14 JUN 2005
- Article first published online: 14 JUN 2005
- (Received 18 November 2004, revised 24 March 2005, accepted 25 April 2005)
- green fluorescent protein;
- muscle glycogen synthase;
- nucleocytoplasmic translocation;
- glucose 6-phosphate
- Top of page
- Experimental procedures
- Supplementary material
Muscle glycogen synthase (MGS) presents a nuclear speckled pattern in primary cultured human muscle and in 3T3-L1 cells deprived of glucose and with depleted glycogen reserves. Nuclear accumulation of the enzyme correlates inversely with cellular glycogen content. Although the glucose-induced export of MGS from the nucleus to the cytoplasm is blocked by leptomycin B, and therefore mediated by CRM1, no nuclear export signal was identified in the sequence of the protein. Deletion analysis shows that the region comprising amino acids 555–633 of human MGS, which encompasses an Arg-rich cluster involved in the allosteric activation of the enzyme by Glc6P, is crucial for its nuclear concentration and aggregation. Mutation of these Arg residues, which desensitizes the enzyme towards Glc6P, interferes with its nuclear accumulation. In contrast, the known phosphorylation sites of MGS that regulate its activity are not involved in the control of its subcellular distribution. Nuclear human MGS colocalizes with the promyelocytic leukaemia oncoprotein and p80-coilin, a marker of Cajal bodies. The subnuclear distribution of MGS is altered by incubation with transcription inhibitors. These observations suggest that, in addition to its metabolic function, MGS may participate in nuclear processes.
Dulbecco's modified Eagle's medium
fetal bovine serum
liver glycogen synthase
muscle glycogen synthase
human muscle glycogen synthase
rat liver glycogen synthase
green fluorescent protein
nuclear export signal
nuclear localization signal
promyelocytic leukaemia oncoprotein
Glycogen synthase (GS; EC 188.8.131.52) catalyzes the addition of α-1,4-linked glucose units to a growing glycogen molecule, a key step in the biosynthesis of the polysaccharide. In mammals, two GS isoforms have been described: the liver form, which is specific to this organ, and the muscle form, which, apart from muscle, is expressed in several tissues, including adipose tissue, kidney, spleen and the nervous system . GS is highly regulated by both covalent phosphorylation and allosteric effectors. In response to hormonal signals that differ depending on the tissue, muscle and liver GS (MGS and LGS) are phosphorylated at several serine residues, a process that leads to the inactivation of the two isoforms . Glc6P is an allosteric activator, which also favours the covalent activation of GS through its dephosphorylation . An arginine-rich cluster in the C-terminal region of the protein is crucial for the conformational switch triggered by Glc6P, and also by dephosphorylation, which is involved in the regulation of the catalytic activity of yeast  and rabbit MGS .
Glycogen and GS do not show uniform intracellular distribution. LGS accumulates near the plasma membrane when cultured hepatocytes are incubated with glucose [6,7]. Consequently, the deposits of the polysaccharide grow from the periphery towards the interior of the cells . In hepatocytes, glycogen synthesis is not homogeneous within the cell, and the spatial distribution of LGS is regulated. Regarding muscle, the protein resulting from fusion of human muscle glycogen synthase (HsMGS) with green fluorescent protein (GFP) shows a regulated nucleocytoplasmic distribution [9,10]. When cells expressing this chimeric protein were incubated in medium containing glucose, the GFP–HsMGS fusion was found throughout the cytoplasm and associated with glycogen. In contrast, when the cells were maintained in a glucose-free medium, the chimeric protein was concentrated in the nucleus and formed spherical structures. The behaviour of the fusion protein was independent of the cell type and hormone supply. Other studies have reported the presence of GS activity and glycogen in the nucleus of ascite tumour-derived cell lines [11,12].
Here we show that MGS accumulates, under certain metabolic conditions, in the nuclear compartment of various cell types that endogenously express the enzyme. We also analyze the subnuclear distribution of the enzyme and we address the molecular mechanisms and determinants involved in its nucleocytoplasmic translocation.
- Top of page
- Experimental procedures
- Supplementary material
MGS is translocated between the nucleus and cytoplasm
We previously reported that the chimeric protein resulting from the fusion of GFP with HsMGS is translocated between the nucleus and cytoplasm when transiently overexpressed in C2C12, COS-1 cells and primary cultured rat hepatocytes [9,10]. Here we have used primary cultured cells from human muscle and the MGS3 antibody, directed against the last nine amino acids of the C-terminus of the protein (see Experimental procedures), to analyze the subcellular distribution of HsMGS in cells that endogenously express this protein. We first characterized the MGS3 polyclonal antibody (Fig. 1). Primary cultured muscle cells, which can be induced to form highly differentiated multinucleated cells (myotubes) that express large amounts of HsMGS, were infected with the AdCMV–GFP or AdCMV–GFP–HsMGS adenovirus, and the cellular lysates were analyzed by western blot using MGS3 as primary antibody. Cells infected with the AdCMV–GFP adenovirus presented a single band, which corresponds to endogenous HsMGS, whereas myotubes infected with the AdCMV–GFP–HsMGS showed an additional more intense band, the molecular mass of which corresponds to that of the enzyme fused to GFP (Fig. 1A). In another experiment, COS-1 cells were transfected with the pEGFP–HsMGS plasmid and the cellular lysates were subjected to immunoprecipitation with the MGS3 antibody. The supernatant resulting from this immunoprecipitation showed a 50% decrease in total GS activity compared with the control supernatant treated in the same conditions but without the antibody. The immunoprecipitates were analyzed by western blot using a distinct MGS antibody, obtained from chickens immunized with purified rabbit MGS . Only the immunoprecipitate obtained in the presence of the MGS3 antibody showed a band corresponding to the GFP–HsMGS fusion protein (Fig. 1B).
Finally, the specificity of the MGS3 antibody in the immunolocalization of MGS was analyzed. L6 myoblasts were infected with the AdCMV–GFP–HsMGS adenovirus and two days after infection were incubated for 16 h in medium devoid of glucose, to promote nuclear accumulation of the GFP–HsMGS chimera. The infected cells were further incubated for 6 h in the absence or presence of 30 mm glucose, and were finally subjected to immunocytochemistry with the MGS3 antibody, before observation in the confocal microscope. The cytoplasmic colocalization of the green fluorescence, arising form the GFP fused to HsMGS, and the red fluorescence, which originates from the immunodetection of the chimeric enzyme with the MGS3 antibody, is very high (Fig. 1C). In the L6 myoblasts incubated with glucose, which do not present any nuclear fluorescence, the enzyme forms characteristic glycogen-bound aggregates  which are labelled by both fluorescent markers and appear as yellow or orange spots in the Merge panel (Fig. 1C). The GFP–HsMGS chimera is also specifically labelled by the MGS3 antibody in the cytoplasm of cells incubated without glucose, but not so in the nucleus. In this compartment the GFP-fused enzyme presents both a nucleoplasmic diffuse distribution and a speckled pattern. The nuclear GFP–HsMGS aggregates are recognized by the MGS3 antibody and appear yellow in the Merge panel (Fig. 1C), whereas the diffuse nucleoplasmic enzyme appears green. This observation indicates that immunodetection of nuclear MGS is not complete, and only the spots where the concentration of the enzyme is high are efficiently labelled by the antibody. In any case, this experiment shows that the antibody specifically recognizes MGS and that, with some limitations, it is useful to perform immunocytochemistry.
Next, we proceeded to analyze the intracellular distribution of endogenous HsMGS in cells that naturally express the protein. Thus, differentiated myotubes were incubated in a glucose-free medium for 16 h and then maintained in medium with or without 30 mm glucose for an additional 6 h. Cells were fixed and processed for immunocytochemistry using the MGS3 antibody. In the myotubes incubated with glucose, HsMGS was detected in the cytoplasm essentially in the form of large aggregates (Fig. 2A). When the cells were deprived of glucose for 22 h, these aggregates were fewer and smaller, and a diffuse signal appeared throughout the cytoplasm (Fig. 2B). When the glucose-deprived myotubes were further incubated for 1 h with 100 µm forskolin, a compound that promotes glycogen degradation through the activation of glycogen phosphorylase, HsMGS distributed between the cytoplasm and the nucleus (Fig. 2C). Some of the cells also showed spherical particles in the nucleoplasma, which were stained with the MGS3 antibody. Incubation with isoproterenol, an adrenergic agonist that also promotes glycogenolysis, provided similar results (not shown).
In addition, the subcellular distribution of MGS was analyzed in 3T3-L1 cells, a murine adipocyte model that also expresses high levels of this isoform of the enzyme. Owing to the high sequence identity between murine and human MGS, the MGS3 antibody cross-reacts with mouse MGS. The results obtained were similar: when glucose was present in the incubation medium, the enzyme was found exclusively in the cytoplasm, whereas, in cells deprived of glucose, MGS was distributed between the cytoplasm and the nucleus, which, in some cases, showed a speckled pattern (not shown). In the 3T3-L1 cell line, which accumulates lower concentrations of glycogen than primary cultured muscle cells, the addition of drugs that promote glycogen degradation was not required to observe nuclear MGS.
Nuclear concentration of MGS correlates inversely with cellular glycogen content
The previous results show that the behaviour of endogenous MGS is identical with that previously described for GFP–HsMGS [9,10], indicating that fusion of the enzyme to the GFP fluorescent marker does not alter its characteristic behaviour. Furthermore, these observations point to an inverse relationship between the cellular glycogen content and the nuclear accumulation of the enzyme. To analyze this relationship, we next performed experiments using COS-1 cells transfected with a plasmid encoding the GFP–HsMGS fusion protein. These cells were then deprived of glucose for 16 h and were incubated for an additional 6 h in a medium containing 30 mm glucose, a medium without glucose, or for 1, 3 or 6 h in a glucose-free medium containing 100 µm forskolin, to attain a range of glycogen concentrations. For each condition, nuclear accumulation was quantified by fluorescence imaging of 100 or more cells, and, in parallel, glycogen content was measured in triplicate as indicated in Experimental procedures. The data presented in Fig. 3 are the result of three independent experiments. Independently of the experimental condition used to attain a given concentration of glycogen, the number of cells that exhibited HsMGS nuclear staining was significant only at low concentrations of the polysaccharide (Fig. 3).
HsMGS is exported from the nucleus to the cytoplasm via the CRM1/exportin pathway
Leptomycin B is an inhibitor of the classical export pathway, which acts by covalently modifying the protein CRM1/exportin 1, an evolutionarily conserved receptor for the nuclear export signal (NES) of proteins . To analyze whether HsMGS was exported from the nucleus through this pathway, L6 myoblasts were infected with an adenovirus that drives the overexpression of GFP–HsMGS and, 32 h after infection, were incubated with Dulbecco's modified Eagle's medium (DMEM) without glucose for 16 h, to achieve nuclear concentration of the fusion protein. Cells were then incubated for an additional 6 h in a medium devoid of glucose (Fig. 4A) or, alternatively, were treated in a medium containing 30 mm glucose, in the presence or absence of 100 nm leptomycin B. In contrast with cells treated with glucose alone, in which nuclei were devoid of GFP fluorescence (Fig. 4B), the nuclei of those treated with leptomycin B showed a strong fluorescent signal (Fig. 4C). Interestingly, whereas in the cells incubated in the absence of glucose GFP–HsMGS presented a speckled pattern (Fig. 4A), the fusion protein showed a diffuse and more homogeneous nuclear distribution in those treated with glucose and leptomycin B (Fig. 4C). The cytoplasmic fluorescence observed in these cells, which in some cases appears as the typical glycogen-bound aggregates (Fig. 4C), arises from the residual HsMGS that does not enter the nucleus, even after prolonged incubation periods in the absence of glucose. Identical results were obtained when the experiment was repeated with COS-1 cells transfected with the GFP–HsMGS construct, indicating the presence of an NES in the sequence of HsMGS or the involvement of an NES-containing protein in its export from the nucleus to the cytoplasm.
To rule out the possibility that the glucose-induced change in subcellular localization of MGS was due to its degradation in the nucleus and targeting of newly synthesized protein to the cytoplasm, we performed experiments in the presence of 2 µm cycloheximide. In COS-1 cells which express the GFP–HsMGS chimera, blocking protein translation 30 min before the addition of glucose did not alter the observed redistribution of the protein from the nucleus to the cytoplasm.
Sequence determinants of the HsMGS nucleocytoplasmic transport
To identify elements of the HsMGS sequence involved in the control of its nuclear import and export, we performed deletion analysis and site-directed mutagenesis on the GFP–HsMGS fusion protein. Several plasmids encoding HsMGS fragments fused to GFP were constructed (Fig. 5A). COS-1 cells were transfected with these constructs and were allowed to express the chimeric proteins for 32 h. The subcellular distribution of the fusion proteins was analyzed by observing the GFP fluorescence in cells deprived of glucose or incubated in medium containing 30 mm glucose.
Among the constructs assayed, the CT1 mutant (amino acids 554–737 of HsMGS) was the only one that was concentrated in the nucleus in both the absence and presence of glucose in the incubation medium (Fig. 5B,C). The complementary mutant, HsMGSΔCT1 (amino acids 1–553 of HsMGS), was excluded from the nucleus and gave rise to small cytoplasmic aggregates in both culture conditions (Fig. 5F,G). However, the HsMGSΔCT1 mutant was found in the nucleus of some cells incubated in a glucose-free medium containing leptomycin B to block nuclear export (Fig. 5H). Furthermore, when the CT1 fragment was exchanged for its homologous counterpart from rat liver GS, the chimera (HsMGS/RnLGS) behaved similarly to HsMGS, and in the absence of glucose showed a nuclear speckled pattern (Fig. 5N), whereas in the presence of the monosaccharide formed cytoplasmic aggregates (Fig. 5M).
The region corresponding to amino acids 554–633 of HsMGS is essential for its nuclear concentration, as the two deletion mutants CT2 (Fig. 5D,E) and CT3 (not shown) which lack this fragment did not accumulate in the nucleus, but rather were distributed uniformly between the nuclear and cytoplasmic compartments, independently of the culture conditions. This conclusion is reinforced by the observation that the truncated protein HsMGSΔCT2, which contains amino acids 553–633, was not excluded from the nucleus in the absence of glucose (Fig. 5J). Furthermore, in this compartment it gave rise to speckles, although smaller and less well defined than those produced by the wild-type enzyme in these conditions (Fig. 7B). Like the full enzyme, in response to glucose, this mutant was translocated to the cytoplasm where it was bound to the glycogen particles (Fig. 5I). Identical behaviour was observed for the two remaining C-terminal deletion mutants examined, HsMGSΔCT3 and HsMGSΔCT4, which also contain amino acids 553–633 (not shown). This last truncated form of the protein ends just before the first C-terminal phosphorylation site (3a, Ser641, see below) and therefore lacks all the C-terminal phosphorylation sites.
Finally, the HsMGSΔNT1 fragment, which, in contrast with HsMGSΔCT2, is truncated at its N-terminus and lacks the first 131 amino acids, also entered the nucleus in the absence of glucose, although it was not concentrated in this compartment (Fig. 5L). In addition, the presence of glucose in the incubation medium induced its translocation to the cytoplasm (Fig. 5K), but this process was not as complete as that observed for the wild-type protein (Fig. 7D).
The phosphorylation sites that regulate HsMGS activity are not involved in the control of its subcellular distribution
Phosphorylation is often involved in the regulation of nucleocytoplasmic transport [15,16]. HsMGS is a multiphosphorylated protein, therefore, we studied the possible role of known phosphorylation sites, which regulate the activity of the enzyme, in the control of the enzyme distribution. The serine residues of MGS that undergo reversible phosphorylation  were mutated to alanine in the GFP–HsMGS plasmid (Fig. 6A), and COS-1 cells were transfected with the corresponding plasmids. All the mutants assayed, from those in which a single residue was changed to that in which all nine serine residues had been replaced by alanine, were catalytically active and showed an increased –Glc6P/+Glc6P activity ratio (Table 1), as previously described for the rabbit muscle protein . However, no differences in their subcellular localization with respect to the wild-type protein were observed (Fig. 6B). In the presence of glucose, the cytoplasm showed the characteristic aggregates that correspond to the glycogen-bound enzyme , and, in the absence of the monosaccharide, all phosphorylation mutants were concentrated in the nucleus and formed discrete particles.
|Mutant||GS activity [mU·(mg protein)−1]||Activity ratio (–/+Glc6P)|
The region involved in Glc6P allosteric regulation is crucial for the nuclear accumulation of HsMGS
Yeast and rabbit muscle GS have an arginine-rich region near the C-terminus, which acts as a switch or fringe and controls the overall conformation and therefore the activity of the protein in response to Glc6P concentration [4,5]. To analyze whether this conformational change also affects the subcellular distribution of the enzyme, the mutants R1 (R579/R580/R582A) and R2 (R586/R588/591A), in which all three of the Arg residues indicated were mutated to Ala (Fig. 7A), were transiently expressed in COS-1 cells as GFP fusion proteins. The R1 and R2 mutant enzymes were catalytically active and showed a very high –Glc6P/+Glc6P activity ratio (Table 1), which is consistent with the results obtained for the rabbit muscle enzyme . Analysis of the subcellular localization of the mutants showed that their behaviour differed from that of the wild-type enzyme. The R1 and R2 mutants accumulated and formed spherical particles in the nucleus in the absence of glucose (not shown). However, their nuclear concentration was not as consistent and repetitive as that of the wild-type enzyme (Fig. 7C). Furthermore, time-course experiments revealed that, in response to glucose, R1 and R2 were quickly translocated from the nucleus to the cytoplasm. No nuclear signal was detected after only 2 h of incubation with glucose, in contrast with the 4 h required for the wild-type enzyme to achieve the same degree of cytoplasmic translocation.
When the two Arg-rich regions were eliminated in the R1/R2 mutant, this behaviour was exacerbated. In the presence of glucose, R1/R2 gave rise to the typical glycogen-bound particles in the cytoplasm (Fig. 7E), but in the absence of the monosaccharide, this mutant was barely detected in the nuclear compartment (Fig. 7C). However, blockage of nuclear export by the addition of leptomycin B to the incubation medium, even in the presence of glucose, showed that R1/R2 was still able to enter the nucleus, although it gave rise to smaller nuclear aggregates (Fig. 7B) than those observed with the wild-type enzyme (Fig. 7F).
In another set of experiments, COS-1 cells transiently expressing the GFP–HsMGS construct were first incubated in the absence of glucose to elicit nuclear accumulation of the protein, and then with 30 mm 6-deoxyglucose. After 6 h of treatment with the deoxysugar, the fusion protein remained in the form of nuclear aggregates (not shown) identical with those observed in the cells kept in the absence of glucose.
Putative nuclear localization and nuclear export signals in the sequence of HsMGS
As shown above, the export of HsMGS from the nucleus to the cytoplasm is blocked by leptomycin B, indicating that this translocation is mediated by a NES. Although the NES consensus sequence is rather ill defined, these signals are usually short hydrophobic sequences with a high content of Leu and Ile residues . Three putative NESs were identified by visual inspection of the protein sequence, and the key hydrophobic amino acids were mutated to Ala (Fig. 7A). Two of these mutants, NES-1 and NES-3, showed identical subcellular distribution to the wild-type protein, both in the absence and presence of glucose. NES-2 showed a uniform cytoplasmic distribution in each condition, but its expression level was very low and showed null activity. These observations indicate that the mutations introduced in NES-2 are deleterious and the resulting protein cannot adopt its native conformation, thus precluding any conclusions to be drawn.
Apart from the Arg-rich sequences mentioned earlier, an additional putative nuclear localization signal (NLS), composed of three contiguous basic residues, was identified by visual inspection of the HsMGS sequence. However, when the key residues of this sequence were mutated, the resulting NLS-1 mutant behaved like the wild-type protein: it showed nuclear spherical aggregates in the absence of glucose and was translocated to the cytoplasm when the sugar was added (not shown).
HsMGS subnuclear distribution
The confocal microscopic images of the GFP–HsMGS transfected cells and primary muscle cells immunostained with MGS3 revealed the presence of nuclear HsMGS particles. This observation implies that the enzyme may be localized in a subnuclear compartment, as described for several nuclear proteins. To check this possibility, we performed double-labelling experiments using the GFP-tagged form of HsMGS or the MGS3 antibody and other antibodies directed against protein markers of several subnuclear compartments. We used antisera against human centromer and human Sm-antigen , and antibodies against the promyelocytic leukaemia oncoprotein (PML) , SC-35  and p80-coilin .
Human primary cultured myotubes were treated, as described previously, to achieve HsMGS nuclear localization, fixed, and subjected to double immunolabelling. For p80-coilin immunostaining, COS-1 cells transfected with the GFP–HsMGS plasmid were used, because there was an incompatibility between the p80 antibody and the fixation method required for MGS3 immunolabelling. No colocalization was detected between HsMGS and the centromer antigen, the Sm-antigen or SC-35 (not shown). In contrast, a masking phenomenon was observed for the PML protein. Myotubes that did not show nuclear staining with MGS3 had the typical 10–30 PML bodies per nucleus , whereas those with a strong HsMGS nuclear signal exhibited null or highly decreased PML fluorescence. In the few cases where both antigens were labelled, PML staining was always adjacent to the MGS3 positive aggregates (Fig. 8A). In the case of p80-coilin, a marker of Cajal bodies, there was overlapping between GFP–HsMGS and the fluorescent signal of the antibody against p80-coilin (Fig. 8B).
In many cases, the formation of subnuclear compartments is dependent on the transcriptional state of the cells. For this reason, transcription inhibitors, such as actinomycin D, produce the reorganization of these compartments and the proteins that form them. To study whether inhibition of transcription affected the subnuclear distribution of HsMGS, cultured myotubes were infected with the GFP–HsMGS adenovirus. Cells were incubated in the absence of glucose for 16 h and for an additional 2 h in the same medium containing 100 µm forskolin, to ensure nuclear localization of the fusion protein. Actinomycin D was then added at 10 µg·mL−1, a concentration that inhibits all RNA polymerases, and time-lapse images of the cells maintained at 37 °C were taken for 4 h, using a confocal microscope. The initial large speckles of GFP–HsMGS disaggregated to form new smaller ones and finally the fluorescent signal was dispersed throughout the nucleus (Fig. 9; additional data in video; AVI file). Identical results were obtained when cells were fixed before observation in the confocal microscope: short treatments with actinomycin D in the absence of glucose led to the formation of smaller and more abundant nuclear speckles, while prolonged treatments caused the complete disaggregation of nuclear GFP–HsMGS.
- Top of page
- Experimental procedures
- Supplementary material
Nucleocytoplasmic translocation of MGS
In this study we show that, as a result of glucose deprivation and depletion of the glycogen reserves, MGS is concentrated and aggregates in the nuclear compartment of cells that naturally express this protein, such as cultured myotubes and adipocytes. When glucose is added back, the enzyme is translocated to the cytoplasm, where it gives rise to large glycogen-bound aggregates. The high affinity of GS for glycogen, its substrate and product, has long been recognized [7,10] and constitutes a key factor in schemes designed for the purification of the enzyme from its natural sources. We have previously shown in COS-1 cells that the GFP–HsMGS chimera binds to intracellular glycogen and that the particulate cytoplasmic pattern shown by the fusion protein is due to its close association with glycogen particles . The present results indicate that a significant degree of nuclear MGS accumulation occurs only in cells with very low or null glycogen content. This explains the difficulty in observing an apparent nuclear accumulation of MGS by other authors  and the need to use drugs that promote glycogenolysis, in addition to glucose deprivation, to observe this phenomenon. In comparison with other cell types, muscle cells accumulate large amounts of the polysaccharide, and muscle glycogen stores are never fully depleted in vivo.
Mechanism of nuclear import/export of HsMGS
The regulation of MGS nucleocytoplasmic transport by its own sequence elements is quite complex. Furthermore, the fact that the enzyme binds strongly to cytoplasmic glycogen greatly complicates this type of study, as an apparent nuclear accumulation of MGS can only be observed in cells that are depleted of the polysaccharide. However, we have been able to show that export of the protein from the nucleus to the cytoplasm is blocked by leptomycin B and is therefore mediated by CRM1 and a NES. We analyzed three mutant forms of HsMGS in which the key residues of three putative NESs were mutated to alanine. However, NES-1 and NES-3 were normally exported from the nucleus, and no conclusion could be drawn from NES-2, as this point mutant does not adopt the native conformation. The behaviour of HsMGSΔNT1 indicates that the N-terminal region of HsMGS is crucial for both efficient nuclear import and export. This deletion mutant, in spite of barely entering the nucleus in the absence of glucose, is not fully exported to the cytoplasm in response to the sugar. The fragment eliminated in HsMGSΔNT1 (amino acids 1–133) does not contain any recognizable NESs, apart from NES-1, which, as indicated above, is not a genuine NES. Thus, either HsMGS possesses an NES in the fragment comprising amino acids 1–133 that we have been unable to identify or it is exported to the cytoplasm through the interaction with an NES-containing protein which acts as a carrier. In the second case, the N-terminal region of the enzyme would be essential for the interaction with the hypothetical transport protein.
Similarly, our attempts to characterize the mechanism of nuclear import of HsMGS have not allowed us to identify a short fragment in the sequence of the protein that is directly responsible for this process. All the deletion and point mutants analyzed in this study can be found in the nucleus, either by simple depletion of glucose or by blocking nuclear export with leptomycin B. The CT1, CT2 and CT3 mutants fused to GFP have molecular masses of 48, 40 and 38 kDa, respectively. These values are below or approach the nuclear pore exclusion limit , allowing, in principle, the passive diffusion of the chimeric proteins between the nuclear and the cytoplasmic compartments. However, the molecular masses of all the other constructs are well above this exclusion limit and therefore they probably enter the nucleus by active transport.
The highly basic segment near the C-terminus, between amino acids 578 and 590, which has been implicated in the allosteric response of the enzyme to Glc6P[4,5], possesses the characteristics of an NLS. However, when this group of basic residues was eliminated by deletion, as in HsMGSΔCT1, or by mutation, as in R1, R2 and R1/R2, the mutant proteins showed impaired nuclear concentration and aggregation, but still entered the nucleus. The NLS-1 mutant, in which the only additional cluster of basic residues found in HsMGS was mutated, behaves identically with the wild-type enzyme. These results imply that the nuclear import of HsMGS is mediated by another class of NLS that is not composed of basic residues , or through interaction with another protein.
The N-terminal region and the Arg-rich cluster are essential for the nuclear accumulation and aggregation of HsMGS
Neither the C-terminal deletion mutant HsMGSΔCT1 (amino acids 1–554) nor HsMGSΔNT1, which lacks the first 133 amino acids, accumulated in the nucleus or formed the typical spherical aggregates observed with the wild-type protein. Thus, elements that are distant in the sequence are required for the protein to be retained and properly structured in the nucleus, suggesting that this phenomenon occurs through the interaction of HsMGS with a nuclear component and that both termini are essential for this interaction. In contrast, the remaining C-terminal deletion mutants examined, HsMGSΔCT2, ΔCT3 and ΔCT4, all of which contain amino acids 553–633 in their respective sequences, exhibited similar behaviour to the wild-type, although the nuclear speckles in these mutants were smaller and less well defined. These observations indicate that amino acids 634–737 are not essential for the nuclear concentration and aggregation of the enzyme. However, they highlight the relevance of the fragment comprising amino acids 553–633, which includes the Arg-rich cluster mentioned above, in the nuclear retention of HsMGS. Interestingly, the HsMGS/RnLGS chimera, in which the CT1 fragment was replaced by that of RnLGS, behaved almost identically with HsMGS. Although this is the region with the lowest degree of identity between the two isoforms of the enzyme, the region encompassing the Arg-rich cluster is almost fully conserved in RnLGS, again indicating that the presence of this cluster is crucial for nuclear retention and aggregation. To further illustrate the complexity of this system, it is worth mentioning that, although RnLGS is translocated from a homogeneous cytoplasmic distribution to a region near the plasma membrane in response to glucose [7,26], it is never concentrated in the nucleus. Thus, the Arg-rich cluster present in the two GS isoforms is necessary but not sufficient to allow the nuclear concentration and aggregation of the enzyme, and the regulated nucleocytoplasmic translocation of HsMGS is dependent on other features that are distributed along the sequence of the muscle enzyme and absent from the liver isoform.
Role of Glc6P in the nucleocytoplasmic translocation of HsMGS
Several lines of evidence indicate that the nucleocytoplasmic shuttling of HsMGS is not regulated by the reversible phosphorylation of the sites  that control its activity. As indicated above, the HsMGSΔCT2, ΔCT3 and ΔCT4 deletion mutants, which lack some or all of the C-terminal phosphorylation sites, behaved almost identically with the wild-type enzyme. In addition, the substitution of all the Ser residues known to undergo reversible phosphorylation by Ala, a nonphosphorylable residue, did not affect the subcellular distribution of the mutant enzyme in any condition.
In contrast, the Arg-rich cluster between amino acids 578 and 590 is crucial for correct subcellular distribution of HsMGS. This segment has been implicated in the allosteric activation of MGS by Glc6P. Hanashiro & Roach  showed that the catalytic activity of two mutant forms of rabbit MGS, R1 (R579A/R580A/R582A) and R2 (R586A/R588A/R591A), was insensitive to the presence of Glc6P. In the present study, the homologous mutants of HsMGS, R1 and R2, were concentrated less efficiently in the nucleus of cells incubated in the absence of glucose, and, in response to the sugar, exited this compartment more quickly than the wild-type protein. The double mutant R1/R2 was hardly retained at all in the nucleus and did not form nuclear speckles in the absence of glucose. One explanation for these observations assumes that the Arg-rich cluster of MGS is part of a complex nuclear retention (and aggregation) system, also involving other regions of the protein, which is regulated by binding of Glc6P to the enzyme. Interestingly, 6-deoxyglucose did not induce the export of nuclear HsMGS, indicating that, if glucose is involved directly in the control of the export process, its 6-OH group is crucial for this function or, more likely, that glucose must be converted into Glc6P to bring about these changes.
When cells showing nuclear HsMGS speckles were simultaneously treated with glucose and leptomycin B to block the nuclear export, the size of the speckles largely decreased, and, in many cells, HsMGS presented a homogeneous nucleoplasmic distribution. These observations indicate that glucose, probably after its conversion into Glc6P, triggers the disaggregation of the nuclear HsMGS speckles and that this process occurs before the export of the enzyme to the cytoplasm.
What is the role of nuclear MGS?
The subcellular distribution of MGS is regulated by glycogen content, one of its substrates, and by glucose or a derived metabolite, probably Glc6P. What is the physiological meaning of this regulated differential localization of MGS?
Some studies have reported the presence of GS activity and glycogen in the nucleus of certain cell lines [11,12]. Owing to the polymeric nature of glycogen and its large molecular mass, the presence of the polysaccharide in the nucleus can be explained only if the enzymes responsible for its synthesis, namely GS and the branching enzyme, can enter this compartment and perform their catalytic function there. In this regard, it has been shown that glycogen is required for the formation of particulate chromatin intermediates in the assembly of functionally active nuclei .
The nuclear accumulation of MGS is not homogeneous, but rather in the form of spherical aggregates, which resemble the so-called nuclear speckles. These nuclear subcompartments, which are not delimited by membranes, are characterized mainly by protein composition and are involved in distinct nuclear tasks. More specifically, PML bodies are small spherical domains which have been implicated in several functions, such as transcriptional regulation , nuclear storage , growth control  and apoptosis . The nucleocytoplasmic-shuttling phosphoprotein p80-coilin is a specific marker of Cajal bodies, which are also small spherical structures often associated with PML bodies, the function of which is unknown . Here we show that the nuclear accumulation of HsMGS-triggered glycogen depletion produces a masking effect in the detection of the PML bodies with a specific antibody and also that nuclear HsMGS colocalizes with p80-coilin. Furthermore, the particulate subnuclear organization of HsMGS, like the morphology and number of many nuclear compartments , is sensitive to the inhibition of transcription by actinomycin D. All these observations suggest that, in addition to its metabolic function, MGS participates in nuclear processes. Glyceraldehyde-3-phosphate dehydrogenase, a classical ‘metabolic’ enzyme, is also involved in the regulation of gene expression and links this process to energy metabolism . In the same regard, MGS may be a sensor of cellular energy reserves. When glycogen stores are depleted, MGS can no longer bind to the polysaccharide that anchors the enzyme in the cytoplasm and is translocated to the nucleus, where it performs the hypothetical additional function(s). The increase in Glc6P concentration, as a result of the influx of glucose in the cell, is the signal for the re-export of MGS. This ends the nuclear activity of HsMGS, which is relocated in the cytoplasm in the presence of its allosteric activator, ready for glycogen synthesis.
- Top of page
- Experimental procedures
- Supplementary material
Plasmid and adenovirus construction
Standard molecular cloning techniques were used throughout . The pEGFP–HsMGS construct coding for GFP–HsMGS has already been described  and named there as pEGFP-C1/HMGS. The full coding sequence of the GFP–HsMGS fusion protein was excised from pEGFP–HsMGS and used to construct AdCMV–GFP–HsMGS, as described by Becker et al. . Point mutations were introduced into pEGFP–HsMGS by site-directed mutagenesis, using the QuikChange kit (Stratagene, La Jolla, CA, USA). Whenever possible, the oligonucleotides used (Table 2, only the sense oligonucleotide for each mutant is listed) introduced (+, underlined) or eliminated (–) a restriction site to facilitate screening.
|Name||5′ primer||Restriction site|
|NES-1||CGATGTCAAGCGCGCCGGGTGCAGAGGACTGGG||–HindIII + BssHII|
|NLS-1||CACAGCACTTGCTCCGGAGGCGCCCAGATATTGTGACC||+BspHI + NarI|
The deletion mutants were constructed as follows: a SalI restriction site was introduced at nucleotide 1817 in the coding sequence of HsMGS  to yield pEGFP–HsMGS(CT1), which encodes a fusion protein with a conservative change (Leu553 to Val) with respect to pEGFP–HsMGS. This construct was digested with SalI, and the fragment encoding the C-terminal sequence of HsMGS was ligated to pEGFP-C1 (SalI) (Clontech, Mountain View, CA, USA) to yield pEGFP-CT1, and the remaining was religated, thereby providing pEGFP–HsMGSΔCT1. pEGFP–HsMGS was digested with KpnI. The fragment encoding the C-terminal sequence of HsMGS was ligated to pEGFP-C2 (KpnI) (Clontech) to yield pEGFP-CT2, and the remaining was religated, providing pEGFP–HsMGSΔCT2. pEGFP–HsMGS was digested with BamHI. The fragment encoding the C-terminal sequence of HsMGS was ligated to pEGFP-C3 (BamHI) (Clontech) to yield pEGFP-CT3, and the remaining was religated, providing pEGFP–HsMGSΔCT3. pEGFP–HsMGSΔCT3 was constructed by mutagenesis of pEGFP–HsMGS, replacing the codon corresponding to Ser641 (the phosphorylation site 3a) by a stop codon. pEGFP–HsMGS was digested with SacI/SalI. The fragment encoding the C-terminal sequence of HsMGS was ligated to pEGFP-C3 (SacI/SalI) (Clontech) to yield pEGFP-ΔNT1. A cDNA encoding RnLGS contained in a pGEM-T vector (Promega, Madison, WI, USA) (pGEM-T/RnLGS) was digested with SalI, and the fragment encoding the C-terminal sequence of RnLGS was ligated to pEGFP–HsMGSΔCT1 (SalI), yielding pEGFP–HsMGS/RnLGS. The DNAs encoding the recombinant proteins were sequenced using the ABI-PRISM DNA sequencing kit and an ABI-PRISM 377 automatic DNA sequencer (Applied Biosystems, Foster City, CA, USA).
Cell culture and transfection
293 Fibroblasts (ATCC No. CRL-1573, Manassas, VA, USA) were grown on plates in DMEM (Gibco, Carlsbad, CA, USA), supplemented with 30 mm glucose, 10% (v/v) fetal bovine serum (FBS; Biological Industries, Kibbutz Beit Haemek, Israel) and penicillin/streptomycin (Biological Industries). These cells were used to construct and propagate the recombinant adenovirus AdCMV–GFP–HsMGS .
L6 myoblast cells (ATCC No. CRL-1458) were grown in DMEM supplemented with 30 mm glucose, 10% (v/v) FBS and penicillin/streptomycin. Viral infection of the cells was performed over 1 h in DMEM containing 30 mm glucose and 2% (v/v) FBS at a multiplicity of infection of 5. The cells were rinsed with phosphate-buffered saline (NaCl/Pi) and incubated with DMEM containing 30 mm glucose and 10% (v/v) FBS at 37 °C in humidified 5% CO2/95% air (v/v). Experiments were performed 32 h after transfection. Leptomycin B was used at 100 nm.
COS-1 cells (ATCC No. CRL-1650) were grown in DMEM, supplemented with 30 mm glucose, 10% (v/v) FBS, and penicillin/streptomycin, on 60-mm dishes with a glass coverslip (for analysis of glycogen content and nuclear concentration of MGS), on glass coverslips in 35-mm dishes or on 100-mm dishes for maintenance. Cells cultured on 60-mm dishes were transfected using 625 mg DEAE-dextran (Sigma, St Louis, MO, USA), 0.5 mmol chloroquine (Sigma), and 10 mg plasmid DNA per dish in DMEM. After a 4-h incubation, cells were treated for 2 min in DMEM containing 10% dimethyl sulfoxide (Sigma) and 10% FBS. They were then washed with DMEM plus 10% FBS and maintained in this medium. Cells grown on coverslips were transfected at 70–80% of confluence using Superfect (Qiagen, Valencia, CA, USA), following the manufacturer's instructions. After transfection (4–5 h) at 37 °C in humidified 5% CO2/95% air, cells were washed in NaCl/Pi and incubated in DMEM supplemented with 30 mm glucose and 10% (v/v) FBS. Cells cultured on 100-mm dishes were transfected at 90% of confluence by electroporation. Briefly, for each plate, cells were detached by treatment with trypsin, washed with NaCl/Pi, and resuspended to a final volume of 800 µL in DMEM without antibiotics but with 100 µg salmon sperm DNA (Stratagene) and 100 µg of the desired plasmid construction. Samples were transferred to a Gene Pulser cuvette (Bio-Rad, Hercules, CA, USA) with an electrode gap of 0.4 cm and electroporated in a Gene Pulser II device (Bio-Rad) at 200 V and 950 µF. Cells were then diluted in the appropriate volume of DMEM, supplemented with 30 mm glucose, 10% (v/v) FBS, and penicillin/streptomycin, and seeded on glass coverslips in 35-mm dishes. Experiments were performed 48–52 h after transfection. Cells were preincubated overnight in DMEM without glucose, and on the day of the experiment were incubated in DMEM without or with glucose at the indicated concentration for various periods of time. At the end of the incubation, cells grown on 60-mm dishes were rinsed twice with NaCl/Pi and the coverslip was frozen in liquid nitrogen. Cells grown on coverslips were fixed for 20 min at room temperature in NaCl/Pi containing 4% paraformaldehyde (Fluka, Buchs SG, Switzerland) and washed several times with NaCl/Pi. Coverslips were air-dried and finally mounted on to glass slides, using the Immuno Fluore mounting medium (MP Biomedicals, Irvine, CA, USA). Human muscle primary cultures were initiated from satellite cells from muscle biopsy specimens from patients considered free of any disease after all diagnostic studies had been reported (biopsies were carried out with informed consent and approval of the Human Research Ethics Committee of the Hospital Clínic, Barcelona). Aneural muscle cultures were established in monolayers following the explant–re-explantation technique, as previously described . The cultures were grown in DMEM/M-199 medium (3 : 1, v/v; Gibco) supplemented with 10% (v/v) FBS, 10 µg·mL−1 insulin (Sigma), 20 mmol·L−1 glutamine, 25 ng·mL−1 fibroblast growth factor, 10 ng·mL−1 epidermal growth factor (BD, Franklin Lakes, NJ, USA), and penicillin/streptomycin/amphotericin B (Biological Industries). Immediately after myoblast fusion, cells were rinsed with Hank's balanced salt solution (Biological Industries) and incubated in a medium devoid of fibroblast growth factor, epidermal growth factor, and glutamine. Glucose deprivation was achieved by incubating differentiated myotubes in DMEM without glucose for variable times followed by incubation in DMEM without glucose plus 100 µm forskolin (Sigma) or 100 µm isoproterenol (Sigma). Actinomycin D (Sigma) was used at 10 µg·mL−1. Cells grown on coverslips for immunocytochemistry were fixed with methanol at −20 °C or with 4% (v/v) paraformaldehyde in NaCl/Pi for 20 min at room temperature and rinsed several times with NaCl/Pi.
3T3-L1 cells (ATCC No. CL-173) were grown in DMEM containing 30 mm glucose, 10% (v/v) heat inactivated FBS, 25 mm Hepes and streptomycin/ampicillin. Cells were maintained undifferentiated by replating them before reaching 80% confluence. The method described by Rubin et al.  was used to differentiate the cells into adipocytes. Differentiation was considered complete after 5–7 days. Cells were incubated in DMEM with or without 30 mm glucose for 16 h, rinsed twice with NaCl/Pi, fixed at room temperature for 20 min with 4% (v/v) paraformaldehyde in NaCl/Pi, and rinsed again with NaCl/Pi several times.
GS activity assays and glycogen content
For the measurement of glycogen content, cell monolayers were scraped into 30% (v/v) KOH, and the extract was then boiled for 15 min and centrifuged at 5000 g for 15 min. Glycogen was measured in the cleared supernatants as described . To determine GS activity, frozen cell monolayers were scraped using homogenization buffer that consisted of 10 mm Tris/HCl (pH 7.0), 150 mm KF, 15 mm EDTA, 15 mm 2-mercaptoethanol, 10 mg·mL−1 leupeptin, 1 mm benzamidine, and 1 mm phenylmethanesulfonyl fluoride. Cells were ruptured by sonication. Protein concentration was measured as described by Bradford  using the Bio-Rad protein assay reagent. GS activity was measured in the presence or absence of 6.6 mm Glc6P as described previously . The activity measured in the absence of Glc6P represents the active form of the enzyme (I or a form), whereas the activity tested in the presence of 6.6 mm Glc6P is a measure of total activity. The ratio of these two activities is an estimate of the activation state of the enzyme.
Immunocytochemistry and antibodies
Coverslips were rinsed three times with NaCl/Pi, and permeabilized for 30 min with NaCl/Pi containing 0.2% (v/v) Triton X-100 (Sigma) and blocked for 10 min with NaCl/Pi containing 0.2% (v/v) Triton X-100 and 3% (w/v) BSA (Sigma). Primary antibodies were diluted in NaCl/Pi containing 3% (w/v) BSA and applied to the cells for 45 min at room temperature. Coverslips were then washed several times with NaCl/Pi and subjected to incubation with secondary antibodies for 30 min at room temperature. Finally, coverslips were washed, air-dried, and mounted on to glass slides using the Immuno Fluore mounting medium.
The MGS3 polyclonal antibody was raised in rabbit against a peptide containing the nine C-terminal amino acids of HsMGS; TSSLGEERN(729–737). Specific recognition of MGS was confirmed by western blot (at 1 : 1000 dilution), immunoprecipitation and measurement of the activity of the immunoprecipitates. The secondary antibodies used were a Texas Red-conjugated anti-rabbit IgG (Molecular Probes, Eugene, OR, USA) at 1 : 200 dilution or a fluorescein isothiocyanate-conjugated anti-rabbit IgG (Chemicon, Temecula, CA, USA) at 1 : 100 dilution. The second polyclonal antibody against MGS (a gift from J.C. Lawrence, University of Virginia School of Medicine, Charlottesville, VA, USA), made by immunizing chickens with purified rabbit MGS, was used in western blots at 1 : 5000 dilution. Incubation with human ANA-centromere autoantibody, at 1 : 1000 dilution, or human ENA-Sm autoantibody (The Binding Site, Birmingham, UK), at 1 : 100 dilution, was followed by incubation of Texas Red-conjugated anti-human IgG (Molecular Probes), at 1 : 200 dilution. A monoclonal antibody against SC-35 (a gift from X.D. Fu, University of California, La Jolla, CA, USA) was used at 1 : 50 dilution. 5E10 ascites supernatant containing a monoclonal antibody against PML (a gift from R. van Driel, E.C. Slater Institute, University of Amsterdam, the Netherlands) was used at 1 : 10 dilution. P80-coilin ascites supernatant (a gift from E.K.L. Chan, The Scripps Research Institute, La Jolla, CA, USA) was used at 1 : 10 dilution. The secondary antibodies used were a Texas Red (Molecular Probes) or a fluorescein isothiocyanate (Chemicon) conjugated anti-mouse IgG, diluted at 1 : 200 and 1 : 50, respectively.
Confocal microscopy and time-lapse microscopy
Fluorescence images were obtained with a Leica TCS 4D (Leica Lasertechnik, Heidelberg, Germany) confocal scanning laser microscope adapted to an inverted Leitz DMIRBE microscope and 63 × or 100 × (NA 1.4 oil) Leitz Plan-Apo objectives or with a Leica SPII Spectral microscope and 63 × (NA 1.3 oil) Leitz Plan-Apo objective. The light source was an argon/krypton laser (75 mW), and optical sections (0.1 µm) were obtained. Time-lapse microscopy was performed with the Leica TCS 4D microscope using the Delta TC3 Culture Dish System with temperature controller and objective heater (Bioptechs, Butler, PA, USA) to maintain the cells at 37 °C. The cell medium was buffered with 100 mm Hepes, pH 7.4, and covered with mineral oil to avoid evaporation.
- Top of page
- Experimental procedures
- Supplementary material
We thank Anna Adrover and Ivette López for technical assistance and Tanya Yates for assistance in preparing the English manuscript. E.C. and D.C. gratefully acknowledge doctoral fellowships from the Generalitat de Catalunya. This work was supported by Grant BMC2002-00705 from the Dirección General de Investigación (Ministerio de Ciencia y Tecnología, Spain), by grants RCMN C03/08 (Red de Centros) and RGDM G03/212 (Red de Grupos), from the Fondo de Investigaciones Sanitarias (Instituto de Salud Carlos III, Spain), and in part by the Juvenile Diabetes Foundation International.
- Top of page
- Experimental procedures
- Supplementary material
- 321989) Molecular Cloning: a Laboratory Manual, 2nd edn. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY., & (
- Top of page
- Experimental procedures
- Supplementary material
The following material is available from http://www.blackwellpublishing.com/products/journals/suppmat/EJB/EJB4738/EJB4738sm.htm
Video S1. Effect of actinomycin D on nuclear GFP-HsMGS in cultured human myotubes recorded by time-lapse confocal microscopy.