PP1 is the only known serine/threonine phosphatase that dephosphorylates GS and GP and changes its activation state [4]. The PP1 catalytic subunit (PP1c) is targeted to the glycogen particles by glycogen-associated regulatory subunits, PP1-GTSs. PP1-GTSs constitute a family of proteins including PPP1R3A, PPP1R3B, PPP1R3C, PPP1R3D and PPP1R3E with low amino acid identity, which is characterized by its conserved PP1c binding motif (the RVXF motif) [5], a glycogen-binding domain [6,7] and a PP1-substrate binding domain [7]. PP1-GTSs display tissue- and specie-specific expression. In human skeletal muscle, several members of the PP1-GTS gene family are expressed: the muscle-specific PPP1R3A/G_M/PP1G/R_GL [8,9]; PPP1R3C/PTG/PPP1R5 [9,10] and PPP1R3B/G_L [11], which are expressed mostly in muscle and liver; the relatively widespread isoform PPP1R3D/PPP1R6 [9]; and PPP1R3E, most abundant in skeletal muscle and heart tissue in humans [7]. In contrast, in rodent skeletal muscle, the PPP1R3B [11] and PPP1R3E [7] genes are not significantly expressed.

The human PPP1R6 subunit was identified by Armstrong et al. [9] in the search for sequences related to the rat G_L and human PPP1R5 proteins in an expressed sequence tag database. These authors demonstrated that the PPP1R6 protein specifically binds to purified PP1 and is associated with glycogen in the glycogen-sarcovesicle fraction from rabbit skeletal muscle. They detected PPP1R6 transcripts in all human tissues examined including brain, placenta, lung, liver, kidney and pancreas. The highest levels were found in skeletal muscle and heart. In a later study, the PPP1R6 gene transcript was detected in several human cancer cell lines, including small cell and non-small cell lung cancers, colorectal, gastric and ovarian cancers, and adult normal lung, colorectal and ovarian tissues, but not normal gastric tissue [12]. Only one previous study has demonstrated a metabolic functional role of PPP1R6 in stimulating glycogen accumulation in Chinese hamster ovary cells stably transformed with the insulin receptor [13].

In contrast, the metabolic control role of G_M and PTG in the muscle glycogen metabolism has been amply described. PP1G/R_GL/G_M regulates basal GS and GP activity ratio in skeletal muscle, as observed in G_M/R_GL knockout mice [14,15] and is essential for GS activation in response to exercise [14], but not necessary for activation of GS by insulin [15]. In cultured human myotubes, G_M also regulates both GS and GP activities and the effect of activating GS is enhanced in glucose/glycogen-depleted cells [16]. PTG regulates basal GS activation, is not required for insulin stimulation of GS and does not affect GP activity in skeletal muscle of mice bearing a heterozygous deletion of the PTG gene [17]. In cultured human myotubes, PTG has a powerful activating effect on GS, which is not glucose-dependent, and on glycogenesis; whereas it does not affect GP activity [18]. In this cell system, the effects of G_L have also been reported, i.e. that G_L activates GS, regardless of glucose, and glycogenesis more powerfully than G_M, and does not affect the GP activation state [19].

Even though PP1-GTSs are presumed to contribute to organizing the glycogen particle [2,3], the ability to target glycogen particles to a specific subcellular location has only been suggested for G_M/R_GL, as it contains a sarcoplasmic reticulum (SR) binding domain [8,20] and is found associated with non-junctional SR membranes, forming a complex with GS and PP1c in rabbit skeletal muscle [21].

In cultured myotubes, PPP1R6 activated GS and reduced its phosphorylation at Ser-641/0, also referred to as site 3a, in a less powerful manner than PTG, but stronger than G_M. GS may be phosphorylated at nine or more sites [39], but the Ser-641/0 site is crucial for the regulation of GS activity and is extensively regulated by various effectors [4]. The selectivity of PP1 for the different sites of GS phosphorylation has not been established, but it has been proposed that PP1-GTSs may confer such selectivity [4]. Our data provide evidence that the degree of dephosphorylation of the Ser-641/0 site achieved in response to PPP1R6, PTG or G_M is tightly associated with that of GS activation, thereby suggesting a similar strong control of enzyme activity through the modulation of the phosphorylation of this site by these PP1-GTSs. The activating effect of PPP1R6 on GS was enhanced by glucose/glycogen depletion, as also observed for G_M [16] but not PTG [18]. In fact, glycogen content is a strong regulator of GS and high glycogen content impairs insulin-stimulated activation of this enzyme, as reviewed in [4]. PPP1R6 did not exert any effect on the cultured myotube GP activity, either the endogenously expressed isoforms, which are predominantly the brain (PYGB) and liver (PYGL) isoforms [32], or the overexpressed muscle (PYGM) isoform. Like PPP1R6, neither does PTG affect endogenous myotube GP activity [18], whereas G_M inactivates GP [16]. As PPP1R6 markedly activated GS in myotubes, it exerted a potent glycogenic effect. This was seen by the increases in the rate of glycogen synthesis; by an increase in the net glycogen content, as quantified after KOH extraction and ethanol precipitation; and cytochemically, by an increase in the glycogen stained as blue fluorescent PA-SH material.

Glycogen accumulation in response to PPP1R6 was less than that achieved with PTG but higher than with G_M. Cellular glycogen is complexed with proteins, including glycogen metabolizing enzymes, PP1 and PP1-GTSs [3], in the form of particles or granules, also named glycosomes [1,2]. The morphology of these granules was originally defined by Drochmans [40] as α-, β- and γ-types. The α particles, commonly found in mammalian liver, are agglomerates of β-particles, have a rosette appearance and large diameters [1], in the order of 100 nm [40]. Single β-particles are found in most normal tissues, including skeletal muscle [1], have a spheroid shape and diameters ranging from 15 to 40 nm. The γ-particles are about 3 nm in diameter and are subunits of β- and α-particles [1]. Glycogen particle sizes in skeletal muscle cells have been consistently shown to follow a continuous distribution pattern. In human skeletal muscle, diameters range from 10 to 44 nm and a mean value of 25 nm at rest [41] and from 8 to 43 nm and a mean particle size of about 13 nm after exercise [42]. In cultured rat myotubes, these particles range from above 15 to above 40 nm (mean value of 29.4 nm) in glycogen-replete cells or from above 10 to below 40 nm (mean value of 24.9 nm) in glycogen-depleted ones [43]. Strikingly, here we show that, despite being in a continuous range, the size distribution pattern of glycogen particles is determined by PP1-GTS scaffolding, with average diameters in the order of PPP1R6 < G_M < PTG. In control myotubes most glycogen particles corresponded to the β-type, with a mean diameter of 29.2 nm. No particles smaller than about 20 nm were observed, but a few above 42 nm were found. Notably, the detection limit for small glycogen granules is estimated to be 5-10 nm (at 20000 × magnification) [42], while the predicted maximum size limit of a granule is 42 nm, consisting of 12 tiers of carbohydrate [3]. PPP1R6 generated small granules, with a mean diameter of 14.4 nm, with most of the granules falling below 20 nm and many between 7 and 15 nm. These granules resembled the γ-type in the formation of filamentous alignments [40]. G_M-derived glycogen particles showed an intermediate mean diameter of 28.3 nm and most were between 20 and 40 nm, which corresponds to the β-particle classification. PTG produced the largest particles, with a mean diameter of 36.9 nm and values ranging from 25 to 50 nm. Many of these particles looked like β-type ones, but we observed a few supramolecular clusters with a rosette-like shape that resembled α-particles. In fact, the larger α-type glycogen particles have not been observed in mammalian tissues other than liver [40]. Since PTG is expressed abundantly in liver, in rats [44] and humans [9], we speculate that it induces the formation of α-particles in this tissue. In this regard, β particles are suggested to be covalently bonded to form α-particles through a hitherto unsuspected enzyme process operative in the liver [45], which could be tissue-dependent.

The number of glycogen particles was increased by the three PP1-GTSs tested. PPP1R6 and PTG induced the greatest increases in particle numbers and quantified glycogen accumulation. The rise in particle numbers per cell area for PPP1R6 was higher than with PTG, although quantification of extracted or cytochemically stained glycogen was not greater, probably because PTG-derived particles have bigger size and tend to form agglomerations of single β-particles. G_M caused the smallest increase in the number of particles per cell area, which fits with glycogen quantification data. In fact, early muscle glycogen resynthesis has been shown to also rely on the synthesis of new particles. Elsner et al. [43] suggested that during early glycogen resynthesis in fasted cultured myotubes, new glycogen molecules are formed while there is a modest increase in glycogen particle diameter. Graham and coworkers [42] showed that, during recovery from prolonged exercise, resynthesis of muscle glycogen in humans is characterized initially by an increase in number and no change in particle size and later by an increase in particle size but not in number.

The subcellular distribution of glycogen-particles promoted by either PP1-GTS has distinct patterns. In PPP1R6-myotubes, granules were closely associated with cellular components. In G_M-myotubes, granules accumulated near the endoplasmic reticulum, which is consistent with G_M association with SR via an SR-binding sequence in its C-terminal region [8,20,21]; however, granules also appeared in other locations, preferentially subsarcolemmal ones. In fact, two forms of G_M, an SR-associated form and a cytosolic glycogen-associated form, can be distinguished in rat skeletal muscle extracts [46]. In contrast, PTG caused the accumulation of glycogen granules in large clusters located in central areas of the cytosol devoid of cellular organelles. In control cells, glycogen granules were distributed throughout the cytosol. Since in control myotubes the PPP1R3C gene is the most expressed gene (of PPP1R3A, PPP1R3B, PPP1R3C and PPP1R6 genes) and PTG is strongly glycogenic, it can be argued that glycogen particles in these cells are mainly PTG-derived, which may explain their preferential cytosolic location. Smaller granules were observed in control myotubes than in PTG-overexpressing cells, probably due to much lower glycogen accumulation in the former. We cannot rule out, however, a role for the still unstudied PPP1R3E protein. In the skeletal muscle tissue of humans [3,41,47,48] and rats [49], electron microscopy studies have shown compartmental distribution of glycogen particles in three main regions: subsarcolemmal, intermyofibrillar and intramyofibrillar. In the subsarcolemmal location, clustered accumulations are observed at the level of mitochondria and SR [47], and nuclei and mitochondria [41]. Aneurally cultured human myotubes are not fully differentiated, and likely this influences the pattern of cellular glycogen distribution. However, with PPP1R6 and G_M we observed clustering of glycogen granules to the membrane of myotube organelles.