Characterization of the Inhibition of Protein Phosphatase-1 by DARPP-32 and Inhibitor-2*

Phospho-DARPP-32 (where DARPP-32 is dopamine- and cAMP-regulated phosphoprotein, M r 32,000), its homolog, phospho-inhibitor-1, and inhibitor-2 are potent inhibitors (IC50 ∼1 nm) of the catalytic subunit of protein phosphatase-1 (PP1). Our previous studies have indicated that a region encompassing residues 6–11 (RKKIQF) and phospho-Thr-34, of phospho-DARPP-32, interacts with PP1. However, little is known about specific regions of inhibitor-2 that interact with PP1. We have now characterized in detail the interaction of phospho-DARPP-32 and inhibitor-2 with PP1. Mutagenesis studies indicate that within DARPP-32 Phe-11 and Ile-9 play critical roles, with Lys-7 playing a lesser role in inhibition of PP1. Pro-33 and Pro-35 are also important, as is the number of amino acids between residues 7 and 11 and phospho-Thr-34. For inhibitor-2, deletion of amino acids 1–8 (I2-(9–204)) or 100–204 (I2-(1–99)) had little effect on the ability of the mutant proteins to inhibit PP1. Further deletion of residues 9–13 (I2-(14–204)) resulted in a large decrease in inhibitory potency (IC50 ∼800 nm), whereas further COOH-terminal deletion (I2-(1–84)) caused a moderate decrease in inhibitory potency (IC50∼10 nm). Within residues 9–13 (PIKGI), mutagenesis indicated that Ile-10, Lys-11, and Ile-13 play critical roles. The peptide I2-(6–20) antagonized the inhibition of PP-1 by inhibitor-2 but had no effect on inhibition by phospho-DARPP-32. In contrast, the peptide D32-(6–38) antagonized the inhibition of PP1 by phospho-DARPP-32, inhibitor-2, and I2-(1–120) but not I2-(85–204). These results indicate that distinct amino acid motifs contained within the NH2 termini of phospho-DARPP-32 (KKIQF, where italics indicate important residues) and inhibitor-2 (IKGI) are critical for inhibition of PP1. Moreover, residues 14–84 of inhibitor-2 and residues 6–38 of phospho-DARPP-32 share elements that are important for interaction with PP1.

Protein phosphatase-1 (PP1) 1 is a major eukaryotic protein serine/threonine phosphatase that regulates diverse cellular processes such as cell cycle progression, protein synthesis, muscle contraction, carbohydrate metabolism, transcription, and neuronal signaling (1)(2)(3)(4). The catalytic subunit of PP1 is regulated by the heat-stable protein inhibitors, inhibitor-1, its homolog DARPP-32 (dopamine-and cAMP-regulated phosphoprotein, M r 32,000), and inhibitor-2 (1,2). Phosphorylation of inhibitor-1 at Thr-35 or of DARPP-32 at Thr-34 by cAMP-dependent protein kinase converts either protein into a potent inhibitor of PP1. In contrast, unphosphorylated inhibitor-2 interacts with the catalytic subunit of PP1 leading first to inhibition of enzyme activity and subsequently to an inactive complex, termed Mg-ATP-dependent PP1 (1,5). The Mg-ATPdependent form of PP1 can then be re-activated following phosphorylation of Thr-72 of inhibitor-2 by glycogen synthase kinase-3 (GSK-3). PP1 is also regulated by its interaction with a variety of protein subunits that act in a manner distinct from the inhibitor proteins and that appear to target the catalytic subunit to specific subcellular compartments (1,2,6,7). These regulatory subunits include the following: the glycogen-targeting proteins, G M and G L (8); the myofibrillar-targeting protein, M110 (8); and the nuclear-targeting protein, PNUTS (9,10). Our recent studies of DARPP-32 (11), studies of G M (12), M110 (12,13), inhibitor-1 (14), and peptide display library analysis (15) have indicated that PP1 interacts with phospho-DARPP-32 and the various binding subunits via a short amino acid motif. The exact sequence of the motif is not identical, but one or more basic amino acids is followed by two hydrophobic residues separated by a variable amino acid. In DARPP-32 the motif is found between residues 6 and 11 (RKKIQF). Furthermore, our studies of DARPP-32 have suggested that residues 6 -11 bind to PP1 at a site removed from the active site and that this interaction is not directly involved in inhibition of enzyme activity (11). Inhibition of PP1 by phospho-DARPP-32 requires phospho-Thr-34, which is likely to occupy the active site of the enzyme in a manner in which catalysis cannot take place (11,16,17).
The identification of the basic/hydrophobic motif in DARPP-32 and the other binding subunits provides a structural basis for their interaction with PP1 in a mutually exclusive manner. However, much less is known about the interaction of inhibitor-2 with PP1. There is no obvious amino acid sequence identity between inhibitor-2 and DARPP-32 or the other binding subunits. Proteolysis studies failed to identify short regions of inhibitor-2 that retained the properties of the holoprotein (18). However, initial truncation mutagenesis studies indicated that residues 1-35 were important for inhibition but not for inactivation and complex formation and that a region of the COOH terminus was required for reactivation (5). Given the lack of information concerning the interaction of inhibitor-2 with PP1, we have carried out a detailed structurefunction analysis of the protein. We have also characterized further the roles of amino acids within the PP1 binding motif of DARPP-32 and those surrounding phospho-Thr-34 of the protein. Finally we have used peptide competition studies to compare the interaction of inhibitor-2 and phospho-DARPP-32 with PP1. The results obtained indicate that distinct amino acid motifs contained within the NH 2 termini of phospho-DARPP-32 (KKIQF, where bold indicates important residues) and inhibitor-2 (IKGI) are critical for inhibition of PP1 and are likely to bind to different sites on the enzyme, both of which are removed from the active site. This study therefore identifies a novel mode of interaction of PP1 with its target proteins and extends our understanding of this growing family of important regulatory molecules that regulate cell signaling through control of serine/threonine dephosphorylation.
Preparation of Wild-type and Mutant Inhibitor-2-Human inhibitor-2 cDNA (21) was used as a template for truncation and site-directed mutagenesis using polymerase chain reaction. Deletion and amino acid substitution mutants were both amplified using appropriate primers from 100 ng of inhibitor-2 cDNA (oligonucleotides were synthesized by Operon). The amplified DNA was purified, digested with NdeI and BamHI, and subcloned into pET-3a (or pET-3cp for T7-tagged inhibitor-2). All mutations were confirmed by DNA sequencing.
Wild-type inhibitor-2 and mutants were purified using a modification of the method described by Helps et al. (21). BL21 (DE3) cells containing the expression plasmids were grown in LB broth with ampicillin (0.1 g/liter) at 37°C until the absorbance at 600 nm was between 0.6 and 1.0. Isopropyl-1-thio-␤-D-galactopyranoside (final concentration 0.4 mM) was added, and the incubation was continued for 4 h. Bacteria were harvested by centrifugation, and the pellet was resuspended in 100 ml of buffer A (20 mM Tris-HCl (pH 7.5), 0.2 mM phenylmethylsulfonyl fluoride, 0.2 mM EDTA, 4.0 mM benzamidine, and 0.1% 2-mercaptoethanol), and cells were lysed using a French press (1000 -1500 psi). The lysate was heated in boiled water for 10 min and was then subjected to centrifugation at 20,000 ϫ g to remove insoluble material. The supernatant was loaded onto a DEAE-cellulose column (100 ml); the column was washed, and bound proteins were eluted using a linear gradient from 0.0 to 0.6 M NaCl (in buffer B, 1000 ml total volume). Fractions containing inhibitor-2 were pooled and concentrated to a volume of 5 ml by ultrafiltration using a YM-10 membrane. The concentrated sample was diluted with 10 ml of buffer B (20 mM Tris-HCl (pH 7.5), 0.2 mM EDTA, 1.0 mM dithiothreitol) and loaded onto a Blue-Sepharose column (40 ml). The column was washed, and bound protein was eluted with a linear gradient from 0.0 -1.0 M NaCl (in buffer B, 500 ml total volume). Fractions containing inhibitor-2 were pooled and diluted by addition of 1 volume of buffer B. The sample was loaded onto a Mono-Q column (Amersham Pharmacia Biotech 10/10); the column was washed, and proteins were eluted with a linear gradient from 0.18 to 0.6 M NaCl (in buffer A). Fractions containing purified inhibitor-2 were pooled, dialyzed against water, and lyophilized. Wild-type inhibitor-2 and all mu-tants were purified to greater than 95% homogeneity as judged by SDS-PAGE (see Fig. 3). The yield of each mutant was similar to that of the wild-type protein (2-5 mg/liter). In general the binding affinity of inhibitor-2 mutants to Blue-Sepharose was proportional to their inhibitory potencies.
Preparation of Wild-type and Mutant DARPP-32-Vector construction, plasmid transformation, and protein expression were performed essentially as described (23). The rat cDNA in pET3a was used as a template for site-directed mutagenesis using polymerase chain reaction. Deletion, insertion, and amino acid substitution mutants were amplified using appropriate primers from 100 ng of DARPP-32 cDNA (all oligonucleotides were synthesized by Operon). The amplified DNA was gel-purified, digested with NdeI and StuI or NdeI and AvaI, and substituted for the appropriate part of the DARPP-32 cDNA in the pET3a vector. All mutations were confirmed by DNA sequencing.
DARPP-32 mutants were purified by heat treatment, anion ionexchange chromatography using DEAE-cellulose, gel filtration using Sephacryl S-200, and fast protein liquid chromatography using a Mono-Q column. Wild-type DARPP-32 and all mutant proteins were purified to greater than 95% homogeneity as judged by SDS-PAGE (data not shown). Wild-type and mutant DARPP-32 was phosphorylated stoichiometrically by cAMP-dependent protein kinase, and phospho-DARPP-32 was purified using a Mono-Q column as described (17).
PP1 Assays-PP1 was assayed using [ 32 P]phosphorylase a as substrate essentially as described (20). Assay mixtures (final volume 30 l) contained 50 mM Tris-HCl, 0.15 mM EGTA, 15 mM 2-mercaptoethanol, 0.01% (w/w) Brij 35, 0.3 mg/ml bovine serum albumin, 5 mM caffeine, 10 M [ 32 P]phosphorylase a, various protein inhibitors, and PP1. For inhibitor-2 assays, PP1 and inhibitor-2 were preincubated at 30°C for 10 min. Dephosphorylation reactions were initiated by the addition of substrate, and reactions were carried out for 10 min at 30°C. Antagonist peptides and inhibitor proteins were premixed before addition of PP1. All reactions were performed in duplicate. All experiments were performed at least two times, with typical errors being less than 20% of the mean.
PP1 Overlay Assay-Binding of PP1 to wild-type and mutant inhibitor-2 was analyzed using a PP1 overlay technique essentially as described (24). Briefly, proteins were separated by SDS-PAGE (using the method of Laemmli, 12.5% acrylamide) and transferred to nitrocellulose filters. The nitrocellulose filters were incubated with a buffer containing 10 mM Tris-HCl (pH 7.4), 2% (w/v) dried milk, 0.1% Tween 20. Filters were washed with phosphate-buffered saline containing 0.2% Nonidet P-40 and then incubated with phosphate-buffered saline/ Nonidet P-40 containing 0.5 g/ml recombinant PP1 for 2 h at 4°C. Filters were washed with phosphate-buffered saline/Nonidet P-40, and bound PP1 was detected using antibody to PP1 as described (25).

Characterization of Inhibition of PP1 by Site-directed Mutants of DARPP-32-Our
previous studies have provided support for a model in which two distinct subdomains in DARPP-32 interact with PP1. Subdomain 1 contains the phosphorylated threonine (Thr-34) and surrounding residues (RRRRPTPMLF; residues 29 -38), and subdomain 2 includes a short stretch of residues at the NH 2 terminus of the protein (RKKIQF; residues 6 -11). Site-directed mutagenesis of fulllength DARPP-32 was therefore used to assess the role of individual amino acids in subdomains 1 and 2 and also the relationship between the two subdomains. Each mutant was phosphorylated stoichiometrically by cAMP-dependent protein kinase, and the inhibitory potencies of the various phosphoproteins was measured.
Within subdomain 2, deletion of amino acids 6 -9 (D32-(⌬6 -9)) or replacement of these residues (D32-(R6S,K7Q,K8Q, I9S)) resulted in a large increase in the IC 50 for inhibition of PP1 ( Fig. 1A and Table I). Further deletion of residues 10 and 11 (D32-(⌬6 -11)) resulted in an additional increase in the IC 50 . Within this region, mutation of Lys-7, but not Lys-8, resulted in a small but significant increase in the IC 50 . Mutation of Ile-9 to glycine resulted in a large increase in the IC 50 ; however, mutation of this residue to alanine had little effect. Mutation of Phe-11 to alanine resulted in a large increase in IC 50 ; however, mutation of this residue to tryptophan had no effect. Within subdomain 1, mutation of Arg-29 or Arg-30 had no effect, although combined mutation of these two residues resulted in a modest increase in IC 50 ( Fig. 1B and Table I). Mutation of Pro-33 and Pro-35 (D32-(P33G,P35G)) resulted in a large increase in IC 50 .
The role of certain of the amino acids that link subdomains 1 and 2 was also investigated. Combined mutation of several hydrophobic amino acids that might have formed an amphipathic ␣-helix (Leu-20, Val-25, and Ile-28; D32-(L20A,V25A,I28A)) resulted in an increase in IC 50 . Deletion of residues 15-18 (D32-(⌬15-18)) also resulted in an increase in IC 50 . In contrast, insertion of three alanine residues between residues 15 and 16 (D32-(ins15/AAA/16)) had no effect. The dephospho-form of DARPP-32 binds to PP1 and inhibits enzyme activity, although with an IC 50 in the micromolar range. The ability of dephosphoforms of some of the mutants to inhibit PP1 was therefore also analyzed (Table I). Deletion or substitution of residues 6 -9 resulted in significant increases in IC 50 . Deletion of residues 15-18 also resulted in a small increase in IC 50 . In contrast, dephospho-D32-(ins15/AAA/16) exhibited a slight decrease in IC 50 . Mutation of amino acids within subdomain 1 resulted in only small increases in IC 50 .
Characterization of Inhibition of PP1 by Truncation Mutants of Inhibitor-2-In an attempt to identify regions of inhibitor-2 that are involved in binding to PP1 and inhibition of enzyme activity, a series of truncation mutants were expressed as recombinant proteins. In addition, wild-type inhibitor-2 was chemically cleaved with 2-nitro-5-thiocyanobenzoic acid to produce two large fragments, I2-(t1-84) and I2-(85-205), that were also analyzed. Dose-response curves for the inhibition of PP1 by the various proteins were obtained ( Fig. 2A), and the IC 50 values were calculated (Table II). Under the assay conditions used, the IC 50 for inhibition of PP1 by wild-type inhibitor-2 was ϳ1 nM, a result very similar to that obtained in previous studies (5). Deletion of up to 105 amino acids at the COOH terminus (I2-(1-99)) had little effect on the IC 50 , although further deletion of 15 residues (I2-(t1-84)) resulted in a small but significant increase in IC 50 . Deletion of the first 8 amino acids at the NH 2 terminus had no effect on the IC 50 ; however, deletion of residues 9 -13 caused a large increase in IC 50 . Further removal of NH 2 -terminal residues had little additional effect.

Characterization of Inhibition of PP1 by Site-directed Mutants of Inhibitor-2-
The results obtained using truncated fragments of inhibitor-2 suggested that amino acids at both the NH 2 and COOH termini of the protein were important for binding to PP1 and inhibition of enzyme activity. Amino acids 9 -13 (PIKGI) appeared to be the most important for inhibition, and the contribution of individual amino acids in this region was further assessed by site-directed mutagenesis (Fig. 2B and Table II). Mutation of Ile-10 (I2-(I10G)) had a small but significant effect on inhibition of PP1. Mutation of either Lys-11 (I2-(K11E)), or Ile-13 (I2-(I13G), I2-(I13A)) resulted in large increases in the IC 50 . Notably, substitution of bulky hydrophobic amino acids for Ile-13 (I13F) and I2-(I13W) caused a larger increase in IC 50 than that of glycine or alanine. Combined mutation of Ile-10 with Lys-11 or Ile-10 with Ile-13 resulted in little additional loss of inhibitory potency relative, respectively, to that of mutation of Lys-11 or Ile-13 alone. However, combined mutation of Lys-11 with Ile-13 resulted in a more than additive loss of inhibitory potency relative to that of mutation of Lys-11 or Ile-13 alone.
The potential contribution of several other residues in inhibitor-2 was also assessed. Mutation of Thr-72 (I2-(T72A)), the residue phosphorylated by GSK-3, had no effect on inhibitory potency. This is the same result as described in a previous study (5). The results shown in Fig. 3 (and data not shown) indicated that residues 141-159 appeared to be important for binding to PP1. This region contains the sequence, KRKLHY, which bears some similarity to the KKIQF motif found in subdomain 2 of DARPP-32 (see above), and it has been suggested by preliminary studies by DePaoli-Roach and co-workers (26) that this region may be important in inhibition of PP1. However, mutation of Tyr-147 (I2-(Y147A)) had no effect on inhibitory potency (Table II). We have previously raised the possibility that KKRQF (residues 134 -139) might also be related to the KKIQF motif of DARPP-32 (11). Again mutation of Phe-139 (I2-(F139A)) had no effect on inhibitory potency.
Comparison of the Inhibition of PP1 by Inhibitor-2 and DARPP-32-The results described above indicated that basic and hydrophobic amino acids in the NH 2 -terminal domains of both inhibitor-2 (residues 11-13) and DARPP-32 (residues 7-11) were critical for inhibition of PP1. However, comparison of the results from site-directed mutagenesis suggested that there may be differences between the binding motifs of the two proteins. Therefore, we compared the ability of synthetic peptides, based on amino acid sequences containing the two bind-

DISCUSSION
The results from the present study suggest that phospho-DARPP-32 and inhibitor-2, two heat-stable inhibitors of PP1, interact with the catalytic subunit via different mechanisms that involve the interaction of common and distinct subdomains within the two inhibitors. Interaction of PP1 with either inhibitor requires at least two subdomains within each protein (Fig. 8A). In the case of phospho-DARPP-32, these two subdomains are included between residues 6 and 38. Residues close to, and including the phosphorylated form of Thr-34, define subdomain 1, and a short motif between residues 7 and 11 (KKIQF) defines subdomain 2. In the case of inhibitor-2, a larger part of the molecule (residue 9 to approximately residue 99) is required for inhibition. A short NH 2 -terminal amino acid motif (IKGI, subdomain 3) is located between residues 10 and 13 of inhibitor-2. Subdomain 2 in DARPP-32 and subdomain 3 in inhibitor-2 both include basic and hydrophobic amino acids that are important for inhibition of PP1. However, these motifs are unrelated to each other and bind to different regions of the catalytic subunit that are both removed from the active site of the enzyme (Fig. 8B). Inhibitor-2 also contains a region between residues 15 and 84 that binds or overlaps the part of PP1 that interacts with subdomain 2 of DARPP-32. For phospho-DARPP-32, inhibition of PP1 requires the interaction of phospho-Thr-34 with the active site of the enzyme or with residues very close to the active site. In the case of inhibitor-2, inhibition of PP1 presumably requires the interaction of residues between 15 and 84 with the active site of the enzyme or with residues close to the active site.
This study extends our investigations of the role of subdomains 1 and 2 of phospho-DARPP-32 in the interaction with  PP1 (11,16). Our previous studies had indicated that phospho-Thr-34 is essential for potent inhibition of PP1; notably a peptide containing phosphoserine in place of phosphothreonine was ineffective as an inhibitor (16). Mutation of residues in the active site of PP1 had a parallel effect on the inhibitory potency of phospho-DARPP-32 and a variety of toxins that are known to interact with the active site (17). In addition, mutation of residues away from the active site of PP1 was able to influence the ability of the enzymes to either be inhibited by phospho-DARPP-32 or alternatively to dephosphorylate phospho-Thr-34 (17). Whereas the underlying structural basis for the latter result is not known, these results suggest in native PP1 that phospho-Thr-34 interacts with residues close to or in the active site in a manner in which it cannot be dephosphorylated. The fact that mutation of Pro-33 and Pro-35 reduces the inhibitory potency of phospho-DARPP-32 suggests that these two residues that flank phospho-Thr-34 may be involved in this interaction.
The results also extend our knowledge of the precise role played by specific amino acids within the basic/hydrophobic motif that is found in subdomain 2 of DARPP-32 and within subdomains of a growing number of PP1-binding proteins (4,8,9,12,(27)(28)(29)(30)(31)(32). The exact sequence of the motif is not identical, but one or more basic amino acids is followed by two hydrophobic residues separated by a variable amino acid. The first of the two hydrophobic amino acids is either valine or isoleucine, and the second hydrophobic residue is phenylalanine (11-15). The molecular basis for the interaction of this BB(V/I)XF motif with PP1 has recently been determined using x-ray crystallography (12). Six residues (RRVSFA) of a 13-residue peptide containing the PP1-binding domain of G M are ordered in the crystal structure and interact in an extended manner with a hydrophobic channel situated on the side opposite from that of the active site of PP1. Interactions are found between the side chains of the valine and phenylalanine in the peptide and solvent-exposed hydrophobic side chains in PP1. Electrostatic interactions are also found between the two arginine residues in the peptide and acidic residues in PP1.
In the present study, mutation of Ile-9 to glycine reduced the inhibitory potency of phospho-DARPP-32. Mutation to alanine was largely accommodated, suggesting that there is some flexibility in the identity of the hydrophobic amino acid at this position. Mutation of Phe-11 to alanine reduced inhibitory potency; however, phospho-DARPP-32 in which Phe-11 was mutated to tryptophan retained full inhibitory potency. A random peptide library analysis has indicated that tryptophan may substitute for phenylalanine (15). In addition, our recent studies have indicated that tryptophan is the residue found in this position in the PP1 binding motif of the nuclear targeting, PP1-binding subunit, PNUTS (9). 2 Mutation of Lys-7 (to glutamate) had a relatively small effect on inhibitory potency. Unexpectedly, mutation of Lys-8 (to glutamate) had no effect. These results suggest that, within the binding motif, basic amino acids play a lesser role than the hydrophobic residues in binding to PP1 and that the relative role of the multiple basic amino acids may vary in different PP1-binding proteins. Importantly, whereas the BB(V/I)XF motif is found in virtually all PP1-binding proteins, the relative affinity of the different proteins varies considerably. For example, the dephospho-and phospho-forms of DARPP-32 bind to PP1 with approximate micromolar K d values (23). As a result, DARPP-32 is not retained by PP1 using affinity chromatography or identified as a PP1-binding protein in immunoprecipitation or overlay assays (data not shown). Mutation of Phe-11 of DARPP-32 to tryptophan did not alter the apparent affinity for PP1, as measured by the overlay assay (data not shown). Thus whereas the BB(V/ I)XF motif is essential for binding of DARPP-32 and other proteins to PP1, additional interaction at one or more sites influences the overall affinity between PP1 and its binding subunits.
The distance between the two subdomains of phospho-DARPP-32 appears to be important since deletion of residues 15-18 reduced inhibitory potency. In contrast, insertion of three amino acids in this region had no effect. Combined mutation of three hydrophobic residues between the two subdomains resulted in a marked reduction in inhibitory potency. Comparison of the amino acid sequences of DARPP-32 and inhibitor-1 between subdomains 2 and 1 reveals limited identity (16). However, Leu-20 and Ile-28 are conserved, and Val-25 is an alanine residue in inhibitor-1, suggesting that hydrophobic contacts with PP1 may be important within this region of the two proteins. Circular dichroism studies of DARPP-32 3 and inhibitor-1 (33) have indicated that both proteins are largely disordered in structure. These hydrophobic residues could therefore interact in an extended conformation with surfaceexposed hydrophobic amino acids in PP1. Alternatively, the region between the subdomains of DARPP-32 may have the ability to form an amphipathic ␣-helix upon binding to PP1.
Assuming that Phe-11 of DARPP-32 interacts with PP1 in the same way as Phe-68 of the G M peptide (12), and phospho-Thr-34 interacts with the active site of the enzyme, there are several alternative ways in which intervening residues could interact with PP1. We have previously suggested, based on initial modeling studies, that the basic amino acid side chains in residues 29 -32 might interact with several acidic amino acids present in a groove close to the active site of PP1 (34). However, this model is not supported by the present results that indicated that Arg-29 and Arg-30 did not make a major contribution to the interaction of DARPP-32 with PP1. In addition, mutation of several of the acidic amino acids in PP1, either singly or in combination, did not reduce the inhibitory potency of phospho-DARPP-32 (17). Given that many of these acidic residues are found only in the catalytic subunit of PP1, but not PP2A and PP2B, it remains an attractive possibility that they play a role in binding other PP1-binding proteins. Alternative modes of binding for phospho-DARPP-32 could include the following: interaction 1) with residues in the COOHterminal groove that also emanates from the active site of the enzyme, or 2) with the ␤12-␤13 loop situated above the active site in a manner similar to that of the autoinhibitory segment of PP2B (35).
The results obtained from the present studies indicate that the first half of inhibitor-2 (approximately residues 9 -99) retains the same inhibitory qualities as the wild-type protein. A subdomain including residues 10 -13 defines a novel PP1 binding motif (IKGI) that is distinct from that identified in subdomain 2 of DARPP-32 (KKIQF). These results extend previous studies that had indicated an important role for the NH 2terminal 35 residues of the protein in inhibition of PP1 (5). In addition to the novel PP1 binding region, residues 15-84 of inhibitor-2 also contain a second region that binds or overlaps the part of PP1 that interacts with DARPP-32, and these distinct but overlapping modes of binding of DARPP-32 and inhibitor-2 to PP1 are consistent with previous kinetic studies (36). Analysis of the primary structure of inhibitor-2 had failed to identify any obvious similarity to DARPP-32, inhibitor-1, or other PP1-binding proteins (11,18,37). Inhibitor-2 is, however, similar in sequence to the Glc8 protein, an apparent functional homolog identified in Saccharomyces cerevisiae (38). The level of amino acid identity in inhibitor-2 and Glc8 is low (28% over 190 residues), but there are several short stretches of the two proteins that are highly conserved, including the region around Thr-72 (of inhibitor-2). Based on a previous alignment of the amino acid sequences of inhibitor-2 and Glc8 (38), residues 10 -14 of inhibitor-2 (IKGI) are IPGL in Glc8 (residues 48 -51). Interestingly, in inhibitor-2 the sequence KKSQKW (residues 41-46) is conserved as EERVQW in Glc8 (residues 88 -93), the latter sequence in Glc8 containing a sequence equivalent to the PP1 binding motif found in subdomain 2 of DARPP-32. Little is known about the biochemical properties of Glc8, but it is possible that these regions of inhibitor-2 and Glc8 represent functionally conserved PP1 binding regions that have diverged through evolution.
An important question raised by these and previous studies relates to whether part of inhibitor-2 interacts with the active 3 H. C. Hemmings, unpublished observations. site of PP1. Kinetic studies indicate that inhibitor-2 is a competitive inhibitor when phosphorylase a is used as a substrate (36). Inhibitor-2 also competes with binding of okadaic acid and microcystin to PP1 (39,40), and both of these toxins have been found to bind to the active site of PP1 (17,34). Moreover, mutation of amino acids between residues 76 and 85 of inhibitor-2 resulted in PP1⅐inhibitor-2 complexes that showed high constitutive activity, in contrast to the normally inactive complex obtained with wild-type proteins (5). These various studies all point to the conclusion that part of inhibitor-2, including residues 76 -85, either interacts directly with the active site of PP1 or with residues close to the active site in a manner that blocks binding of peptide substrate (see Fig. 8B).
The overlay analysis suggested that residues 9 -14 contribute to a high affinity interaction between inhibitor-2 and PP1. Whereas our studies focused on the mechanism of inhibition and not of complex formation or inactivation of PP1, preliminary results indicated that residues 1-99 of inhibitor-2 were able to form a stable complex with PP1. Potentially, binding of residues 9 -14 to PP1 might represent a first step in the interaction of inhibitor-2 with PP1 that ultimately results in complex formation and inactivation following the binding of other parts of the two proteins. The overlay analysis also suggested that residues 141-159 of inhibitor-2 are involved in binding to PP1. However, deletion and site-directed mutagenesis and peptide competition studies all indicate that residues 141-159 are not involved in inhibition of PP1. Notably, this region is well conserved in Glc8, so it is likely to play some important functional role. Previous studies have indicated that residues 146 -204 are important for reactivation of PP1 (5), and it therefore seems likely that the high affinity interaction between residues 141 and 159 and PP1 is involved in complex formation and/or reactivation of the enzyme. Recent studies have indicated that inhibitor-2 is present in both the cytosol and nucleus and that the expression of the protein is regulated in a cell cycle-dependent manner (41,42). It is not known if all of the cellular inhibitor-2 is complexed with PP1; however, the localization of the protein appears to be controlled by both nuclear localization and nucleus export signals that are located between residues 140 and 160 (42). The interaction of inhibitor-2 with PP1 via this region may therefore modulate, or be modulated, by the trafficking of inhibitor-2 into and out of the nucleus.