Báo cáo khoa học: Determinants of the nucleocytoplasmic shuttling of muscle glycogen synthase potx

Determinants of the nucleocytoplasmic shuttling of muscle glycogen synthase Emili Cid 1, *, Daniel Cifuentes 1,2 , Susanna Baque ´ 1 , Juan C. Ferrer 1 and Joan J. Guinovart 1,2 1 Departament de Bioquı ´ mica i Biologia Molecular, Universitat de Barcelona, Spain 2 Institut de Recerca Biome ` dica de Barcelona-Parc Cientı ´ fic de Barcelona, Universitat de Barcelona, Spain Glycogen synthase (GS; EC 2.4.1.11) catalyzes the addition of a-1,4-linked glucose units to a growing gly- cogen 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 [1]. GS is highly regulated by both covalent phosphorylation and allo- steric 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 [2]. Glc6P is an allosteric activator, which also favours the covalent activation of GS through its dephosphorylation [3]. 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 [4] and rab- bit MGS [5]. Glycogen and GS do not show uniform intracellular distribution. LGS accumulates near the plasma mem- brane when cultured hepatocytes are incubated with glucose [6,7]. Consequently, the deposits of the polysac- charide grow from the periphery towards the interior of the cells [8]. In hepatocytes, glycogen synthesis is not homogeneous within the cell, and the spatial distri- bution of LGS is regulated. Regarding muscle, the pro- tein resulting from fusion of human muscle glycogen Keywords glycogen; green fluorescent protein; muscle glycogen synthase; nucleocytoplasmic translocation; glucose 6-phosphate Correspondence J. J. Guinovart, Institut de Recerca Biome ` dica de Barcelona-Parc Cientı ´ fic de Barcelona, c ⁄ Josep Samitier, 4-5, E-08028 Barcelona, Spain Fax: +34 934037114 Tel: +34 934037163 E-mail: guinovart@pcb.ub.es *Present address The Rockefeller University, New York, NY, USA (Received 18 November 2004, revised 24 March 2005, accepted 25 April 2005) doi:10.1111/j.1742-4658.2005.04738.x 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 encom- passes 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. Abbreviations DMEM, Dulbecco’s modified Eagle’s medium; FBS, fetal bovine serum; GS, glycogen synthase; LGS, liver glycogen synthase; MGS, muscle glycogen synthase; HsMGS, human muscle glycogen synthase; RnLGS, rat liver glycogen synthase; GFP, green fluorescent protein; NES, nuclear export signal; NLS, nuclear localization signal; PML, promyelocytic leukaemia oncoprotein. FEBS Journal 272 (2005) 3197–3213 ª 2005 FEBS 3197 synthase (HsMGS) with green fluorescent protein (GFP) shows a regulated nucleocytoplasmic distribu- tion [9,10]. When cells expressing this chimeric protein were incubated in medium containing glucose, the GFP–HsMGS fusion was found throughout the cyto- plasm 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 pres- ence 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. Results 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 C 2 C 12 , 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 subcel- lular distribution of HsMGS in cells that endogen- ously 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 adeno- virus, and the cellular lysates were analyzed by west- ern 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 corres- ponds 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 super- natant 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 rab- bit MGS [13]. Only the immunoprecipitate obtained in the presence of the MGS3 antibody showed a band corresponding to the GFP–HsMGS fusion pro- tein (Fig. 1B). C AB 250 160 105 75 50 Fig. 1. Characterization of the MGS3 antibody. (A) Human cultured myotubes were infected with the pCMV–GFP (left lane) or pCMV–GFP–HsMGS (right lane) adenovirus, at a multiplicity of infection of 10, and the cellular lysates were analyzed by western blot using the MGS3 antibody at a 1 : 1000 dilution. (B) COS-1 cells were transfected with the pEGFP–HsMGS plasmid, and the cellular lysates were subjected to immunoprecipitation in the absence (left lane) or presence (right lane) of the MGS3 antibody. The immuno- precipitates were analyzed by western blot using a different MGS antibody. (C) Confocal microscopy images of L6 myoblasts infected with the pCMV-GFP–HsMGS adenovirus incubated with or without 30 m M glucose, as indicated. Panels labelled with GFP show the green fluorescence of the GFP–HsMGS chimera. Panels labelled with MGS3 show the immunodetection of MGS with MGS3 (1 : 1000 dilution) as primary antibody and a secondary antibody of anti-rabbit IgG (1 : 200 dilution) conjugated to Texas Red. Panels labelled ‘Merge’ show the superimposition of the green and red images. The bar represents 20 lm. Nucleocytoplasmic translocation of muscle glycogen synthase E. Cid et al. 3198 FEBS Journal 272 (2005) 3197–3213 ª 2005 FEBS Finally, the specificity of the MGS3 antibody in the immunolocalization of MGS was analyzed. L6 myo- blasts 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 micro- scope. 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 myo- blasts incubated with glucose, which do not present any nuclear fluorescence, the enzyme forms characteris- tic glycogen-bound aggregates [10] 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 com- partment the GFP-fused enzyme presents both a nucleo- plasmic 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 immuno- detection 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 recog- nizes MGS and that, with some limitations, it is useful to perform immunocytochemistry. Next, we proceeded to analyze the intracellular dis- tribution of endogenous HsMGS in cells that natur- ally 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 glu- cose-deprived myotubes were further incubated for 1 h with 100 lm forskolin, a compound that pro- motes 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 nucleo- plasma, 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 cyto- plasm, 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 Fig. 2. MGS immunolocalization in primary cultured human muscle cells. Differentiated human myotubes were incubated for 16 h in DMEM without glucose and then maintained in medium with (A) or without (B) 30 m M glucose for an additional 6 h. Cells in (C) were deprived of glucose for 22 h and further incubated for 1 h with 100 l M forskolin in the absence of glucose. Immunocytochemistry was performed, using MGS3 as primary antibody, and confocal sections were taken as indicated in Experimental procedures. Arrows indicate the position of the nuclei. The bar represents 10 lm. E. Cid et al. Nucleocytoplasmic translocation of muscle glycogen synthase FEBS Journal 272 (2005) 3197–3213 ª 2005 FEBS 3199 lower concentrations of glycogen than primary cul- tured 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 endo- genous 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 obser- vations 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 pro- tein. 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 lm forskolin, to attain a range of glycogen con- centrations. 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 experi- mental 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 pro- tein CRM1 ⁄ exportin 1, an evolutionarily conserved receptor for the nuclear export signal (NES) of pro- teins [14]. To analyze whether HsMGS was exported from the nucleus through this pathway, L6 myoblasts were infected with an adenovirus that drives the over- expression of GFP–HsMGS and, 32 h after infection, were incubated with Dulbecco’s modified Eagle’s med- ium (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 trea- ted in a medium containing 30 mm glucose, in the presence or absence of 100 nm leptomycin B. In con- trast 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 homo- geneous nuclear distribution in those treated with glu- cose 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, indi- cating 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 lm 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 per- formed deletion analysis and site-directed mutagenesis Fig. 3. Correlation between nuclear concentration of MGS and cellular glycogen content. COS-1 cells transiently expressing GFP–HsMGS were cultured in several conditions (see text) to attain a range of glycogen concentrations. Glycogen content is expressed as a percentage of the highest value obtained in three independent experiments. Nuclear concentration of the GFP–HsMGS fusion pro- tein was evaluated by confocal fluorescence microscopy. For every point, 100 or more cells were counted and the result is expressed as the percentage of cells that showed some degree of nuclear fluorescence. Nucleocytoplasmic translocation of muscle glycogen synthase E. Cid et al. 3200 FEBS Journal 272 (2005) 3197–3213 ª 2005 FEBS on the GFP–HsMGS fusion protein. Several plasmids encoding HsMGS fragments fused to GFP were con- structed (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 incu- bated 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, HsMGSDCT1 (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 HsMGSDCT1 mutant was found in the nucleus of some cells incubated in a glu- cose-free medium containing leptomycin B to block nuclear export (Fig. 5H). Furthermore, when the CT1 fragment was exchanged for its homologous counter- part 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 conclu- sion is reinforced by the observation that the truncated protein HsMGSDCT2, 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 trans- located 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, HsMGSDCT3 and HsMGSDCT4, 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 phos- phorylation sites. Finally, the HsMGSDNT1 fragment, which, in con- trast with HsMGSDCT2, 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 addi- tion, 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 [2] 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 Fig. 4. Inhibition of HsMGS nuclear export by leptomycin B. L6 myoblasts were infected with the AdCMV–GFP–HsMGS adenovirus and 32 h after infection cells were incubated for 16 h in a medium devoid of glucose. L6 myoblasts were then treated for 6 h in DMEM (A), in DMEM with 30 m M glucose (B) or in DMEM with 30 mM glucose containing 100 nm leptomycin B (C). The bar represents 20 lm. E. Cid et al. Nucleocytoplasmic translocation of muscle glycogen synthase FEBS Journal 272 (2005) 3197–3213 ª 2005 FEBS 3201 A BC D FG H I J KL MN E Fig. 5. Intracellular distribution of HsMGS fragments in COS-1 cells. COS-1 cells were transfected with the constructs indicated in (A). Cells overexpressing the GFP-fused chimeras were incubated in the presence or absence of 30 m M glucose and in some cases with 100 n M leptomycin B, as indica- ted. The columns labelled ‘GFP’ show the intracellular localization of the chimeras in green, and the ‘GFP + Hoechst’ columns show the nuclear counterstaining with Hoechst-33342 in red, superimposed on the GFP fluorescence. The bar represents 50 lm. Nucleocytoplasmic translocation of muscle glycogen synthase E. Cid et al. 3202 FEBS Journal 272 (2005) 3197–3213 ª 2005 FEBS 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 [17]. However, no differences in their subcellular locali- zation 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 [10], and, in the absence of the monosaccharide, all phosphoryla- tion mutants were concentrated in the nucleus and formed discrete particles. 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 there- fore the activity of the protein in response to Glc6P con- centration [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 resi- dues 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 catalyti- cally active and showed a very high –Glc6P ⁄ +Glc6P activity ratio (Table 1), which is consistent with the results obtained for the rabbit muscle enzyme [5]. Analy- sis 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 concen- tration was not as consistent and repetitive as that of the wild-type enzyme (Fig. 7C). Furthermore, time- A B Fig. 6. Phosphorylation sites in HsMGS and its mutated forms. (A) The grey bar repre- sents HsMGS with the relative positions of the known phosphorylation sites indicated by filled circles and vertical lines. The local amino-acid sequences, the classical designa- tion of the phosphorylation sites (under- lined), as well as the residue number are indicated in the middle part of the figure. The phosphorylation mutants constructed in this study are shown in the lower part of the figure, indicating in each case the Ser residues that were replaced by Ala residues. (B) Confocal images of COS-1 cells transi- ently expressing a fusion of GFP to the 22a3abc451ab mutant of HsMGS, in which all known serine residues that undergo reversible phosphorylation have been replaced by alanine residues. Cells were incubated in the absence or presence of 30 m M glucose, as indicated. The columns labelled ‘GFP’ show the intracellular localiza- tion of the chimera in green, and the ‘GFP + Hoechst’ columns show the nuclear count- erstaining with Hoechst-33342 in red, super- imposed on the GFP fluorescence. The bar represents 50 lm. E. Cid et al. Nucleocytoplasmic translocation of muscle glycogen synthase FEBS Journal 272 (2005) 3197–3213 ª 2005 FEBS 3203 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 transi- ently expressing the GFP–HsMGS construct were first incubated in the absence of glucose to elicit nuclear accumulation of the protein, and then with A BC E D FG Fig. 7. NES, NLS and Glc6P desensitized mutants of HsMGS. (A) The relative positions of the putative NLS and NES sequences, and the two Arg-rich regions (see text) are indicated as vertical insertions in the bar that represents HsMGS. The local amino-acid sequences, the residue number of the mutated amino acid and the designation of the mutated forms are also shown. (B) COS-1 cells overexpressing the wild-type (left two columns) and R1 ⁄ R2 mutant (right two columns) forms of HsMGS fused to GFP were incubated in the absence of glucose (B,C), in the presence of 30 m M glucose (D,E) or with 30 mM glucose plus 100 nM leptomycin B (F,G), as indicated. The columns labelled ‘GFP’ show the intracellular localization of the chimeras in green, and the ‘GFP + Hoechst’ columns show the nuclear counterstain- ing with Hoechst-33342 in red, superimposed on the GFP fluorescence. The bar represents 50 lm. Nucleocytoplasmic translocation of muscle glycogen synthase E. Cid et al. 3204 FEBS Journal 272 (2005) 3197–3213 ª 2005 FEBS 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 nuc- leus to the cytoplasm is blocked by leptomycin B, indi- cating 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 [18]. Three putative NESs were identified by visual inspection of the protein sequence, and the key hydro- phobic 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 con- formation, thus precluding any conclusions to be drawn. Apart from the Arg-rich sequences mentioned ear- lier, 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 immuno- stained 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 [19], and antibodies against the promyelocytic leukaemia oncoprotein (PML) [20], SC-35 [21] and p80-coilin [22]. Human primary cultured myotubes were treated, as described previously, to achieve HsMGS nuclear localization, fixed, and subjected to double immuno- labelling. For p80-coilin immunostaining, COS-1 cells transfected with the GFP–Hs MGS 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 pro- tein. Myotubes that did not show nuclear staining with MGS3 had the typical 10–30 PML bodies per nucleus [23], 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 fluores- cent signal of the antibody against p80-coilin (Fig. 8B). In many cases, the formation of subnuclear com- partments 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 Table 1. Activity ratio of selected GFP–HsMGS point mutants expressed in COS-1 cells. COS-1 cells were transfected by the DEAE-dextran method and incubated for 42 h in DMEM supple- mented with 30 m M glucose and 10% FBS. Cells overexpressing the indicated mutant protein were then collected and homogenized. GS activity was measured in the absence and of 6.6 m M Glc6P,as indicated in Experimental procedures, and the ratio of these two activities was calculated. Mutant GS activity [mUÆ(mg protein) )1 ] Activity ratio (– ⁄ +Glc6P) –Glc6P +Glc6P GFP–HsMGS 20.7 51.3 0.40 1a 47.3 104.4 0.45 1b 21.2 54.4 0.39 22a 31.3 67.7 0.46 3abc 34.4 60.3 0.57 45 39.4 96.0 0.41 22a3abc45 36.5 74.6 0.49 22a3abc451ab 30.7 53.9 0.57 R1 22.8 34.8 0.66 R2 60.0 52.9 1.1 R1 ⁄ R2 32.4 36.1 0.9 E. Cid et al. Nucleocytoplasmic translocation of muscle glycogen synthase FEBS Journal 272 (2005) 3197–3213 ª 2005 FEBS 3205 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 lm forskolin, to ensure nuclear localization of the fusion protein. Actinomycin D was then added at 10 lgÆmL )1 , a concentration that inhibits all RNA polymerases, and time-lapse images of the cells main- tained 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 treat- ments 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. Discussion Nucleocytoplasmic translocation of MGS In this study we show that, as a result of glucose depri- vation and depletion of the glycogen reserves, MGS is concentrated and aggregates in the nuclear compart- ment of cells that naturally express this protein, such as cultured myotubes and adipocytes. When glucose is added back, the enzyme is translocated to the cyto- plasm, 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. Fig. 8. Co-localization of HsMGS with markers of subnuclear com- partments. (A) Differentiated human myotubes were incubated as in Fig. 1C to ensure nuclear concentration of endogenous HsMGS. Cells were then subjected to double immunolabelling, using MGS3 and aPML as primary antibodies to detect HsMGS (green) and PML (red), respectively. (B) COS-1 cells transiently expressing the GFP–HsMGS fusion protein, in green, were incubated for 22 h in a medium devoid of glucose, fixed and immunolabelled with an anti- body against p80-coilin, in red. The bar represents 10 lm. Fig. 9. Time-lapse microscopy of MGS in cultured human muscle cells. Differentiated human myotubes were infected with the AdCMV– GFP–HsMGS adenovirus and treated as in Fig. 8A to achieve nuclear concentration of the GFP–HsMGS fusion protein. Myotubes were then incubated in DMEM with 10 lgÆmL )1 actinomycin D to inhibit all RNA polymerases. Confocal images were taken after 0, 2 and 4 h. The bar represents 20 lm. Nucleocytoplasmic translocation of muscle glycogen synthase E. Cid et al. 3206 FEBS Journal 272 (2005) 3197–3213 ª 2005 FEBS [...]... 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. .. 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 [2] that control its activity As indicated above, the HsMGSDCT2, DCT3 and DCT4 deletion mutants, which lack some or all of the C-terminal phosphorylation... However, they highlight the relevance of the fragment comprising amino acids 553–633, which includes the Arg-rich cluster mentioned above, in the nuclear 3207 Nucleocytoplasmic translocation of muscle glycogen synthase 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. .. 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... absence of 6.6 mm Glc6P as described previously [39] 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... Guinovart JJ (1997) The role of glucose 6-phosphate in the control of glycogen synthase FASEB J 11, 544–558 4 Pederson BA, Cheng C, Wilson WA & Roach PJ (2000) Regulation of glycogen synthase Identification of residues involved in regulation by the allosteric ligand glucose-6-P and by phosphorylation J Biol Chem 275, 27753–27761 3211 Nucleocytoplasmic translocation of muscle glycogen synthase 5 Hanashiro... accumulation of MGS by other authors [24] 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 [24] Mechanism of nuclear import ⁄ export of HsMGS The regulation of MGS nucleocytoplasmic. .. apoptosis [30] 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 [23] Here we show that the nuclear accumulation of FEBS Journal 272 (2005) 3197–3213 ª 2005 FEBS E Cid et al Nucleocytoplasmic translocation of muscle glycogen synthase HsMGS-triggered glycogen. .. 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 Experimental procedures Plasmid and adenovirus construction Standard molecular cloning techniques were used throughout [32] The pEGFP–HsMGS construct... 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 . GS, glycogen synthase; LGS, liver glycogen synthase; MGS, muscle glycogen synthase; HsMGS, human muscle glycogen synthase; RnLGS, rat liver glycogen synthase; . Regarding muscle, the pro- tein resulting from fusion of human muscle glycogen Keywords glycogen; green fluorescent protein; muscle glycogen synthase; nucleocytoplasmic translocation;