Cellular mechanisms regulating protein phosphatase-1. A key functional interaction between inhibitor-2 and the type 1 protein phosphatase catalytic subunit.

Inhibitor-1 (I-1) and inhibitor-2 (I-2) selectively inhibit type 1 protein serine/threonine phosphatases (PP1). To define the molecular basis for PP1 inhibition by I-1 and I-2 charged-to-alanine substitutions in the Saccharomyces cerevisiae, PP1 catalytic subunit (GLC7), were analyzed. Two PP1 mutants, E53A/E55A and K165A/E166A/K167A, showed reduced sensitivity to I-2 when compared with wild-type PP1. Both mutants were effectively inhibited by I-1. Two-hybrid analysis and coprecipitation or pull-down assays established that wild-type and mutant PP1 catalytic subunits bound I-2 in an identical manner and suggested a role for the mutated amino acids in enzyme inhibition. Inhibition of wild-type and mutant PP1 enzymes by full-length I-2(1-204), I-2(1-114), and I-2(36-204) indicated that the mutant enzymes were impaired in their interaction with the N-terminal 35 amino acids of I-2. Site-directed mutagenesis of amino acids near the N terminus of I-2 and competition for PP1 binding by a synthetic peptide encompassing an I-2 N-terminal sequence suggested that a PP1 domain composed of amino acids Glu-53, Glu-55, Asp-165, Glu-166, and Lys-167 interacts with the N terminus of I-2. This defined a novel regulatory interaction between I-2 and PP1 that determines I-2 potency and perhaps selectivity as a PP1 inhibitor.

Type 1 protein phosphatases (PP1) 1 are major serine/threonine phosphatases in all eukaryotic cells and are characterized by their unique sensitivity to the two protein inhibitors, inhibitor-1 (I-1) and inhibitor-2 (I-2). Inhibition by these mammalian proteins has become a fundamental criterion for classifying type 1 phosphatases from many different species (1,2). I-1 and I-2 inhibit PP1 from yeast (3), flies (4), and humans (5) in an identical manner. The unique sensitivity of these enzymes to I-1 and I-2 most likely reflects the high structural conservation of PP1 catalytic subunits through evolution.
I-1 and I-2 have been used to define of role of PP1 in cell division (6,7), ion gating (8), gene expression (9,10), and neuronal plasticity (11,12). Despite their common properties as PP1 inhibitors, I-1 and I-2 share no primary sequence homology. They also differ in their regulation by reversible protein phosphorylation. I-1 inhibits PP1 activity only when it has been phosphorylated on a specific threonine residue by protein kinase A (13). I-2, on the other hand, is a very effective PP1 inhibitor in its unphosphorylated state and forms a stable inactive complex with the PP1 catalytic subunit. The PP1/I-2 complex is reactivated through a complex series of phosphorylations on I-2 that increases phosphatase activity without dissociating PP1 from I-2 (14).
Structure-function analyses of I-1 (13) and its structural homologue, DARPP-32 (15,16), have highlighted a pivotal role for a conserved N-terminal tetrapeptide sequence, KIQF, in PP1 inhibition. A similar sequence is present in many other PP1 regulators and represents a key PP1-binding site (17,18). I-2 binds PP1 through multiple sequences, and one of these contains a KLHY sequence that could approximate to the KIXF PP1 binding motif (16,19). However, this sequence can be eliminated without reducing I-2 activity as a PP1 inhibitor (16,20), and the mechanism for PP1 inhibition by I-2 remains unknown.
Disparities in structure-function of I-1 and I-2 has made it difficult to understand their common function as PP1 inhibitors. In recent studies, we turned our attention to analyzing the PP1 catalytic subunit and defining domains recognized by the protein inhibitors to gain further insights into the cellular mechanisms that regulate PP1 activity. Our initial studies (21) introduced point mutations in the yeast PP1 catalytic subunit and identified residues that abrogated PP1 binding and inhibition by I-1. This work was extended using a truncated as well as a chimeric PP1 catalytic subunit containing C-terminal sequences from the I-1/ I-2-insensitive or type 2 serine/threonine phosphatase, PP2A (22). These studies defined the ␤12-␤13 loop as a critical structure for PP1 inhibition by toxins and I-1 and I-2. To identify additional interactions between PP1 catalytic subunit and I-1 and I-2, in this study we have analyzed PP1 mutants generated by substituting alanines in place of charged amino acids on the surface of PP1 to disrupt its interactions with protein regulators. This has identified a PP1 domain that uniquely defines the ability of I-2 to inhibit this enzyme. Mutant I-2 proteins then showed that the newly identified domain associates with an N-terminal sequence conserved in I-2 in many species. The protein-protein interactions between I-2 and PP1 that contribute to enzyme inhibition are discussed.

MATERIALS AND METHODS
Phosphorylase b was purchased from Calzyme, and phosphorylase kinase was from Life Technologies, Inc. [␥ 32 P]ATP was obtained from NEN Life Science Products. Heparin-Sepharose was purchased from Amersham Pharmacia Biotech. Anti-GAL4 DNA binding domain antibody was obtained from Santa Cruz Biotechnology.
Yeast Two-hybrid Interactions-GLC7 alleles with various chargedto-alanine substitutions (23) were polymerase chain reaction-amplified from pNC160 using Taq or Pfu polymerase, the forward primer, 5ЈCC-GCCATGGGAATGGACTCACAACCAGTTGACG3Ј, and the reverse primer, 5ЈGGCGGATCCGATTTAGGACGTGAATC3Ј. This introduced NcoI and BamHI sites at the 5Ј and 3Ј ends of the GLC7 coding sequence. The GLC7 mutants were subcloned by blunt-end ligation into a unique SrfI site in pBluescript SK(ϩ) using the pCR-Script SK(ϩ) cloning kit (Stratagene, La Jolla, CA). The GLC7 alleles were then excised with BamHI/NcoI and ligated into the BamHI/NcoI sites of pAS1-CYH2 (24). To ensure that no additional sequences were introduced during polymerase chain reaction amplification, all GLC7 subclones were sequenced at the Iowa State Nucleic Acid Facility (Ames, Iowa). All GLC7 constructs were individually transformed into Y187 yeast strains and mated with Y190 yeast containing either pACTII-hI-1 (21) or pACTII-hI-2 (25). Diploid yeast were grown on plates containing the selection media and transferred to liquid culture for quantitative analysis of PP1-inhibitor interaction.
To quantitate the two-hybrid interaction, liquid cultures were grown to A 600 0.6. An aliquot of the culture (1.5 ml) was centrifuged, and the yeast were resuspended in 200 l of Z buffer (100 mM H 3 PO 4 , 10 mM KCl, 1 mM MgSO 4 , pH 7.0). To this, we added 50 l of 0.1% (w/v) SDS and 25 l of chloroform to permeabilize the cells, and the lysate was assayed for ␤-galactosidase activity using O-nitrophenyl-S-D-galactoside as a substrate (26).
Purification of Yeast PP1-Purification of Glc7p was carried out as described previously (21). Briefly, yeast with a mutant GLC7 gene as the sole source of PP1 were grown in 1 liter of yeast extract/peptone/ dextrose media to near saturation (A 600 0.6). The cells were collected by centrifugation at 3,000 ϫ g, resuspended in 10 ml of lysis buffer (50 mM Tris-HCl, pH 7.5 containing 1 mM EDTA and 3 mM ␤-mercaptoethanol), and lysed using either a bead beater (3 ϫ 1 min) or by three passages through a French press (1,000 psi). The lysate was centrifuged at 3,300 ϫ g for 15 min, and the supernatant was adjusted to 70% saturated ammonium sulfate. The suspension was centrifuged at 10,000 ϫ g for 20 min, and the protein pellet was resuspended in 10 ml of lysis buffer and dialyzed overnight against the same buffer. The dialyzed sample was applied to heparin-Sepharose. The column was washed with 3 volumes of lysis buffer followed by 5 volumes of 50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 3 mM ␤-mercaptoethanol, and 50 mM NaCl. PP1 was eluted using 50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 500 mM NaCl, 3 mM ␤-mercaptoethanol, and 20% (v/v) glycerol. PP1 activity was assayed using phosphorylase a as substrate. Fractions containing enzyme activity were pooled and dialyzed against 50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM dithiotheitol, and 50% (v/v) glycerol for storage at Ϫ80°C. Recombinant rabbit I-2 (14,20), human I-2 (16) and human I-1 (27) were produced as described previously.
Protein Phosphatase Assay-Glc7p was assayed by a modified protein phosphatase assay. The enzyme was incubated with phosphorylase a (2 mg/ml) for 20 min at 30°C. The assay was terminated by the addition of 200 l of 20% (w/v) trichloroacetic acid and 50 l of bovine serum albumin (10 mg/ml). The protein suspension was centrifuged, and [ 32 P]phosphate released into the supernatant was measured by liquid scintillation counting. In assays with the inhibitor proteins, I-1 was added immediately before assaying for PP1 activity. I-2, on the other hand, was preincubated with the phosphatase for 20 min at 30°C before addition of the substrate, phosphorylase a.
Coprecipitation or Pull-down Assays-Equivalent amounts of purified PP1 catalytic subunits (defined by phosphorylase phosphatase activity) in Tris-buffered saline were incubated with human I-2 coupled to CNBractivated Sepharose. After rocking with the I-2 beads for 1 h at 4°C, the beads were removed by centrifugation and washed twice with equal volumes of Tris-buffered saline. The supernatant and washes were pooled and assayed for remaining phosphorylase phosphatase activity.
Western Immunoblotting-Yeast cultures expressing Glc7p mutants fused to the GAL4 DNA binding domain were grown to log phase, centrifuged, and lysed. The cytoplasmic fraction (10 g of total protein) was subjected to 10% SDS-polyacrylamide gel electrophoresis and electrophoretically transferred to a polyvinylidene difluoride membrane. The Glc7p-GAL4 fusion protein was detected with an antibody against the GAL4 DNA binding domain (Santa Cruz). The antibodies were visualized by the ECL reaction (Amersham Pharmacia Biotech).

RESULTS
Mutagenesis of the type 1 protein phosphatase gene (GLC7) has highlighted the multifunctional role of PP1 in yeast (28) and generated mutant phosphatases that have been used to study PP1 regulation by mammalian proteins. Previously, we analyzed chemically mutagenized Glc7p and identified the ␤12-␤13 loop as an essential determinant of PP1 inhibition by I-1 and other phosphatase inhibitors (21). In the present study, we investigated 19 additional mutations in Glc7p produced by charged-to-alanine substitutions (23). A subset of these mutant enzymes was selected for detailed biochemical studies of PP1 inhibition by I-1 and I-2. This selection was based on three criteria. First, the mutated residues were conserved in all PP1 catalytic subunits from yeast to man but not present in type 2 phosphatases, which are insensitive to I-1 and I-2. Second, the mutated residues were located on the surface of the PP1 catalytic subunit as judged by the crystallography of PP1␣ (29), and third, the mutant alleles complimented a GLC7 deficiency that leads to lethality in yeast (23), thereby indicating that the mutant genes encoded active PP1 catalytic subunits.
Inhibition of Glc7p Mutants by I-1 and I-2-Nine mutated Glc7p enzymes were extensively purified and analyzed for their activity and regulation by I-1 and I-2 (Table I). Eight mutants were stable and active phosphorylase phosphatases, with specific activities indistinguishable from WT Glc7p. The D137A/ E138A mutant lost significant activity during purification. This instability may be consistent with its temperature-sensitive phenotype in yeast (23). Inhibition of seven PP1 mutants by I-1 was similar to that seen with WT PP1 (Table I and Fig. 1A), with IC 50 values of 2 nM. The only exception was D165A/ E166A/K167A, which showed a slightly increased IC 50 for I-1 between 8 and 10 nM.
Several Glc7p mutants showed significant changes in their inhibition by I-2. Although WT Glc7p was inhibited by I-2 with an IC 50 of 3 nM (Fig. 1B), the E53A/E55A mutant was inhibited with an IC 50 of 25 nM. D165A/E166A/K167A showed an even greater loss in its sensitivity to I-2, with an IC 50 close to 1 M. Thus, the surface domain defined by these two adjacent sets of charged residues may be particularly important for PP1 regulation by I-2.  (23) were analyzed for their biochemical properties. The individual alleles, substituted amino acids, measured protein phosphatase activity using phosphorylase a as substrate, and sensitivity to the two mammalian inhibitors, I-1 and I-2, are summarized below. The IC 50 values represent an average of at least three independent experiments with a standard error of less than 5%. Association of Yeast PP1 Mutants with I-1 and I-2-To address whether the reduced sensitivity to I-2 reflected changes in I-2 binding to the mutant PP1 catalytic subunits, we undertook a yeast two-hybrid assay with human I-2 as bait. All yeast phosphatases, mutant and wild type, effectively bound human I-2 ( Fig. 2A). The E53A/E55A mutant showed a slightly increased association with I-2, whereas D165A/E166A/K167A showed a modest reduction in I-2 binding when compared with WT Glc7p. These differences in Glc7p/I-2 interaction were not significant and in large part attributed to variations in expression of the individual GAL4-Glc7p fusion proteins in yeast (Fig. 2B).
Association of WT and mutant PP1 enzymes with human I-2 was also assessed using a coprecipitation or pull-down assay in which equivalent amounts of phosphorylase phosphatase activity was incubated with increasing concentrations of human I-2 coupled to CNBr-activated Sepharose (Fig. 3). As reported in earlier studies for PP1 binding to a fluorescent rabbit I-2 (30), a biphasic curve was obtained for PP1 binding to the immobilized human I-2. Most relevant for these studies, WT and mutant PP1 catalytic subunits associated with I-2 in an essentially identical manner. This suggested that the 10-and 500fold reduction in sensitivity of the mutant PP1 enzymes to I-2 was not simply due to a loss in I-2 binding.
Analysis of Mutant I-2 Proteins-We then turned our attention to defining regions of I-2 that might be recognized by this newly identified regulatory domain in PP1c. Previous studies had suggested that the N terminus of I-2 was particularly important for PP1 inhibition (20). Thus, we analyzed the inhibition of WT and mutant yeast PP1 enzymes by two recombinant rabbit I-2 proteins. Truncation of C-terminal sequences yielded I-2(1-114) that inhibited WT PP1, E53A/E55A, and the D165A/E166A/K167A in a similar manner to full-length I-2. In other words, E53A/E55A and D165A/E166A/K167A were increasingly resistant to inhibition by I-2(1-114) when compared with WT PP1c (Fig. 4A). As seen with mammalian PP1c (20), deletion of N-terminal 35 amino acids that produced I-2(36 -204) markedly increased its IC 50 for inhibition of WT yeast PP1 to nearly 400 nM. I-2(36 -204) inhibited the WT and mutant yeast phosphatases with nearly identical IC 50 values (Fig. 4B). This suggested that the differences in inhibition of WT and mutant Glc7p enzymes by full-length I-2 most likely arose from their differing interactions with the Nterminal 35 amino acids in I-2.
Recent studies identified the tetrapeptide sequence Ile 10 -Lys-Gly-Ile 13 in I-2 as a potential site of interaction with the PP1 catalytic subunit. Mutating two amino acids, Lys 11 and Ile 13 , in human I-2 resulted in a marked right-shift in IC 50 for the inhibition of rabbit skeletal muscle PP1c (16). We analyzed this double mutation in recombinant human I-2 and found that like I-2(36 -204), the Lys 11 Glu/Ile 13 Gly mutant was a less effective PP1 inhibitor. More importantly, I-2(Lys 11 Glu/Ile 13 Glu)

FIG. 1. Inhibition of WT and mutant yeast PP1 by I-1 and I-2.
The activity of WT PP1 (diamonds) and the mutants E53A/E55A (triangles) and D165A/E165A/K167A (squares) was assayed using phosphorylase a as substrate in the presence of increasing concentrations of I-1 (panel A) and I-2 (panel B). PP1 activity is expressed as the percentage of activity measured in the absence of inhibitors. failed to distinguish between WT Glc7p and the two mutants, E53A/E55A and D165A/E166A/K167A, inhibiting all three enzymes in an identical manner (Fig. 5A).

FIG. 2. Two-hybrid interaction of yeast PP1 enzymes with human I-2. Panel
Another way to assess the impact of PP1 mutations on its regulation by I-2 was by competition with a synthetic peptide, I-2(6 -20), which represents a potential PP1-binding site. Previous studies showed that the synthetic peptide, I-2(6 -20) (at concentrations up to 25 M), was not inhibitory toward mammalian PP1c but increased the IC 50 for its inhibition by WT I-2 by more than 20-fold (16). Similarly, 25 M I-2(6 -20) had no effect on the phosphorylase phosphatase activity of WT and mutant yeast PP1 enzymes. However, at this concentration, I-2(6 -20) peptide produced a 20-fold right shift in the inhibition of WT Glc7p by full-length I-2, increasing the IC 50 from 2 to 40 nM (Fig. 5B). In similar assays, I-2(6 -20) had a lesser effect on the inhibition of E53A/E55A by I-2, with only a 2-3fold shift in the IC 50 and no effect on the inhibition of D165A/ E166A/K167A. This suggested that both PP1 mutants were compromised in their association with the I-2 (6 -20) peptide. DISCUSSION Several mammalian protein inhibitors inhibit the type 1 protein serine/threonine phosphatases. I-1 (13), DARPP-32 (15), and CPI-17 (31), in their dephosphorylated state, weakly associate with the PP1 catalytic subunit and are not effective phosphatase inhibitors. After their phosphorylation, all three proteins are converted to potent PP1 inhibitors. In contrast, nanomolar concentrations of I-2 (20), NIPP-1 (32), and I-3/HCG V (33), even in their dephosphorylated state, inhibit PP1 activity. Moreover, I-2 and NIPP-1 form stable inactive complexes with the PP1 catalytic subunit that are reactivated by phosphorylation of the inhibitor proteins. However, reactivation maintains the association of PP1 with both inhibitors (30). This suggests that distinct interactions mediate the stable association and inhibition of PP1c by these protein inhibitors. Substituting clusters of charged residues on the surface of the yeast PP1 catalytic subunit with alanines, we have monitored both the binding and inhibition of mutant phosphatases by I-1 and I-2. This has identified two clusters of charged amino acids whose substitution has only minor effects (0 -4-fold reduction) on PP1c inhibition by human I-1. These mutant enzymes were, however, 20 -500-fold more resistant to inhibition by rabbit and human I-2. This points to a novel regulatory interaction that preferentially defines the inhibition of PP1 by I-2. As the mutated amino acids are unique to type 1 phosphatases, they may also dictate the specificity of I-2 as a PP1 inhibitor.
We have also defined the region of I-2 that most likely mediates the regulatory interaction with the newly identified domain in the PP1 catalytic subunit. N-and C-terminal truncations of I-2 focused our attention on the N-terminal 35 amino acids of I-2. Systematic structure-function analysis of I-2 had previously identified residues 9 -14 as particularly important for PP1 inhibition by nanomolar concentrations of I-2 (16). The combined substitutions, Lys 11 -Glu and Ile 13 -Gly, produced a 600-fold right shift in the IC 50 for inhibition of WT Glc7p by I-2. Inability of E53A/E55A and D165A/E166A/K167A to discriminate between WT I-2 and I-2(Lys 11 Glu/Ile 13 Gly) suggested the charged-to-alanine substitutions modified the recognition site for an I-2 sequence that contains Lys 11 and Ile 13 . This issue was also addressed by competition with a synthetic peptide, I-2 (6 -20), which reduced the efficacy of I-2 as a PP1 inhibitor (16). This peptide was much less effective in modifying the inhibition of mutant PP1 enzymes by I-2, consistent with the idea that the peptide binds to the region modified in the mutant PP1 enzymes. The specificity of this peptide was established by the fact that I-2(6 -20) did not compete for PP1 inhibition by I-1 (data not shown) or DARPP-32 (16). Based on the preferential effect of the charged-to-alanine mutations on PP1 regulation by I-2 and the possible interaction of this PP1 domain with a sequence unique to I-2, we have called the region defined by amino acids Glu-53, Glu-55, Asp-165, Glu-166, and Lys-167 an I-2-preferring or simply "I-2 domain." Human, rabbit, rat (35), and Drosophila I-2 (36,37) are all potent PP1 inhibitors (IC 50 1-2 nM) and contain the N-terminal sequence, PXKGILK (termed Site 1 in Fig. 6B). In contrast, an I-2-related protein from Caenorhabditis elegans (36,37), rat I-2␤ (35), and Drosophila testes I-t (38) lack this sequence (Fig.  6B), and Saccharomyces cerevisiae Glc8 has a modified sequence, MGGILK, at its N terminus (39). Drosophila I-t (38), yeast Glc8 (40), and rat I-2␤ (35) are all weaker PP1 inhibitors, with IC 50 values greater than 200 nM. That rat I-2␤ does not compete with I-2␣ as a PP1 inhibitor suggests that the site 1 interaction increases the affinity of I-2 for PP1c (35). Consistent with this, substituting the N-terminal 20 amino acids in FIG. 6. Modeling the protein-protein interactions of PP1c and I-2. Panel A shows a schematic of mammalian I-2 structure with the N-terminal domain that mediates PP1 inhibition, shown as a shaded box, and the C-terminal sequence not required for PP1 inhibition is marked by hatched bars. The N-terminal sequence, Ile 10 -Lys-Glu-Ile 13 , has been shown in these and other experiments (16) to be essential for PP1 inhibition by low nanomolar concentrations of I-2. The threonine 72 is phosphorylated by glycogen synthase kinase-3 and is conserved in many I-2 and I-2-like proteins (36,37). Panel B shows the alignment of N-terminal sequences in human (hI-2), rabbit (rbI-2), and rat (rtI-2␣) with other I-2-like proteins. Our studies suggest that the conserved sequence, highlighted in a box and designated Site 1 represents a PP1-regulatory site that is only conserved in I-2 proteins that inhibit PP1c at low nanomolar concentrations. A related GILK sequence located at the N terminus in Glc8 is underlined by a bracket. All amino acid identities are shown as white letters in black boxes, and homologous residues are in shaded boxes. Dashes represent gaps introduced to obtain an optimal alignment. Panel C shows the three-dimensional structure of PP1␣ catalytic subunit (a space-filling model displayed using Rasmol 2. Drosophila I-t with 35 residues from the N terminus of human I-2 enhanced its potency as a PP1 inhibitor by more than 20-fold (38). Our data, however, suggest that mutations in PP1c that alter the putative site 1 recognition domain have no apparent effect of I-2 binding. Thus, the site 1 sequence most likely plays a unique role in PP1 inhibition by I-2.
The canonical PP1 binding sequence KIXF binds in a hydrophobic pocket (16) close to the I-2 domain (Fig. 6C). KIXF peptides modeled on DARPP-32 (15,16) and NIPP-1 (41) attenuate PP1 inhibition by I-2. This either suggests that I-2 binds in the hydrophobic pocket, perhaps through a less recognizable sequence, or there is interplay between the hydrophobic pocket and the I-2 domain. One of the amino acids analyzed in this study, Asp-165, lies at the boundary between the hydrophobic pocket and the I-2 domain. Asp-165 cooperates with Arg-64 and Arg-65 in the hydrophobic pocket to create a "favorable electrostatic environment" for the binding of a KIXF peptide (17). This could explain the modest change in sensitivity of the D165A/E166A/K167A mutant to I-1, which absolutely requires the KIQF sequence for PP1 inhibition.
The charged amino acids that attenuate PP1 inhibition by I-2 are separated in the primary structure of PP1c by more than 100 residues, yet lie within 11 angstroms of each other in the three-dimensional structure of PP1␣ (Fig. 6C). Glu-53 and Glu-55 are located at the end of the ␤1 sheet, and the Asp-165/ Glu-166/Lys-167 cluster is in the loop connecting the ␤5 and ␤6 sheets. Both of these structures contribute to the architecture of the PP1 active site that dictates metal binding. It is interesting to note that PP1 inhibition by I-2 converts it to an enzyme that requires added divalent cations for activity (34), suggesting a distortion of the PP1 catalytic center by I-2. In any case, I-2 binds at multiple sites on the PP1 catalytic subunit (19), and only some of these interactions mediate PP1 inhibition. Thus far, two domains, the ␤12-␤13 loop and the I-2 domain, have been shown to define the efficacy of I-2 as a PP1 inhibitor. These domains along with the hydrophobic pocket are roughly aligned along the back the PP1 catalytic subunit (Fig. 6C). At least two models can be proposed to allow I-2(1-114) to associate at one or more of sites and inhibit PP1 activity (Fig. 6, D and E).
In the full-length I-2, Park and DePaoli-Roach (20) identify additional interactions of PP1c with sequences C-terminal to I-2(1-114) that mediate the slow inactivation of PP1c. This inactivation is reversed by glycogen synthase kinase-3-mediated phosphorylation of threonine 72 in I-2 and has been proposed to define the function of I-2 as a PP1 "chaperone" that refolds denatured PP1 catalytic subunits in vitro (5,36,42). A dual role for I-2, as an inhibitor and an "activator" of PP1c, might explain the paradoxical observation that Glc8 both augments and inhibits PP1 functions in budding yeast (40). In any case, the KIXF peptide from NIPP-1 (41) attenuates both PP1 inhibition and inactivation by I-2, suggesting that a subset of the protein-protein interactions that occur between PP1 and I-2 mediate both events. Clearly, more work is required to define the full extent of the interactions that regulate the PP1/I-2 complex in mammalian tissues.
Cellular levels of I-2 (43) and Glc8 (44) fluctuate during cell division in mammalian and yeast cells, respectively. I-2 also shuttles in and out of the nucleus during the mammalian cell division cycle (45), and this shuttling is modulated by the phosphorylation of I-2 on threonine 72. Consequences of altered subcellular distribution and/or I-2 levels for PP1 regulation are unclear, but further structure-function analyses of I-2 may also resolve this issue. In conclusion, our data provide experimental support for the hypothesis that structurally diverse PP1 regulators recognize distinct structural elements on the PP1 catalytic subunit to modulate enzyme activity. By defining the molecular determinants that mediate PP1c regulation by I-2, we should be better equipped to elucidate the physiological functions of different I-2 proteins and gain an improved understanding of the role of I-2 in cell signaling.