| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Afdeling Biochemie, Faculteit Geneeskunde, Katholieke Universiteit Leuven, Leuven, Belgium
ABSTRACT I. INTRODUCTION II. THE STRUCTURE OF PROTEIN PHOSPHATASE-1 A. The Catalytic Subunit B. Protein Interactors of PP1 III. CELL DIVISION AND MEIOSIS A. Reversal of Signaling by Protein Kinase Aurora(-B) 1. Mitotic substrates of aurora(-B) and PP1 2. Aurora(-B) and PP1 in cytokinesis 3. Meiotic substrates of Aurora(-B) and PP1 4. R subunits that target PP1 to Aurora(-B) substrates B. Delay of Centrosome Splitting Until the G2/M Transition C. PP1 at the M/G1 Transition D. Exit From the Pachytene Stage in Yeast Meiosis IV. CELL CYCLE ARREST AND APOPTOSIS V. METABOLISM A. Reversal of Starvation-Induced Metabolic Shifts B. Glycogen Metabolism 1. Glycogen-associated substrates of PP1 2. The mammalian G subunits 3. Control of hepatic glycogen metabolism 4. Glycogen metabolism in skeletal muscle VI. PROTEIN SYNTHESIS A. Transcription B. mRNA Processing C. Translation VII. ACTIN AND ACTOMYOSIN REORGANIZATION A. Neurabin-Associated PP1 B. Mypt-Associated PP1 1. Myosin phosphatase 2. Myosin and Mypt kinases 3. Actomyosin (de)phosphorylation in cytokinesis 4. Regulation of cell shape and cell adhesion by myosin kinases/phosphatase 5. Myosin (de)phosphorylation and muscle contraction C. Scd5-Associated PP1 VIII. RECEPTORS, ION CHANNELS, AND ION PUMPS A. Intracellular Ca2+ Release Channels and Ca2+ Pumps 1. The ryanodine and inositol trisphosphate receptors 2. Sarcoplasmic reticulum Ca2+-ATPase 3. PP1 and cardiac (dys)function B. Transforming Growth Factor-{beta} Receptor-I C. Regulation of Ionotropic Glutamate Receptors D. Regulation of Other Channels and Transporters IX. INHIBITION AND MATURATION OF PROTEIN PHOSPHATASE-1 A. Inhibition 1. The PKA-activated inhibitors 2. The PHIs B. Maturation X. CONCLUSIONS
 |
ABSTRACT |
|---|
 Top NextReferences |
|---|
|
I. INTRODUCTION |
|---|
|
TopPreviousNextReferences |
|---|
100 protein tyrosine kinases and protein tyrosine phosphatases. However, the numbers of protein serine/threonine kinases (400) and protein serine/threonine phosphatases (25) are hugely different (294), and this has been accounted for by distinct diversification strategies during evolution (74). Indeed, while the number of protein kinases has steadily increased during eukaryotic evolution, serine/threonine phosphatases have not flourished to the same extent, but the diversity of their interacting polypeptides has increased enormously. Thus the true diversity of protein serine/threonine phosphatases is only seen at the holoenzyme level and largely stems from the variety of regulators that can interact with a given catalytic subunit. When holoenzymes are considered, protein serine/threonine kinases and phosphatases show a similar diversity. Protein serine/threonine phosphatases are currently divided into three structurally unrelated families. The PPM family comprises Mg2+-dependent enzymes, including protein phosphatase (PP) 2C. The FCP family contains only one member, which is also Mg2+ dependent. All other protein serine/threonine phosphatases are classified in the PPP family, consisting of the subfamilies PP1, PP2A (including PP4 and PP6), PP2B, and PP5, which all have a structurally related core and a similar catalytic mechanism. This review only deals with PP1, in particular with its functions in various cellular processes. Other recent reviews on PP1 have mainly focused on the structure of the enzyme and the diversity of its regulators (3, 33, 48, 74, 86).
|
II. THE STRUCTURE OF PROTEIN PHOSPHATASE-1 |
|---|
|
TopPreviousNextReferences |
|---|
PP1 (35–38 kDa) is one of the most conserved eukaryotic proteins. This is nicely illustrated by the early branching eukaryote Giardia lamblia, which expresses an isoform of PP1 that is 72% identical to the mammalian PP1 isoforms (74). Also, the phenotypes associated with mutations of PP1 in fungi could be (partially) complemented by expression of mammalian PP1 (113, 311), indicating that PP1 is also functionally conserved. Eukaryotic genomes contain one (Saccharomyces cerevisiae) to eight genes (Arabidopsis thaliana) encoding PP1 isoforms. More than 70% of the residues in the central three-quarters of these isoforms are virtually invariant, yet the flanking NH2- and COOH-terminal sequences show more divergence. Mammals have three PP1 genes, encoding the isoforms PP1
, PP1
, and PP1
/
. Two splice variants can be generated from the PP1 gene, PP11 and PP12. With the exception of the testis-enriched PP12, the mammalian isoforms are ubiquitously expressed.
The crystal structure of PP1 shows a compact fold with a central -sandwich that excludes only the COOH terminus and the extreme NH2 terminus (Fig. 1). A number of invariant residues coordinate two metals, presumably Fe2+ and Zn2+, near the front edge of the -sandwich, and these metals are thought to contribute to catalysis by enhancing the nucleophilicity of metal-bound water and the electrophilicity of the phosphorus atom (117, 148). The active site is situated at the bifurcation point of an extended Y-shaped surface depression. The arms of this depression are denoted as the COOH-terminal groove, the acidic groove, and the hydrophobic groove (Fig. 1). Crystallographic studies also suggested the mechanism of inhibition of PP1 by some cell-permeable toxins that are widely used for functional studies. Thus the cyclic heptapeptide microcystin LR interacts with two of the metal-bound water molecules and thereby blocks the binding of substrates to the catalytic site. Furthermore, it interacts with the hydrophobic groove and binds covalently to Cys-273 in the 12-13 loop, which over-hangs the catalytic site. The polyether fatty acid okadaic acid binds to the hydrophobic groove and forms hydrogen bonds with Tyr-272 in the 12-13 loop and with basic residues in the catalytic site (247). Another polyether fatty acid, calyculin A, contains a phosphate group that interacts with the metal binding site, but calyculin A also forms a tight network of interactions with the hydrophilic and acidic grooves (207).
|
The COOH-terminal fragment of PP1 (30 residues) is excluded from the globular structure but contains threonine residues that are phosphorylated in a cell cycle-dependent manner, resulting in a reduced activity of PP1 (see sects. IIIC and IV). It has been suggested that this inhibition is caused by the binding of phosphothreonine at the catalytic site and the interaction of basic residues in the COOH terminus with acidic residues that surround the catalytic site (117).
The catalytic subunits of PP1 do not exist freely in the cell, but they associate with a host of different regulatory (R) polypeptides (Table 1) to form a variety of distinct multimeric holoenzymes. Thus many of the identified interactors of PP1 have been characterized as regulators. For other interactors, such as phosphofructokinase, the retinoblastoma protein, and Sla1, it is not yet clear whether they are regulators and/or substrates of PP1, or whether they bind directly to PP1 or via another interactor. Regulators of PP1 can be divided in primary and secondary regulators (74), according to whether they originated as regulators of PP1 or acquired this PP1 binding function only later in evolution. Primary regulators (e.g., inhibitor-2, NIPP1, and Sds22) typically contain (putative) PP1-binding sites in all eukaryotic lineages where they occur. Secondary regulators (e.g., AKAP149, Nek2, Bcl2), on the other hand, share functional domains with homologs that lack binding sites for PP1, which indicates that these sites were acquired later in evolution by proteins with an originally unrelated function. Some PP1 interactors appear to have evolved late in the evolution of a particular eukaryotic lineage, as no homologs can be identified in other lineages. For example, the PKA-activated inhibitors are vertebrate specific, while some Drosophila (Bifocal, Klp38B) or fungal (Reg1/2, Gip1) regulators have no obvious vertebrate counterparts. The protein interactors of PP1 can also be classified based on their function (48) into substrate-independent activity regulators [e.g., inhibitor-1, dopamine and cAMP-regulated phosphoprotein of 32 kDa (DARPP-32), and inhibitor-2], targetting subunits/substrate specifiers (e.g., G subunits, Mypts) or substrates (Aurora kinases, Nek2). The limitations of the latter classification are that the exact function of many protein interactors is still unknown and that some interactors, e.g., Reg1, function both as a targetting subunit and as a substrate.
|
|
Some regulatory binding sites of PP1 have been mapped (Fig. 1). The best characterized is the so-called "RVXF" binding channel, which is a hydrophobic groove remote from the catalytic site and is formed by the top rear edges of the two central -sheets (118). Most regulators of PP1 contain an RVXF motif, which actually conforms to the consensus sequence [RK]x0–1[VI]{P}[FW], where x can be any residue and {P} refers to any residue but proline (74, 118, 378, 411). Binding of the RVXF motif per se is not associated with major conformational changes of PP1 and does not have significant effects on the catalytic activity. The available data rather suggest that the RVXF motif serves as an anchor for the initial binding of the R subunits to PP1 and thereby promotes, sometimes cooperatively, the binding of secondary sites, which often bind with lower affinity but affect the activity and substrate specificity of PP1 (48, 378). The 12-13 loop forms a second, flexible binding site of PP1, one that is essential for inhibition of PP1 by both toxins (see sect. IIA) and protein inhibitors (inhibitor-1, DARPP-32, inhibitor-2, and NIPP1) (91). Still another interaction site for R subunits is the triangular region delineated by the 4-, 5-, and 6-helices of PP1, which we have recently identified as a major interaction site for Sds22 (75). Finally, an interaction site for the conserved NH2-terminal K-[GS]-I-L-K motif of inhibitor-2 has been mapped near the entrance of the RVXF-binding channel (90). Some R subunits, such as the Neurabins and the Mypts (see sect. VII, A and B), interact with PP1 in an isoform-specific manner, indicating that PP1 also contains isoform-specific regulatory binding sites.
The R subunits bring PP1 in close proximity to its substrates by anchoring the phosphatase in specific cellular compartments via targetting motifs or domains. Some R subunits block the activity of PP1 by acting as pseudosubstrates (see sects. VIIB and IXA) or by inducing conformational changes (see sect. IXB). The substrate-specifying effect of some R subunits (G subunits, Mypts, AKAP149) implies both an increased activity toward some substrates and a decreased activity toward other substrates. The surface of PP1 is relatively open, and no peptide binding cleft is evident, in accordance with its broad substrate specificity (33). One can therefore envisage that the binding of R subunits to PP1 restricts the accessibility of the catalytic site, either by causing steric hindrance or by inducing conformational changes. At least in some instances, the substrate-specifying activity may stem from the fact that the R subunits are themselves substrates (see sects. IIIB and VIC) or have binding sites for specific substrates (see sect. VB).
|
III. CELL DIVISION AND MEIOSIS |
|---|
|
TopPreviousNextReferences |
|---|
A. Reversal of Signaling by Protein Kinase Aurora(-B)
1. Mitotic substrates of aurora(-B) and PP1
Protein kinases of the Aurora family have multiple mitotic substrates (281), and increasing evidence suggests that PP1 reverses the action of these protein kinases. One of these substrates is histone H3 (Fig. 2), which is phosphorylated on Ser-10 by the unique Aurora protein kinase in yeast and the Aurora-B protein kinase in animals (1, 146), and is an established mitotic substrate of PP1 (176, 269). Various studies have reported a correlation between the phosphorylation of histone H3 along chromosomes in G2 and chromosome condensation (146, 176, 375), and also between chromosome decondensation in telophase and PP1 activity or histone H3 dephosphorylation (24, 151, 368). These observations have led to the hypothesis that chromosome (de)condensation requires histone H3 (de)phosphorylation. Accordingly, in fission yeast and in animals, phosphorylation of histone H3 is involved in the recruitment to chromosomes of a component of the heteropentameric condensin complex (Fig. 2), which has been implicated in chromosome condensation (146, 191, 257). However, mutation of Ser-10 of histone H3 did not cause any observable growth defect in budding yeast (176), and neither Ser-10 nor the entire NH2-terminal tail of Xenopus histone H3 is essential for chromosome condensation (100). An alternative hypothesis proposes that a checkpoint labels chromosomes that are ready to go through anaphase and telophase by phosphorylation of histone H3 and that this checkpoint impinges on the balanced activity of Aurora(-B) and PP1 (100). The histone H3 kinase activity of Xenopus Aurora-B depends on the latters' phosphorylation by an unknown kinase, which may well be Aurora-B itself, as its yeast counterpart autophosphorylates (44, 269). Interestingly, this Aurora-B activation is antagonized by PP1 (269), and PP1 interacts physically with Aurora-B (337).
|
Recent complementary work in yeast and in animals suggests that Aurora(-B) and PP1 may also act antagonistically in the complex control of the layered protein interface between the centromeres and the mitotic spindle that ensures biorientation of sister kinetochores, spindle integrity, and chromosome segregation (Fig. 2). First, the histone H3 homolog CENP-A, which substitutes for histone H3 in centromeric nucleosomes, is phosphorylated by Aurora-B at a site similar to that of H3 (407). CENP-A phosphorylation starts in mitotic prophase and decreases in anaphase and appears to be correlated with kinetochore maturation. It remains to be explored whether CENP-A is also a substrate of PP1. Second, yeast PP1 and Aurora control the phosphorylation state of the kinetochore protein Ndc10, which binds directly to the centromere (44, 314). Hyperphosphorylation of Ndc10 impairs the attachment of microtubules to the kinetochore (314). Third, both in yeast and in animals Aurora(-B) phosphorylates the inner centromere protein INCENP (194). Strikingly, the temperature-sensitive mitotic defects of a yeast INCENP mutant are attenuated by overexpression of a dominant-negative truncated version of PP1 (204), in accordance with the proposed antagonism between Aurora(-B) and PP1. Together with Survivin, which interacts with Ndc10 (400), Aurora(-B) and INCENP form the chromosomal passenger complex. Like PP1, this complex has been implicated in chromosome segregation and cytokinesis. The passenger complex migrates from the centromeres to the spindle midzone and the cleavage furrow after the transition to anaphase (1, 47, 387, 407). Interestingly, disruption of the phosphorylation site of CENP-A disturbs the subcellular localization of Aurora(-B), INCENP, and PP1 in the latter half of mitosis (407). Given that PP1 and Aurora(-B) interact physically (337), these findings lead to the enticing hypothesis that PP1 is a component of the chromosomal passenger complex.
The Aurora substrate Dam1 is a component of the multimeric spindle-associated DASH complex that is required for biorientation of sister kinetochores and for mitotic spindle integrity (185, 222). Dam1 binds to Aurora and INCENP (Fig. 2). Interestingly, overexpression of PP1 exacerbates the temperature-sensitive growth defect of dam1 mutant cells, indicating that Dam1 may also be a substrate of PP1 (194).
2. Aurora(-B) and PP1 in cytokinesis
Although considerable progress has been made in deciphering the role of Aurora(-B) and PP1 in spindle integrity and chromosome segregation, their function in cytokinesis remains largely elusive. It is known that both the passenger complex and PP11 are present at the cleavage furrow at the end of the M phase (47, 407). A conditional mutation of PP1 in yeast was associated with various cell cycle defects, including a perturbed cytokinesis, which correlated with the absence of an actin ring at the bud neck (16). Furthermore, functional deficiencies of the passenger complex (1, 190, 221) or microinjection of antisense PP11 oligonucleotides (79) resulted in a severe defect in cytokinesis. Only a single candidate target has thus far been identified, i.e., the regulatory light chain of myosin II, which is an in vitro Aurora-B substrate (267). The regulatory light chain is also a well-established substrate of Mypt-containing holoenzymes of PP1 (see sect. VIIB). However, the latter holoenzymes contain the -isoform of PP1 rather than the 1-isoform (256). Furthermore, a functional depletion of Mypt in Caenorhabditis resulted in a rather mild cytokinetic phenotype with ectopic furrowing and an accelerated furrow ingression (293). Therefore, it is likely that Aurora-B and PP1 share still other substrates that play an important and conserved role in cytokinesis.
3. Meiotic substrates of Aurora(-B) and PP1
In addition to their involvement in the progression of mitosis and cytokinesis, Aurora(-B) and PP1 have also been implicated in meiosis. Caenorhabditis oocytes depleted of Aurora-B (or Survivin) by RNA interference fail to separate homologous chromosomes in meiosis I and sister chromatids in meiosis II (304). It has been proposed that Aurora-B promotes chromosome separation by the phosphorylation of the meiosis-specific cohesin Rec8 and that this phosphorylation results in the cleavage of Rec8 by Separase. Accordingly, Rec8 is an in vitro substrate for Aurora-B (304), and Aurora-B is targetted to the remaining points of contact between separating chromosomes in metaphase I and II (191, 304). Interestingly, the latter subchromosomal regions also exhibit a pronounced phosphorylation of histone H3 on Ser-10 (191). Aurora-B could thus be the meiotic counterpart of the Polo-like kinase, which phosphorylates the mitotic cohesin and thereby marks it for Separase-dependent cleavage (9). Like most other Aurora-B functions, this role as meiotic cohesin kinase appears to be conserved and antagonized by PP1. Thus fission yeast Rec8 is phosphorylated during meiosis I and II (290), and PP1 depletion by RNA interference causes precocious separation of sister chromatids at the onset of anaphase I (191, 304). The latter effect correlated with an increased presence of Aurora-B on meiotic chromosomes and a decrease in the level of chromosomal Rec8 (304). It remains to be studied whether PP1 directly dephosphorylates Rec8 or impinges on the targetting or activity of Aurora-B.
4. R subunits that target PP1 to Aurora(-B) substrates
The R-subunit(s) that are involved in the dephosphorylation of Aurora(-B) substrates by PP1 remain(s) unknown, but Sds22, an established interactor of PP1 in both yeast and mammals (75, 107, 171, 234, 334), seems an attractive candidate. Indeed, the Sds22 encoding gene was identified independently in fission and in budding yeast as an extra-copy suppressor of temperature-sensitive mitotic arrest phenotypes that are associated with particular mutations of PP1 (171, 234, 286). Deletion of the Sds22 encoding gene caused a similar arrest, and this phenotype could be complemented by the overexpression of PP1 (171, 234, 286). Also, the conditionally lethal phenotype in budding yeast that was conferred by a loss-of-function mutation of the yeast Aurora kinase was largely relieved by the expression of certain temperature-sensitive mutant versions of Sds22 or PP1 (136, 291). The mutant Sds22 version that rescued the conditional Aurora phenotype showed a decreased ability to interact with PP1. The expression of this mutant Sds22 did not affect the cellular levels of PP1 or Sds22, but drastically reduced the nuclear level of PP1 and caused a redistribution of the nuclear pool of PP1 (291). Whether Sds22 is also involved in meiosis is not known, but it can certainly not be ruled out as Sds22 has been identified in a ternary complex with the mammalian PP12 isoform (82, 179).
B. Delay of Centrosome Splitting Until the G2/M Transition
Centrosomes duplicate during S phase, but they remain paired and continue to function as a single microtubule-organizing center during G2 (281). Shortly before the onset of mitosis, the duplicated centrosomes separate and form the poles of the bipolar spindle apparatus. At least two kinases that have been implicated in the induction of this separation are inactivated by PP1, which suggests that PP1 may prevent precocious splitting of the centrosomes. One of these is Aurora-A, a homolog of Aurora-B (195). While Aurora-A is required for centrosome separation in animals (147), the unique yeast Aurora kinase does not appear to subserve this function, as a conditional loss-of-function mutation of this enzyme did not affect spindle pole body separation (356). Like Aurora-B, Aurora-A interacts with PP1, and this interaction peaks at mitosis. A mechanism of regulation of Aurora-A by PP1 that is similar to that of Aurora-B (269) is suggested by the observation that PP1 dephosphorylates and thereby inactivates Aurora-A in vitro. Interestingly, PP1 is also an in vitro substrate for Aurora-A, and this phosphorylation results in the inactivation of the phosphatase.
A second kinase involved in centrosome splitting is Nek2, a member of the NIMA family of protein kinases. Nek2 is activated by autophosphorylation and is thought to phosphorylate the centrosomal protein C-Nap1, resulting in the dissolution of the structure that keeps the centrosomes together. The counterplayer of Nek2 is PP1, which dephosphorylates both C-Nap1 and Nek2 itself (165). Furthermore, Nek2, PP1, and C-Nap1 can form a ternary complex in vitro, and Nek2 contains an RVXF motif that is essential for its interaction with PP1 (165). Recently, it was reported that inhibitor-2 also interacts with the Nek2/PP1 complex via PP1 and that the expression of inhibitor-2 increases Nek2 kinase activity and promotes centrosome splitting (124). Conversely, the overexpression of PP1 strongly suppresses Nek2-mediated centrosome splitting (250). Interestingly, a parallel can be drawn between the regulatory relationships of PP1 with Aurora-A and Nek2, as PP1 is also a substrate for the associated Nek2 and phosphorylation of COOH-terminal site(s) reduces its phosphatase activity (165). This suggests that the separation of centrosomes may depend on both the activation of the inducing kinases and the inactivation of associated PP1. In this respect, it is worthy of note that PP1 is also inactivated through phosphorylation by cyclin-dependent protein kinase 1 (Cdk1) in early to mid-mitosis at a COOH-terminal site that is different from the Nek2 and Aurora-A phosphorylation site(s) (195, 213, 226, 297).
The splitting of the centrosomes is accompanied by the recruitment of -tubulin ring complexes, which function as nucleation sites for microtubules (281). The recruitment of these complexes is mediated by AKAP450, which also contains binding sites for a host of different protein kinases and phosphatases, including PP1 (348, 349). The functions of these AKAP450-associated signaling enzymes remain unknown.
PP1 contributes to the reassembly of the nuclear envelope at the end of mitosis by acting as a lamin-B phosphatase (362). Lamin-B is a component of the nuclear lamina, and its phosphorylation at the onset of mitosis leads to the disassembly of the nuclear lamina. More recently, Collas and co-workers (331) showed in an elegant series of experiments that PP1 is targetted to Lamin-B by AKAP149, an integral membrane protein of the endoplasmic reticulum and the nuclear envelope. They also found that the recruitment of PP1 by AKAP149 is a prerequisite for the reassembly of the nuclear lamina and that a failure to recruit PP1 results in apoptosis (330). Human AKAP149 binds PP1 via an RVXF motif and, importantly, also functions as a lamin-B specifying subunit (329a).
The burst of protein dephosphorylation at the M/G1 transition not only involves proteins that function in the execution of mitosis per se, but also affects numerous proteins that play a role in such diverse processes as replication, transcription, pre-mRNA splicing, cell survival, and cell cycle progression (49). One of these is the antiapoptotic protein Bcl-2, an integral membrane protein of the mitochondria and the endoplasmic reticulum (see also sect. IV), which is targetted for proteasome-mediated degradation by dephosphorylation (60). A late-mitotic Bcl-2 phosphatase was biochemically identified as PP1 and, moreover, PP1 was found to coimmunoprecipitate with mitochondrial Bcl-2 during late mitosis. Furthermore, it has been shown that Bcl-2 contains a functional PP1-binding RVXF motif (26). Another late-mitotic substrate of PP1 is the retinoblastoma protein (Rb), which is hyperphosphorylated from the S phase until the end of mitosis. During G1, in contrast, Rb is hypophosphorylated, and this allows sequestration of key stimulators of the G1/S transition transition, such as the E2F transcription factors. PP1 was found to function as the Rb phosphatase in mitotic cell lysates (276), and Rb was shown to bind PP1 in two-hybrid (116) and coprecipitation assays (297, 306, 352). The sensitivity of the Rb phosphatase in intact cells to various cell-permeable cytotoxins also points to PP1 (396).
D. Exit From the Pachytene Stage in Yeast Meiosis
In yeast, a premature exit from the pachytene stage after the initiation of meiotic recombination is prevented by the so-called "pachytene checkpoint" (reviewed in Ref. 303). An active checkpoint results in the phosphorylation and activation of protein kinase Mek1, which keeps its substrate Red1 phosphorylated (29, 102). When recombination has ended in late pachytene, the checkpoint is inactivated by the dephosphorylation of Red1 by PP1 (30). Overexpression of PP1 bypasses the checkpoint precociously.
The nature of the regulatory subunit(s) associated with this meiotic function of PP1 remains unclear. A number of findings originally pointed to Gip1, a PP1-binding protein that is specifically expressed in middle meiosis and that is essential for sporulation (372). Thus it was reported that 1) Gip1 was required for the targetting of PP1 to chromosomes in late pachytene, 2) yeast cells lacking Gip1 displayed a pachytene arrest that was similar to that of cells with constitutively active Mek1 or with a deficient version of PP1, and 3) this arrest was alleviated by overexpression of PP1 (30). However, in a more recent study, deletion of the Gip1-encoding gene was found not to affect meiotic progression, but instead to interfere with the normal localization of sporulation-specific septins and the deposition of spore wall material (347). Strikingly, replacement of PP1 by a mutant version that fails to interact with Gip1 yielded a similar phenotype.
|
IV. CELL CYCLE ARREST AND APOPTOSIS |
|---|
|
TopPreviousNextReferences |
|---|
, which results from the Cdk-mediated phosphorylation on Thr-320. This is strikingly illustrated by the observation that the expression of the constitutively active T320A mutant of PP1, but not that of the wild-type PP1, prevented the Rb phosphorylation in late G1 and caused cell cycle arrest (35). Moreover, expression of the PP1 mutant T320A in Rb-negative cells did not impede cell cycle progression, indicating that this effect on cell cycle progression was Rb dependent. Cell cycle arrest and/or apoptosis induced by genotoxins is also correlated with a dephosphorylation of the Rb protein (115, 122, 211, 274, 382). Under these conditions, Rb dephosphorylation is accounted for by a decreased activity of Cdks and by an activation of PP1 via the dephosphorylation of the inhibitory COOH-terminal Cdk site (49, 150). PP1 inhibitors such as calyculin A or inhibitor-2 prevent the induction of cell cycle arrest and apoptosis, which underlines the crucial role of PP1 in this cellular response to stress (115, 382). It has recently been demonstrated that both PP1 and the p70 S6 kinase interact with the vitamin D receptor (37). However, the p70 S6 kinase was only recruited in its phosphorylated form and in the absence of ligand. The binding of PP1 was ligand independent, but PP1 activity increased in a ligand-dependent manner. Ligand-activated PP1 was shown to dephosphorylate p70 S6 kinase, resulting in the inactivation of the kinase and its dissociation from the receptor/phosphatase complex. Because p70 S6 kinase is essential for the G1/S transition, it was argued that its inactivation by PP1/PP2A contributes to the vitamin D-induced cell cycle arrest.
The Bcl-2, Bcl-xL, and Bcl-w proteins have mainly been described as positive regulators of cell survival, but in conjunction with a dephosphorylated form of the proapoptotic protein Bad, they can also induce apoptosis via the activation of proteases of the caspase family. Bcl-2/xL/w contain a PP1-binding RVXF motif, and they can occur in a ternary complex with PP1 and Bad (25–27). Furthermore, the dephosphorylation of Bad and the apoptosis induced by interleukin deprivation from hematopoietic cells were both alleviated by the inhibition of PP1. Combined with the observation that the Bad phosphatase is mainly associated with Bcl-2/xL/w, these data suggest that the Bcl-2/xL/w proteins target Bad for dephosphorylation by PP1.
Recently, PP1 has also been implicated in the ceramide-induced shift of the splicing pattern of the Bcl-x and caspase 9-encoding genes, causing these genes to produce the proapoptotic splice variants Bcl-xS and caspase 9 rather than the antiapoptotic variants Bcl-xL and caspase 9b (77). Increased ceramide levels are thought to induce the dephosphorylation by PP1 of splicing factors of the SR family, which are indeed involved in the regulation of alternative splicing (76).
The COOH-terminal half of another interactor of Bcl-2, ASPP2, contains a PP1-binding RVXF motif of its own. However, the binding of PP1 and Bcl2 to ASPP2 are mutually exclusive (163, 273). ASPP2 and its RVXF-containing homolog ASPP1 also bind to p53 and thereby specifically stimulate the transactivation of proapoptotic genes by p53 (310), but addition of PP1 dissociated p53 from the COOH-terminal half of ASPP2 (163), leaving the function of the PP1-ASSP interaction unknown.
|
V. METABOLISM |
|---|
|
TopPreviousNextReferences |
|---|
An ancient eukaryotic response to nutrient starvation and hypoxia (reviewed in Refs. 72 and 202) is orchestrated by conserved trimeric protein kinases that consist of a catalytic -subunit, a substrate-defining and targeting -subunit (249, 317), and a regulatory -subunit. These protein kinases, termed Snf1 in yeast and AMP-activated kinase (AMPK) in animals, have a common mechanism of regulation by reversible phosphorylation. Glucose deprivation and other stress factors bring about phosphorylation of the -subunit on a conserved threonine residue by an upstream kinase (72). Subsequently, the -subunit binds to the autoinhibitory domain of the -subunit and thereby activates the catalytic domain. Activated Snf1/AMPK in turn promotes 1) glucose import; 2) gluconeogenesis; 3) respiration; 4) the use of alternative sugars and other carbon sources like fatty acids, ethanol, glycerol, pyruvate, and lactate; and 5) the downregulation of anabolic pathways (72, 158). These effects are achieved via direct phosphorylation or transcriptional control of key metabolic enzymes. Two mechanisms are involved in the transcriptional control: phosphorylation-dependent nuclear exclusion of transcriptional repressors (106) and phosphorylation at specific promotors of serine-10 of histone H3, which facilitates acetylation of lysine-14 and transcription (230).
The phosphatase that reverts the -subunit of AMPK to its inactive state in vivo remains unknown, but Snf1 is dephosphorylated by a PP1 holoenzyme. Two noncatalytic subunits have been identified in this PP1 complex, namely, Reg1 and Sip5 (371). The RVXF-containing Reg1 binds constitutively to PP1 and to Sip5 (Fig. 3), and this ternary complex is targetted to the activated Snf1 at limiting glucose concentrations (313). The Snf1 kinase then phosphorylates Reg1. The phosphorylation of Reg1 is antagonized by Reg1-associated PP1, but at low glucose concentrations, the balance is tipped in favor of a net phosphorylation of Reg1 by the hexokinase Hxk2, which is itself phosphorylated on Ser-15 in these conditons (299). Hxk2 interacts (weakly) with both Snf1 and Reg1, but it is not clear yet whether Hxk2 promotes the phosphorylation of Reg1 by the stimulation of Snf1 and/or by the inhibition of the associated PP1 (313). When the availability of glucose increases, an hitherto unidentified trigger promotes the net dephosphorylation of Snf1 by the associated PP1 complex, resulting in the release of the latter complex from Snf1 (313). Phosphorylation of Reg1 is a prerequisite for the dephosphorylation of Snf1 and for the release of the phosphatase complex from the kinase complex. Indeed, deletion of the Hxk2 encoding gene, which is associated with a hypophosphorylation of Reg1, renders the Snf1 complex constitutively active. Also, increased glucose levels fail to dissociate Reg1 from a genetically inactivated Snf1. Interestingly, Hxk2 is also dephosphorylated by PP1 in a Reg1- and glucose-dependent manner (13). After its release from the Snf1 complex, PP1 rapidly reverts Reg1 to its unphosphorylated state.
|
By the dephosphorylation of Snf1, Reg1-associated PP1 reestablishes glucose as the preferred source of energy and reinitiates anabolic pathways. Because PP1 is also a well-established histone H3 phosphatase (see sect. IIIA1), it may also downregulate Snf1 signaling by directly dephosphorylating Snf1 substrates, such as histone H3. As histone H3 phosphorylation on Ser-10 has also been proposed as a mechanism for the regulation of transcription by other histone H3 kinases, such as Msk1 (363), this would add to the importance of transcriptional repression as a function for PP1. In Caenorhabditis too, PP1 was found to antagonize more than one histone H3 kinase (191).
One of the effects of glucose-induced Snf1 inactivation is a loss of maltose permease activity, both by transcriptional repression and by a posttranscriptional mechanism termed glucose(-induced) inhibition (177). Ubiquitin-mediated proteolysis of maltose permease constitutes a third, Snf1-independent mechanism for the downregulation of maltose import. More recently, a novel function in the proteolysis of maltose permease has been proposed for Reg1 and for the distantly related PP1 interactor Reg2, which is not involved in Snf1-mediated signaling (186). Reg1- and Reg2-associated PP1 have been suggested to promote the proteolysis of maltose permease, possibly via the regulation of an as of yet unidentified maltose permease kinase.
The study of glycogen metabolism has contributed enormously to our understanding of the structure and regulation of PP1 (53, 54, 85, 180). For example, these studies have led to the concepts that R subunits of PP1 function as targetting and substrate-specifying subunits and that the activity of PP1 is largely controlled by phosphorylation and allosteric regulation of its R subunits. Also, to this day, glycogen phosphorylase is by far the most widely used substrate for the assay of PP1 in vitro.
The rate-limiting enzymes of glycogen synthesis and breakdown are glycogen synthase and phosphorylase, respectively. The phosphorylation of glycogen synthase is generally associated with an inactivation of the enzyme, whereas phosphorylase is activated by phosphorylation. A host of protein kinases phosphorylate multiple residues in the extremities of glycogen synthase, but phosphorylase is only phosphorylated on one NH2-terminal serine by a single protein kinase, namely, phosphorylase kinase. The latter is itself activated by phosphorylation of its regulatory - and -subunits and by the binding of Ca2+ to the regulatory -subunit, which is identical to calmodulin. Glycogen synthase, phosphorylase, and, to a lesser extent, phosphorylase kinase are bound to the glycogen particles, and their dephosphorylation is believed to be (partially) mediated by species of PP1 that are anchored to the glycogen particles via glycogen-targetting subunits (G subunits). The dephosphorylation of the rate-limiting enzymes of glycogen metabolism by PP1 results in the storage of glycogen, in accordance with the proposed function of PP1 as an energy conserving enzyme (see sect. X).
1. Glycogen-associated substrates of PP1
Overwhelming evidence implicates PP1 in the activation of glycogen synthase in vivo. In budding yeast, mutants of PP1 with a reduced affinity for the G subunit Gac1 or loss-of-function mutants of Gac1 were glycogen deficient and had a low activity of glycogen synthase (298, 336). The additional deletion of Pig1, one of the three other yeast G subunits, exacerbated the glycogen deficiency of a Gac1 null strain (80). Conversely, a higher expression level of Gac1 was associated with an increased activity of glycogen synthase. In rat hepatocytes, a loss of glycogen-associated PP1, as seen for example in insulin-dependent diabetes, was associated with an impaired glucose-induced activation of glycogen synthase (54). On the other hand, a superactivation of glycogen synthase was noted in liver cells from rats with hyperthyroidism, which showed an increased level of glycogen-associated PP1 (217). Furthermore, the overexpression of various G subunits in cultured cells correlated with an activation of glycogen synthase (144, 219, 395). Finally, mice lacking the skeletal muscle specific GM subunit had a low basal activity of glycogen synthase and showed a deficient exercise-induced activation of the enzyme (22).
PP1 is also likely to function as a phosphorylase phosphatase in vivo, since alterations in the expression levels of PP1 or of specific G subunits resulted in corresponding changes in the activity of phosphorylase (22, 84, 144, 395). However, activated phosphorylase (phosphorylase a) is also known as an excellent in vitro substrate for PP2A, and it cannot be ruled out that the latter also contributes significantly to the dephosphorylation of phosphorylase in vivo. There is no information available on the protein phosphatases that dephosphorylate phosphorylase kinase in vivo. It should be noted, however, that only the -subunit of phosphorylase kinase is an in vitro substrate for dephosphorylation by PP1.
Mammalian genomes contain no less than seven genes that encode G subunits, but only four of these have been characterized at the protein level (74). For the purpose of conformity, we suggest that the G subunits are differentiated by a capital subscript, which refers to their gene name, except for GM and GL where the subscript refers to the tissues (striated muscle and liver, respectively) where they are expressed most abundantly (Table 2). With the exception of GM and GL, the G subunits are expressed ubiquitously, albeit at variable levels (20, 110, 215). GL displays a remarkable species-dependent distribution in that it is absent from sketal muscle of rats while it is highly expressed in human skeletal muscle (263).
Two modules are conserved in all G subunits, i.e., a PP1-binding RVXF motif and a targetting module with binding sites for glycogen and PP1 substrates (21, 80, 134, 228, 392, 393). The binding of the G subunits to both PP1 and its substrates seems to be required as the disruption of either binding site abolished the activation of glycogen synthase that is associated with the expression of the GM or GC subunits in cultured cells (228, 393). Interestingly, the binding of glycogen synthase to GM was reported to be modulated by phosphorylation of glycogen synthase (228). The better characterized GM and GL subunits have also been shown to contain a COOH-terminal domain that is involved in the binding to phospholamban in the membranes of the sarcoplasmic reticulum (see sect. VIIA2) and to the allosteric inhibitor phosphorylase a, respectively. Furthermore, GM and GL modulate the substrate specificity of PP1 in that they inhibit the dephosphorylation of phosphorylase but increase the specific glycogen synthase phosphatase activity (6, 110, 180). Likewise, the GC subunit decreases the phosphorylase phosphatase activity of PP1 (111).
3. Control of hepatic glycogen metabolism
The liver functions as a glucose sensor, and hepatic glycogen metabolism contributes to the control of the blood glucose homeostasis (53). A postprandial rise in blood glucose results in the inactivation of phosphorylase and activation of glycogen synthase. Conversely, when glucose levels drop below a given threshold, phosphorylase is activated and glycogen synthase is inactivated. The glucose-induced inactivation of phosphorylase is at least in part explained by the binding of glucose to phosphorylase a, which turns the latter into a better substrate for dephosphorylation (53). The glucose-induced activation of glycogen synthase is associated with a translocation of glycogen synthase to the cell periphery (142) and may be partially mediated by a phosphatidylinositol 3-kinase-dependent signaling pathway (209, 217). However, glucose also elicits hepatic synthase phosphatase activity, both by the removal of the allosteric inhibitor phosphorylase a and by the generation of the stimulator glucose 6-phosphate (66, 328). Glucose 6-phosphate probably acts via an allosteric increase of the substrate quality of glycogen synthase. Phosphorylase a is only inhibitory to GL-associated PP1, the major glycogen-associated synthase phosphatase (63). A glucose-induced activation of glycogen synthase by the latter phosphatase is also in accordance with reports that the loss of GL in the liver of diabetic or adrenalectomized and starved rats (63, 109) is associated with an impaired activation of glycogen synthase by glucose (51, 52). Moreover, the restoration of the GL level by the administration of insulin or by refeeding closely correlates with an improved activation of glycogen synthase.
It seems unlikely that GL-associated PP1 also contributes to the glucose-induced inactivation of phosphorylase in vivo, since the loss of GL in diabetic and in adrenalectomized starved rats did not hamper the inactivation of phosphorylase by glucose in hepatocytes (51, 52) and had but a moderate effect on the glycogen-associated phosphorylase phosphatase activity (54, 109). Moreover, GL is inhibitory to the phosphorylase phosphatase activity of associated PP1, and the allosteric GL inhibitor phosphorylase a does not affect its own dephosphorylation by hepatic protein phosphatases (21, 110). It thus appears that the control of glycogen synthase and phosphorylase by glucose is mediated by different protein phosphatases. The nature of the G subunit(s) that targets PP1 to phosphorylase in the liver remains to be explored. The loss of the GC subunit from the diabetic liver, which retains its ability to inactivate phosphorylase in response to a glucose load, argues against an involvement of GC in this process (63). A role for the GD subunit is also unlikely given that this protein is only expressed at very low levels in the liver (53). Perhaps one of the poorly characterized G subunits (GE, GF, or GG) is involved in the targetting of PP1 to phosphorylase. Alternatively, hepatic phosphorylase a may be dephosphorylated by species of PP1 (or PP2A) that are not or only transiently associated with glycogen.
In view of the contribution of the G subunit(s) to the hepatic uptake of glucose, it has been proposed that the therapeutic expression of (fragments of) these proteins may serve to lower blood glucose in diabetes (279). One benefit of (over)expressing G subunits rather than glycolytic enzymes is that they are not expected to increase the circulating level of lipids. As further support for their proposal, Newgard and co-workers (398) noted that the (over)expression of G subunits in cultured hepatocytes or in rat liver stimulated glycogen deposition, albeit with different sensitivity to glycogenolytic agents and potency. For example, the hepatic overexpression of GC resulted in a 70% increase in the hepatic storage of glycogen and an improved whole body glucose homeostasis, but the deposited glycogen was not broken down during fasting. However, the glycogen pool that was synthesized as a result of the expression of a truncated version of GM was responsive to glycogenolytic stimuli, and the expresssion of this GM fragment moreover normalized glucose tolerance in rats on a high-fat diet (143). Whether the (over) expression of (fragments of) G subunits will ever be used in the treatment of diabetes will of course depend on the availability of suitable gene delivery methods. Drugs that promote the functions of endogenous G subunits would constitute an alternative strategy. For example, a drug that alleviates the allosteric inhibition of GL by phosphorylase a would have great potential in this respect (21).
4. Glycogen metabolism in skeletal muscle
Glycogen in skeletal muscle serves as a source of energy to sustain contractions. In the period following exercise the glycogen stores are replenished, which correlates with an increased glucose uptake and activation of glycogen synthase. Mice lacking GM had a very low basal activity of glycogen synthase and an increased level of phosphorylase a (340). Conversely, GM-overexpressing mice showed an increased activity of muscle glycogen synthase, but their phosphorylase activity was not affected (22). Importantly, the GM null mice failed to activate glycogen synthase following exercise or electrically induced muscle contraction. These data clearly show that GM is essential for the regulation of glycogen synthase under basal conditions and in response to contractile activity. The mechanism by which muscle contraction affects GM remains to be explored.
Epinephrine promotes glycogenolysis and inhibits glycogen synthesis in skeletal muscle. This hormone elevates cAMP and activates protein kinase A (PKA), which in turn promotes glycogenolysis via the phosphorylation of phosphorylase kinase. In addition, PKA functions as a glycogen synthase kinase, and it has been shown to dissociate and thereby inactivate the GM/PP1 complex via the phosphorylation of Ser-67 of GM (380), which occupies position X of the RVXF motif. PKA also phosphorylates Ser-48 of GM, which increases the synthase phosphatase activity of GM-associated PP1. However, the phosphorylation of Ser-67 has an overriding effect since it disrupts the holoenzyme structure. It has been suggested that the phosphorylation of Ser-48 serves as a mechanism for maximal glycogen synthesis in the recovery period after adrenergic stimulation, when PP1 reassociates with GM (380).
Insulin is a major stimulator of glycogen synthesis in skeletal muscle, in particular in the postprandial phase (340). This insulin effect is partially accounted for by signaling via protein kinase B, which results in the inhibition of glycogen synthase kinase 3 (GSK-3). In addition, insulin activates a glycogen-associated species of PP1 that dephosphorylates glycogen synthase (101, 340). Analysis of two, independently generated GM null mice led to different conclusions as to the role of GM in the insulin-mediated control of glycogen metabolism. Suzuki et al. (340) concluded that the GM-PP1 complex is unlikely to mediate the control of glycogen synthesis by insulin since their GM null mice were lean, glucose tolerant, and still responded normally to insulin with an activation of glycogen synthase. In contrast, the GM knock-out mice that were generated by Delibegovic et al. (101) were obese, glucose intolerant, and insulin resistant, suggesting a key role for the GM protein in the metabolic control by insulin. The latter group also reported that insulin still caused a mild activation of glycogen synthase in the GM-deficient mice and that this activation was correlated with a stimulation of the GC-PP1 complex, a response that was not seen in the wild-type animals.
|
VI. PROTEIN SYNTHESIS |
|---|
|
TopPreviousNextReferences |
|---|
The transcription of protein-encoding genes by RNA polymerase II relies on the reversible multisite phosphorylation of heptapeptide repeats in the COOH-terminal domain (CTD) of the largest subunit of the polymerase. Phosphorylation of the CTD domain by the cyclin-dependent protein kinases Cdk7 and Cdk9 is needed for promotor clearance, transcriptional elongation, and recruitment of mRNA processing factors, while its dephosphorylation is required for the regeneration of initiation-competent RNA polymerase II. Although originally the phospho-serine phosphatase FCP1 had been characterized as the CTD phosphatase, recent studies have suggested that PP1 may also contribute to the dephosphorylation of the CTD domain (383). Thus CTD dephosphorylation in cultured cells was inhibited by okadaic acid, which blocks PP1 but does not affect FCP1. Moreover, PP1 was shown to act as a major CTD phosphatase in nuclear extracts and to affinity-purify with RNA polymerase II. Both PP1 and the nuclear regulator NIPP1 have also been identified as components of the Tat-associated RNA polymerase II complex, which regulates transcription from the human immunodeficiency virus type 1 promoter (40, 275).
The transcription factor CREB mediates the expression of cAMP-induced genes by binding to a conserved cAMP-responsive element. Phosphorylation of CREB on Ser-133, e.g., by PKA, promotes the recruitment of the histone acetyltransferase CBP, which facilitates access of the promoter region to the transcriptional machinery. Attenuation of CREB signaling results from the dephosphorylation of CREB by PP1 (5, 45, 154), but the involved targetting subunit is unknown. Interestingly, it was recently reported that the histone deacetylase HDAC1 is part of a CREB-associated complex that also includes PP1 and that promotes the dephosphorylation of CREB (69). The importance of PP1 as a CREB phosphatase is illustrated by the finding that brain-targetted genetic inhibition of PP1 in mice correlated with an enhanced learning capability that involved the hyperphosphorylation of a number of proteins, including CREB (145) (see also sect. VIIIC). In normal conditions, two additional serines that control the stability of CREB are kept dephosphorylated by PP1 (359). However, the decreased expression of PP1 following hypoxia results in the hyperphosphorylation and subsequent ubiquitin-mediated degradation of CREB.
The heat shock factor (HSF) is a key transcriptional activator of stress-inducible genes and is activated by phosphorylation (174). The yeast homolog interacts with the G subunit Gac1 (224), suggesting that Gac1 may contribute to the recovery from stress by promoting the PP1-mediated dephosphorylation of HSF.
Initial evidence also implicates PP1 in the regulation of chromatin remodeling. Indeed, PP1 was identified as an antagonistic regulator of the trithorax protein in Drosophila, a component of a protein complex that is required for the maintenance of normal expression of homeotic genes (307). We have recently found that the nuclear PP1 interactor NIPP1 also binds to Eed (186a) which, as a member of the Polycomb group proteins, acts antagonistically to the trithorax protein and maintains transcriptional repression of homeotic genes by histone deacetylation and methylation. Moreover, like Eed, NIPP1 functioned as a transcriptional repressor of targetted genes. NIPP1, Eed, and PP1 can form a ternary complex, suggesting that NIPP1 targets Eed or an Eed-associated protein for dephosphorylation by PP1. Another trimeric complex that is presumably involved in chromatin remodeling consists of PP1, GADD34, and the SNF5 protein (391). GADD34 is a stress-induced protein that facilitates cell cycle arrest, while SNF5 is a component of a SWI/SNF chromatin remodeling complex that acts by repositioning nucleosomes. Both SNF5 and GADD34 interact directly with PP1, and SNF5 functions as a positive regulator of the GADD34/PP1 complex in vitro. GADD34 binds to PP1 via a canonical RVXF motif that is also required for binding of SNF5. Nevertheless, SNF5 does not compete with PP1 for the same binding site on GADD34. By analogy with the established targetting function of GADD34 in translation (see sect. VIC), these data may reflect that GADD34 promotes the dephosphorylation of a component of a SWI/SNF remodelling complex by PP1.
At least four PP1 interactors are established splicing factors or are known to colocalize with splicing factors. One of these is the splicing factor PSF, which is involved in the second catalytic step of splicing (169) but has recently also been implicated in the control of transcription and gene silencing (319). The nuclear PP1 targetting subunits PNUTS (11) and SNP70 (93) show a punctate nuclear distribution typical for splicing factors and coprecipitate with spliceosomes during splicing in nuclear extracts (M. Lloriam, M. Beullens, I. Andrés, J.-M. Ortiz, and M. Bollen, unpublished observations). The PP1 interactor NIPP1 contains a forkhead-associated domain that is associated with phosphorylated forms of the splicing factors Cdc5L (57), SAP155 (58), and Melk, a novel protein kinase of the AMP-activated kinase family that blocks pre-mRNA splicing in nuclear extracts (V. Vulsteke, M. Beullens, and M. Bollen, unpublished data). We have recently identified NIPP1 as a splicing factor that is involved in a late step of spliceosome assembly but, surprisingly, this function appears to be unrelated to its ability to bind PP1 (38).
The inhibition of PP1 does not affect pre-mRNA splicing in nuclear extracts (38), indicating that PP1 is not involved in spliceosome assembly and splicing catalysis per se. Initial evidence suggests that PP1 may be involved in alternative 5'-splice site selection, possibly by dephosphorylating splicing factors of the SR family (71, 76, 77) (see also sect. IV). PP1 has also been proposed to contribute to spliceosome disassembly and/or the shuttling of splicing factors from the spliceosomes to the splicing factor compartments, which are thought to be storage or assembly sites for splicing factor complexes (252, 270). Consistent with this proposal, it was reported that the addition of either PP1 or inhibitors of PP1 to permeabilized cells interferes with the subnuclear distribution of splicing factors (253).
The RNA binding protein Staufen is a component of ribonucleoprotein complexes that move bidirectionally along dendritic microtubules and are believed to represent local storage compartments for mRNAs under translational arrest (210). Recently, Staufen has been identified as an RVXF-containing interactor of PP1, indicating that the Staufen complexes may be subject to regulation by reversible protein phosphorylation and that this regulation involves PP1 (255).
The phosphorylated form of the eukaryotic translation initiation factor eIF2 integrates general translational repression with the induction of stress-responsive genes in various stress conditions (284, 391). Phosphorylated eIF2 inhibits the assembly of translation initiation complexes by sequestering eIF2B, a translation factor that is needed for the regeneration of GTP-bound eIF2. Paradoxically, phosphorylated eIF2 enhances the translation of the activating-transcription-factor-4, which is involved in the induction of stress-responsive genes. At least four structurally related protein kinases, each activated by specific stress stimuli, phosphorylate eIF2. One of these is protein kinase R, which is activated by autophosphorylation and dimerization following the binding of double-stranded RNA. Recently, it was reported that protein kinase R binds directly to PP1 and is dephosphorylated by associated PP1 (354). In accordance with protein kinase R being a physiological substrate of PP1, these investigators found that the activation of the kinase is prevented by the coexpression of PP1.
Considerable evidence indicates that PP1 also contributes to the cellular recovery from stress by acting as an eIF2 phosphatase. PP1 was first identified as an eIF2 phosphatase in reticulocyte lysates (121). In yeast, a genetic antagonism was established between PP1 and an eIF2 kinase (385). More recently, the hyperphosphorylation of mammalian eIF2 during stress has been linked to a reduction in a PP1-derived eIF2 phosphatase activity (262). The targetting of PP1 to eIF2 appears to be mediated by GADD34 (92, 284). Indeed, the GADD34/PP1 complex is an efficient eIF2 phosphatase in vitro, and the level of phospho-eIF2 is diminished in GADD34-overexpressing cells. Remarkably, a heterotrimeric complex was identified consisting of GADD34, inhibitor-1, and PP1 (92). In this complex, GADD34 interacted directly with PP1 and with inhibitor-1 regardless of the latters' phosphorylation state, but PP1 only interacted with inhibitor-1 when the latter was phosphorylated by PKA. Hence, it was argued that the earlier described PKA-mediated inhibition of translational initiation in reticulocyte lysates could potentially be explained by the phosphorylation of inhibitor-1, resulting in the inhibition of the GADD34/PP1/inhibitor-1 complex and a hyperphosphorylation of eIF2. Connor et al. (92) also reported that the GADD34/PP1 and GADD34/inhibitor-1 interactions in the brain of ground squirrels were lost during hibernation, and this correlated with a hyperphosphorylation of eIF2
